diff --git a/split_1000_1500/image_1048.json b/split_1000_1500/image_1048.json
new file mode 100644
index 0000000000000000000000000000000000000000..d0962eafd965ea360bedc688ec0d1b40f99c6b34
--- /dev/null
+++ b/split_1000_1500/image_1048.json
@@ -0,0 +1 @@
+{"index": 1048, "query": "A Bowflex SelectTech 552 Adjustable Dumbbells (Pair), black/red, adjusts from 5 to 52.5 lbs, dial selection system.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Bowflex SelectTech 552 Adjustable Dumbbells (Pair), black/red, adjusts from 5 to 52.5 lbs, dial selection system..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/zebra/n3jziBpfRp9_3nn5Gf9gp.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1050.txt b/split_1000_1500/image_1050.txt
new file mode 100644
index 0000000000000000000000000000000000000000..0f277f1868b55c67710c0ef2fee4f9af35a84bce
--- /dev/null
+++ b/split_1000_1500/image_1050.txt
@@ -0,0 +1,21 @@
+query: A Milwaukee M18 Fuel Hackzall Reciprocating Saw, red and black, brushless motor, one-handed design (tool only).
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Milwaukee M18 Fuel Hackzall Reciprocating Saw, red and black, brushless motor, one-handed design (tool only)..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Milwaukee M18 Fuel Hackzall Reciprocating Saw, red and black, brushless motor, one-handed design (tool only)..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/kangaroo/2dYosyYlc8OxDBK8W3G_U.png
diff --git a/split_1000_1500/image_1051.txt b/split_1000_1500/image_1051.txt
new file mode 100644
index 0000000000000000000000000000000000000000..8e58bed671c83530751d65a94dbf6b033b84ab6c
--- /dev/null
+++ b/split_1000_1500/image_1051.txt
@@ -0,0 +1,21 @@
+query: An IKEA KALLAX Shelf Unit, white, 30 3/8 x 57 7/8 inches, particleboard construction, can be placed vertically or horizontally.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as An IKEA KALLAX Shelf Unit, white, 30 3/8 x 57 7/8 inches, particleboard construction, can be placed vertically or horizontally..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as An IKEA KALLAX Shelf Unit, white, 30 3/8 x 57 7/8 inches, particleboard construction, can be placed vertically or horizontally..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/zebra/9XFq2oJvOMuSkiv3Jv-PQ.png
diff --git a/split_1000_1500/image_1052_5_twicemask.json b/split_1000_1500/image_1052_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..e5fabf0021bef2a479db20b2aae7aa995d3cb7e3
--- /dev/null
+++ b/split_1000_1500/image_1052_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1052,
+  "prompt": "query: A Sephora Collection Pro Foundation Brush #56, synthetic bristles, medium density, black handle.",
+  "mask_path": "objects/image_1052_5_twicemask.png",
+  "input_image": "objects/image_1052_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1053.txt b/split_1000_1500/image_1053.txt
new file mode 100644
index 0000000000000000000000000000000000000000..a9ac8af7d61a35aa2baeae14fba18139b751b1de
--- /dev/null
+++ b/split_1000_1500/image_1053.txt
@@ -0,0 +1,21 @@
+query: A bottle of Estée Lauder Double Wear Stay-in-Place Makeup, 1 oz, shade 2N1 Desert Beige, matte finish, glass bottle.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A bottle of Estée Lauder Double Wear Stay-in-Place Makeup, 1 oz, shade 2N1 Desert Beige, matte finish, glass bottle..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A bottle of Est\u00e9e Lauder Double Wear Stay-in-Place Makeup, 1 oz, shade 2N1 Desert Beige, matte finish, glass bottle..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/lion/3eo9_1mmFh8XYRQUyFapa.png
diff --git a/split_1000_1500/image_1056.txt b/split_1000_1500/image_1056.txt
new file mode 100644
index 0000000000000000000000000000000000000000..5358aa2bd82c28a46d64dc9c5b68e4b6c7002297
--- /dev/null
+++ b/split_1000_1500/image_1056.txt
@@ -0,0 +1,21 @@
+query: A Dash Cam VIOFO A129 Plus Duo, front and rear camera, 2K 1440P resolution, Wi-Fi GPS logger, night vision.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Dash Cam VIOFO A129 Plus Duo, front and rear camera, 2K 1440P resolution, Wi-Fi GPS logger, night vision..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Dash Cam VIOFO A129 Plus Duo, front and rear camera, 2K 1440P resolution, Wi-Fi GPS logger, night vision..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "21:9", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/kangaroo/BjRodX8YswziHXpU0XhtW.png
diff --git a/split_1000_1500/image_1061.json b/split_1000_1500/image_1061.json
new file mode 100644
index 0000000000000000000000000000000000000000..b6e86b5a43f9175cb32a583e7fb5b3d75af23e0a
--- /dev/null
+++ b/split_1000_1500/image_1061.json
@@ -0,0 +1 @@
+{"index": 1061, "query": "A Logitech MX Master 3S Performance Wireless Mouse, graphite, 8K DPI sensor, quiet clicks, magspeed scrolling wheel.", "aspect_ratio": "3:2", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Logitech MX Master 3S Performance Wireless Mouse, graphite, 8K DPI sensor, quiet clicks, magspeed scrolling wheel..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:2", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/panda/-5B8KrwznwyJtCVy6gJqC.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1064.txt b/split_1000_1500/image_1064.txt
new file mode 100644
index 0000000000000000000000000000000000000000..aebc8ec6921ba59c5a71516a09d2e7f75f6a9ddb
--- /dev/null
+++ b/split_1000_1500/image_1064.txt
@@ -0,0 +1,21 @@
+query: A Philips Hue White and Color Ambiance Smart Light Bulb Starter Kit, 3 bulbs (A19), hue bridge included, voice control compatible.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Philips Hue White and Color Ambiance Smart Light Bulb Starter Kit, 3 bulbs (A19), hue bridge included, voice control compatible..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Philips Hue White and Color Ambiance Smart Light Bulb Starter Kit, 3 bulbs (A19), hue bridge included, voice control compatible..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/lion/yFqidPKVDvJePhpJxUWOz.png
diff --git a/split_1000_1500/image_1071_5_twicemask.json b/split_1000_1500/image_1071_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..147079f38302581a9ea3c683f9028a45c5f420a1
--- /dev/null
+++ b/split_1000_1500/image_1071_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1071,
+  "prompt": "query: A Canon EOS R6 Mark II Mirrorless Camera Body, black, full-frame CMOS sensor, 4K video, subject tracking autofocus.",
+  "mask_path": "objects/image_1071_5_twicemask.png",
+  "input_image": "objects/image_1071_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1080.txt b/split_1000_1500/image_1080.txt
new file mode 100644
index 0000000000000000000000000000000000000000..d691ea52b09d8dc1f3c7afa9625bf70885fc54a3
--- /dev/null
+++ b/split_1000_1500/image_1080.txt
@@ -0,0 +1,21 @@
+query: A Samsung Family Hub 3-Door French Door Refrigerator, 24 cu. ft., stainless steel, built-in touchscreen, ice and water dispenser.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Samsung Family Hub 3-Door French Door Refrigerator, 24 cu. ft., stainless steel, built-in touchscreen, ice and water dispenser..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Samsung Family Hub 3-Door French Door Refrigerator, 24 cu. ft., stainless steel, built-in touchscreen, ice and water dispenser..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/koala/8hiyytHjx7TVhL8zcWbFc.png
diff --git a/split_1000_1500/image_1090.json b/split_1000_1500/image_1090.json
new file mode 100644
index 0000000000000000000000000000000000000000..eca917055acf65d09c16f6abf91702f420c1686f
--- /dev/null
+++ b/split_1000_1500/image_1090.json
@@ -0,0 +1 @@
+{"index": 1090, "query": "A Shark Navigator Lift-Away Deluxe Upright Vacuum, blue, HEPA filter, swivel steering, detachable canister.", "aspect_ratio": "3:4", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Shark Navigator Lift-Away Deluxe Upright Vacuum, blue, HEPA filter, swivel steering, detachable canister..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:4", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/panda/nHaEkS0K93-Q-ndUwWWFs.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1093.txt b/split_1000_1500/image_1093.txt
new file mode 100644
index 0000000000000000000000000000000000000000..c4134e1050fe6d9b1b3c89afb7257ab5a2370350
--- /dev/null
+++ b/split_1000_1500/image_1093.txt
@@ -0,0 +1,21 @@
+query: A bottle of Paula's Choice Skin Perfecting 2% BHA Liquid Exfoliant, 4 oz, salicylic acid, unclogs pores, grey bottle.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A bottle of Paula's Choice Skin Perfecting 2% BHA Liquid Exfoliant, 4 oz, salicylic acid, unclogs pores, grey bottle..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A bottle of Paula's Choice Skin Perfecting 2% BHA Liquid Exfoliant, 4 oz, salicylic acid, unclogs pores, grey bottle..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/kangaroo/FSjvNoRT7P2PpBLQT9NGM.png
diff --git a/split_1000_1500/image_1094.txt b/split_1000_1500/image_1094.txt
new file mode 100644
index 0000000000000000000000000000000000000000..827ebae32f1d09af4616bde97366e910e84d9338
--- /dev/null
+++ b/split_1000_1500/image_1094.txt
@@ -0,0 +1,21 @@
+query: A Logitech C920x HD Pro Webcam, black, 1080p video calling, stereo audio, built-in privacy shutter.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Logitech C920x HD Pro Webcam, black, 1080p video calling, stereo audio, built-in privacy shutter..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Logitech C920x HD Pro Webcam, black, 1080p video calling, stereo audio, built-in privacy shutter..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:2", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/panda/NLjnwFuEn51rD3CXWmwXl.png
diff --git a/split_1000_1500/image_1096.txt b/split_1000_1500/image_1096.txt
new file mode 100644
index 0000000000000000000000000000000000000000..334795c1e9ac2e6946a5e2c44319fcd1985e57da
--- /dev/null
+++ b/split_1000_1500/image_1096.txt
@@ -0,0 +1,21 @@
+query: A Breville Barista Express Espresso Machine, brushed stainless steel, built-in conical burr grinder, steam wand.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Breville Barista Express Espresso Machine, brushed stainless steel, built-in conical burr grinder, steam wand..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Breville Barista Express Espresso Machine, brushed stainless steel, built-in conical burr grinder, steam wand..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/penguin/tmYCeO2LN2fxGPkO8fr68.png
diff --git a/split_1000_1500/image_1097.json b/split_1000_1500/image_1097.json
new file mode 100644
index 0000000000000000000000000000000000000000..4d9f6ea2fcf299c05e9544a7ac6838761a8c93a5
--- /dev/null
+++ b/split_1000_1500/image_1097.json
@@ -0,0 +1 @@
+{"index": 1097, "query": "A Herschel Supply Co. Novel Duffel Bag, navy/tan, signature shoe compartment, waterproof zipper, 42.5L capacity.", "aspect_ratio": "21:9", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Herschel Supply Co. Novel Duffel Bag, navy/tan, signature shoe compartment, waterproof zipper, 42.5L capacity..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "21:9", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/koala/wagL5nPbdGQLMUgRNw4-l.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1121.txt b/split_1000_1500/image_1121.txt
new file mode 100644
index 0000000000000000000000000000000000000000..6e6442a57675da1c6444d6e097d8839e7fda86d0
--- /dev/null
+++ b/split_1000_1500/image_1121.txt
@@ -0,0 +1,21 @@
+query: A set of Bose SoundLink Micro Bluetooth Speaker, black, waterproof (IP67), rugged silicone strap, 6-hour battery.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A set of Bose SoundLink Micro Bluetooth Speaker, black, waterproof (IP67), rugged silicone strap, 6-hour battery..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A set of Bose SoundLink Micro Bluetooth Speaker, black, waterproof (IP67), rugged silicone strap, 6-hour battery..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "2:3", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/lion/Y5-Xzy2GXj76BezHW1QP_.png
diff --git a/split_1000_1500/image_1123.txt b/split_1000_1500/image_1123.txt
new file mode 100644
index 0000000000000000000000000000000000000000..9d2ae389cba83ffb1b98a3c0354abbd8f7665c0c
--- /dev/null
+++ b/split_1000_1500/image_1123.txt
@@ -0,0 +1,21 @@
+query: An LG 4.5 cu. ft. Front Load Washer, white, AI sensor dry, ultra-large capacity, stackable design.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as An LG 4.5 cu. ft. Front Load Washer, white, AI sensor dry, ultra-large capacity, stackable design..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as An LG 4.5 cu. ft. Front Load Washer, white, AI sensor dry, ultra-large capacity, stackable design..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/monkey/YvdVAKIPcBwhO2gFMlPA5.png
diff --git a/split_1000_1500/image_1124.json b/split_1000_1500/image_1124.json
new file mode 100644
index 0000000000000000000000000000000000000000..bc54d08a4e839f42f160ea4361335cf1d1365d02
--- /dev/null
+++ b/split_1000_1500/image_1124.json
@@ -0,0 +1 @@
+{"index": 1124, "query": "A Black+Decker 20V MAX Cordless Drill/Driver, orange and black, 11-position clutch, built-in LED light.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Black+Decker 20V MAX Cordless Drill/Driver, orange and black, 11-position clutch, built-in LED light..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/rabbit/drbM4WP_hRshW31zlMbPf.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1127.json b/split_1000_1500/image_1127.json
new file mode 100644
index 0000000000000000000000000000000000000000..0ef20d41b989324f199fa821569acde646017814
--- /dev/null
+++ b/split_1000_1500/image_1127.json
@@ -0,0 +1 @@
+{"index": 1127, "query": "A PetSafe ScoopFree Ultra Self-Cleaning Cat Litter Box, taupe, covered design, uses crystal litter trays, health counter.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A PetSafe ScoopFree Ultra Self-Cleaning Cat Litter Box, taupe, covered design, uses crystal litter trays, health counter..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/penguin/91JBoZPJcTGOxqAKpx1o9.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1130.txt b/split_1000_1500/image_1130.txt
new file mode 100644
index 0000000000000000000000000000000000000000..0f8c0c48fcc7119a6a9d4d17d7a8aac30cf92ed1
--- /dev/null
+++ b/split_1000_1500/image_1130.txt
@@ -0,0 +1,21 @@
+query: A crate of S.Pellegrino Sparkling Natural Mineral Water, 12 pack of 1-liter plastic bottles, green packaging.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A crate of S.Pellegrino Sparkling Natural Mineral Water, 12 pack of 1-liter plastic bottles, green packaging..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A crate of S.Pellegrino Sparkling Natural Mineral Water, 12 pack of 1-liter plastic bottles, green packaging..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "16:9", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/penguin/YL5mjTmaYHN3dVve-5xim.png
diff --git a/split_1000_1500/image_1131.json b/split_1000_1500/image_1131.json
new file mode 100644
index 0000000000000000000000000000000000000000..d77c5cc687fb13f71d9c84f9c684f2adac31968d
--- /dev/null
+++ b/split_1000_1500/image_1131.json
@@ -0,0 +1 @@
+{"index": 1131, "query": "A Google Nest Cam (Outdoor or Indoor), battery-powered, snow white, 1080p HDR video, magnetic mount.", "aspect_ratio": "16:9", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Google Nest Cam (Outdoor or Indoor), battery-powered, snow white, 1080p HDR video, magnetic mount..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "16:9", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/tiger/f0u3UsiWQ2gXoROMTmAQy.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1131_5_twicemask.json b/split_1000_1500/image_1131_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..c76d184e77c75c8ffa2cd817b50223ec3c57ecc1
--- /dev/null
+++ b/split_1000_1500/image_1131_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1131,
+  "prompt": "query: A Google Nest Cam (Outdoor or Indoor), battery-powered, snow white, 1080p HDR video, magnetic mount.",
+  "mask_path": "objects/image_1131_5_twicemask.png",
+  "input_image": "objects/image_1131_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1134.json b/split_1000_1500/image_1134.json
new file mode 100644
index 0000000000000000000000000000000000000000..529f04e80f5fcbf1a4e07e4b9a9c62e2cf29c8af
--- /dev/null
+++ b/split_1000_1500/image_1134.json
@@ -0,0 +1 @@
+{"index": 1134, "query": "A Hori Nintendo Switch Split Pad Pro, black, full-size controller experience in handheld mode, programmable rear buttons.", "aspect_ratio": "3:2", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Hori Nintendo Switch Split Pad Pro, black, full-size controller experience in handheld mode, programmable rear buttons..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:2", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/zebra/nuWnThyLwdeFg3FGQqvhp.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1138_5_twicemask.json b/split_1000_1500/image_1138_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..82d4ee7c85935aed6ec45c2b99d7f7f231ea5c97
--- /dev/null
+++ b/split_1000_1500/image_1138_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1138,
+  "prompt": "query: An ASUS ROG Zephyrus G14 Gaming Laptop, 14\", moonlight white, AMD Ryzen 9, RTX 4060, 165Hz display.",
+  "mask_path": "objects/image_1138_5_twicemask.png",
+  "input_image": "objects/image_1138_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1140.txt b/split_1000_1500/image_1140.txt
new file mode 100644
index 0000000000000000000000000000000000000000..5b5d002adf5807454412ee6d419c035864578007
--- /dev/null
+++ b/split_1000_1500/image_1140.txt
@@ -0,0 +1,21 @@
+query: A SodaStream Terra Sparkling Water Maker, black, includes 60L CO2 cylinder and 1L dishwasher safe bottle.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A SodaStream Terra Sparkling Water Maker, black, includes 60L CO2 cylinder and 1L dishwasher safe bottle..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A SodaStream Terra Sparkling Water Maker, black, includes 60L CO2 cylinder and 1L dishwasher safe bottle..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "2:3", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/zebra/0Iu5GT9BYx4mzAKWmf03u.png
diff --git a/split_1000_1500/image_1143.json b/split_1000_1500/image_1143.json
new file mode 100644
index 0000000000000000000000000000000000000000..78f9eb6ea5f94cb9b098f58b9c6806bfa9d0a45e
--- /dev/null
+++ b/split_1000_1500/image_1143.json
@@ -0,0 +1 @@
+{"index": 1143, "query": "A pair of Beats Studio Pro Wireless Headphones, sandstone, active noise cancelling, transparency mode, USB-C audio.", "aspect_ratio": "4:3", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A pair of Beats Studio Pro Wireless Headphones, sandstone, active noise cancelling, transparency mode, USB-C audio..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "4:3", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/elephant/eEy8tAIF2a7QlifrdvEyZ.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1146.txt b/split_1000_1500/image_1146.txt
new file mode 100644
index 0000000000000000000000000000000000000000..7f216593b8e371f75e370bb7f5c0baba34f04733
--- /dev/null
+++ b/split_1000_1500/image_1146.txt
@@ -0,0 +1,21 @@
+query: A bottle of Yves Saint Laurent Libre Eau de Parfum, 1.6 oz, lavender and orange blossom, asymmetric black cap with gold logo.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A bottle of Yves Saint Laurent Libre Eau de Parfum, 1.6 oz, lavender and orange blossom, asymmetric black cap with gold logo..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A bottle of Yves Saint Laurent Libre Eau de Parfum, 1.6 oz, lavender and orange blossom, asymmetric black cap with gold logo..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "5:4", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/panda/wk2xWUpTAyy9dalmMzXkw.png
diff --git a/split_1000_1500/image_1151.json b/split_1000_1500/image_1151.json
new file mode 100644
index 0000000000000000000000000000000000000000..a8c1a900c63272ff04d26305ea9c10125e8ee0a8
--- /dev/null
+++ b/split_1000_1500/image_1151.json
@@ -0,0 +1 @@
+{"index": 1151, "query": "A tube of Maybelline Lash Sensational Sky High Mascara, Very Black, flexible tower brush, bamboo extract.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A tube of Maybelline Lash Sensational Sky High Mascara, Very Black, flexible tower brush, bamboo extract..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/penguin/cXMzl07MnAS7Xz5yHj-qr.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1155.json b/split_1000_1500/image_1155.json
new file mode 100644
index 0000000000000000000000000000000000000000..0649ad8aea6bdb1fa367814172bb54d60481cdcc
--- /dev/null
+++ b/split_1000_1500/image_1155.json
@@ -0,0 +1 @@
+{"index": 1155, "query": "A bottle of Heinz Tomato Ketchup, 32 oz, squeeze bottle, made from red ripe tomatoes.", "aspect_ratio": "16:9", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A bottle of Heinz Tomato Ketchup, 32 oz, squeeze bottle, made from red ripe tomatoes..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "16:9", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/kangaroo/mSLascE8Vfs7oaF9O1sVP.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1164_5_twicemask.json b/split_1000_1500/image_1164_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..fb44ab60e4c50360b045c28a19ff32040555157d
--- /dev/null
+++ b/split_1000_1500/image_1164_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1164,
+  "prompt": "query: A pair of Sony WF-1000XM5 Wireless Earbuds, black, noise canceling, hi-res audio, bone conduction sensors.",
+  "mask_path": "objects/image_1164_5_twicemask.png",
+  "input_image": "objects/image_1164_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1167_5_twicemask.json b/split_1000_1500/image_1167_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..73600480ae3ee731ad3f9d46d29d599a3add33bf
--- /dev/null
+++ b/split_1000_1500/image_1167_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1167,
+  "prompt": "query: A Garmin Edge 530 GPS Cycling Computer, black, performance mapping, dynamic performance monitoring, 20-hour battery.",
+  "mask_path": "objects/image_1167_5_twicemask.png",
+  "input_image": "objects/image_1167_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1168.json b/split_1000_1500/image_1168.json
new file mode 100644
index 0000000000000000000000000000000000000000..943b3d2df2b6d5da244e39df0cad4b4cc69ee48d
--- /dev/null
+++ b/split_1000_1500/image_1168.json
@@ -0,0 +1 @@
+{"index": 1168, "query": "A tube of Colgate Optic White Renewal Teeth Whitening Toothpaste, 3 oz, 3% hydrogen peroxide, high impact white.", "aspect_ratio": "5:4", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A tube of Colgate Optic White Renewal Teeth Whitening Toothpaste, 3 oz, 3% hydrogen peroxide, high impact white..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "5:4", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/panda/04Sv8Gp0MeDHZoPlzwogA.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1181.txt b/split_1000_1500/image_1181.txt
new file mode 100644
index 0000000000000000000000000000000000000000..2a91905580bf7132fc7c345403c1ecffc0056f71
--- /dev/null
+++ b/split_1000_1500/image_1181.txt
@@ -0,0 +1,21 @@
+query: A Bissell Little Green Multi-Purpose Portable Carpet and Upholstery Cleaner, green, includes spot & stain formula.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Bissell Little Green Multi-Purpose Portable Carpet and Upholstery Cleaner, green, includes spot & stain formula..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Bissell Little Green Multi-Purpose Portable Carpet and Upholstery Cleaner, green, includes spot & stain formula..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:2", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/zebra/P9_UoUl502D4YmNR955Jh.png
diff --git a/split_1000_1500/image_1186.json b/split_1000_1500/image_1186.json
new file mode 100644
index 0000000000000000000000000000000000000000..178d9b85e68a035b97ea55b0e961ab393065596f
--- /dev/null
+++ b/split_1000_1500/image_1186.json
@@ -0,0 +1 @@
+{"index": 1186, "query": "A Logitech G Pro X Mechanical Gaming Keyboard, black, swappable GX Blue clicky switches, RGB lighting.", "aspect_ratio": "4:3", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Logitech G Pro X Mechanical Gaming Keyboard, black, swappable GX Blue clicky switches, RGB lighting..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "4:3", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/penguin/kcRjdRsD8YOnDTkWfjbYe.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1186_5_twicemask.json b/split_1000_1500/image_1186_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..61072dcbfc53e698d6b81aa1cb95648167315c1b
--- /dev/null
+++ b/split_1000_1500/image_1186_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1186,
+  "prompt": "query: A Logitech G Pro X Mechanical Gaming Keyboard, black, swappable GX Blue clicky switches, RGB lighting.",
+  "mask_path": "objects/image_1186_5_twicemask.png",
+  "input_image": "objects/image_1186_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1199_5_twicemask.json b/split_1000_1500/image_1199_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..6ed54cb1800e1f34c022b78cbbcbe19e2694f4d4
--- /dev/null
+++ b/split_1000_1500/image_1199_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1199,
+  "prompt": "query: A bottle of Windex Original Glass Cleaner, 23 fl oz, blue liquid, trigger spray, streak-free shine.",
+  "mask_path": "objects/image_1199_5_twicemask.png",
+  "input_image": "objects/image_1199_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1201.txt b/split_1000_1500/image_1201.txt
new file mode 100644
index 0000000000000000000000000000000000000000..938ade0d4532c00845e484938921243c2f4e7d0f
--- /dev/null
+++ b/split_1000_1500/image_1201.txt
@@ -0,0 +1,21 @@
+query: A Samsung Galaxy S23 Ultra smartphone in phantom black, featuring 256GB storage, a 6.8-inch AMOLED display, and an integrated S Pen.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Samsung Galaxy S23 Ultra smartphone in phantom black, featuring 256GB storage, a 6.8-inch AMOLED display, and an integrated S Pen..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Samsung Galaxy S23 Ultra smartphone in phantom black, featuring 256GB storage, a 6.8-inch AMOLED display, and an integrated S Pen..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/tiger/VgXpXQmfFOXjeGS9hql0S.png
diff --git a/split_1000_1500/image_1206.json b/split_1000_1500/image_1206.json
new file mode 100644
index 0000000000000000000000000000000000000000..4085fb7b3a0a644eeaab04ca681f3ff8968e3f52
--- /dev/null
+++ b/split_1000_1500/image_1206.json
@@ -0,0 +1 @@
+{"index": 1206, "query": "A luxe La Mer Cr\u00e8me de la Mer moisturizing cream in a 60ml porcelain jar, formulated with Miracle Broth for dry skin.", "aspect_ratio": "4:3", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A luxe La Mer Cr\u00e8me de la Mer moisturizing cream in a 60ml porcelain jar, formulated with Miracle Broth for dry skin..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "4:3", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/zebra/AOvg5Qn1_Un0Waecl9enI.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1206.txt b/split_1000_1500/image_1206.txt
new file mode 100644
index 0000000000000000000000000000000000000000..b5197449d49bbd0dead526b35b06d39ce0da2695
--- /dev/null
+++ b/split_1000_1500/image_1206.txt
@@ -0,0 +1,21 @@
+query: A luxe La Mer Crème de la Mer moisturizing cream in a 60ml porcelain jar, formulated with Miracle Broth for dry skin.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A luxe La Mer Crème de la Mer moisturizing cream in a 60ml porcelain jar, formulated with Miracle Broth for dry skin..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A luxe La Mer Cr\u00e8me de la Mer moisturizing cream in a 60ml porcelain jar, formulated with Miracle Broth for dry skin..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "4:3", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/zebra/AOvg5Qn1_Un0Waecl9enI.png
diff --git a/split_1000_1500/image_1211.txt b/split_1000_1500/image_1211.txt
new file mode 100644
index 0000000000000000000000000000000000000000..cd148c96616a24d59af554ddcb82118398349ddd
--- /dev/null
+++ b/split_1000_1500/image_1211.txt
@@ -0,0 +1,21 @@
+query: An 18k yellow gold vermeil pendant necklace featuring a genuine emerald cut malachite stone on a delicate chain.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as An 18k yellow gold vermeil pendant necklace featuring a genuine emerald cut malachite stone on a delicate chain..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as An 18k yellow gold vermeil pendant necklace featuring a genuine emerald cut malachite stone on a delicate chain..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "4:3", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/penguin/7RmVtlV8C252DjgIBxi8B.png
diff --git a/split_1000_1500/image_1213.txt b/split_1000_1500/image_1213.txt
new file mode 100644
index 0000000000000000000000000000000000000000..5da8ff5d60bc6b47d0dfe81474d543e94ae079ae
--- /dev/null
+++ b/split_1000_1500/image_1213.txt
@@ -0,0 +1,21 @@
+query: A yoga mat by Lululemon, 5mm thick, reversible with a marble swirl pattern in pink and white, featuring a textured non-slip grip.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A yoga mat by Lululemon, 5mm thick, reversible with a marble swirl pattern in pink and white, featuring a textured non-slip grip..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A yoga mat by Lululemon, 5mm thick, reversible with a marble swirl pattern in pink and white, featuring a textured non-slip grip..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/rabbit/lFGYsryY6S9y1o5llYS9N.png
diff --git a/split_1000_1500/image_1216.json b/split_1000_1500/image_1216.json
new file mode 100644
index 0000000000000000000000000000000000000000..e00c035506f35a9098c66386a36aeec644ee4bcd
--- /dev/null
+++ b/split_1000_1500/image_1216.json
@@ -0,0 +1 @@
+{"index": 1216, "query": "A bottle of Chanel No. 5 Eau de Parfum, 100ml, featuring notes of aldehydes, jasmine, and bourbon vanilla in a classic glass bottle.", "aspect_ratio": "9:16", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A bottle of Chanel No. 5 Eau de Parfum, 100ml, featuring notes of aldehydes, jasmine, and bourbon vanilla in a classic glass bottle..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "9:16", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/elephant/2B6c6GcPMig4DJUxU5mfa.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1227.json b/split_1000_1500/image_1227.json
new file mode 100644
index 0000000000000000000000000000000000000000..872443f688ef36cedb9cb5029f0f7b6a88bfb902
--- /dev/null
+++ b/split_1000_1500/image_1227.json
@@ -0,0 +1 @@
+{"index": 1227, "query": "A Macat 10-piece stainless steel knife block set, including a chef's knife, santoku, and honing steel with ergonomic black handles.", "aspect_ratio": "9:16", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Macat 10-piece stainless steel knife block set, including a chef's knife, santoku, and honing steel with ergonomic black handles..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "9:16", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/lion/olb7ydCY37G-8XTiBnKjA.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1227_5_twicemask.json b/split_1000_1500/image_1227_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..926d8e5da17b02ee9f090e204aeb08f044f2b92c
--- /dev/null
+++ b/split_1000_1500/image_1227_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1227,
+  "prompt": "query: A Macat 10-piece stainless steel knife block set, including a chef's knife, santoku, and honing steel with ergonomic black handles.",
+  "mask_path": "objects/image_1227_5_twicemask.png",
+  "input_image": "objects/image_1227_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1229_5_twicemask.json b/split_1000_1500/image_1229_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..8cca12c50fd4dca86b5675bb7f38adb7ba39f4ef
--- /dev/null
+++ b/split_1000_1500/image_1229_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1229,
+  "prompt": "query: An Instant Pot Duo 7-in-1 electric pressure cooker, 6-quart capacity, stainless steel with 13 one-touch smart programs.",
+  "mask_path": "objects/image_1229_5_twicemask.png",
+  "input_image": "objects/image_1229_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1234.txt b/split_1000_1500/image_1234.txt
new file mode 100644
index 0000000000000000000000000000000000000000..4f989bc8ab7c82c76776934c478f326e9b461151
--- /dev/null
+++ b/split_1000_1500/image_1234.txt
@@ -0,0 +1,21 @@
+query: A bag of Lavazza Super Crema whole bean coffee blend, 2.2 lbs, featuring notes of hazelnut and brown sugar.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A bag of Lavazza Super Crema whole bean coffee blend, 2.2 lbs, featuring notes of hazelnut and brown sugar..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A bag of Lavazza Super Crema whole bean coffee blend, 2.2 lbs, featuring notes of hazelnut and brown sugar..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/kangaroo/O2-W17iADhQGemERPPq_s.png
diff --git a/split_1000_1500/image_1235.json b/split_1000_1500/image_1235.json
new file mode 100644
index 0000000000000000000000000000000000000000..0bb53702787c68e63240b55e48e427a94b156878
--- /dev/null
+++ b/split_1000_1500/image_1235.json
@@ -0,0 +1 @@
+{"index": 1235, "query": "A heavy-duty camping tent by Coleman, sundome 4-person design in green, with a rainfly and large windows for ventilation.", "aspect_ratio": "9:16", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A heavy-duty camping tent by Coleman, sundome 4-person design in green, with a rainfly and large windows for ventilation..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "9:16", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/tiger/2jFAfM37SdRopHjgQjnUW.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1243.txt b/split_1000_1500/image_1243.txt
new file mode 100644
index 0000000000000000000000000000000000000000..af59cffbee700ca0ed901c3539714397d0e523dd
--- /dev/null
+++ b/split_1000_1500/image_1243.txt
@@ -0,0 +1,21 @@
+query: An LG 65-inch Class C3 Series OLED evo 4K UHD Smart webOS TV with Dolby Vision and sleek bezel-less design.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as An LG 65-inch Class C3 Series OLED evo 4K UHD Smart webOS TV with Dolby Vision and sleek bezel-less design..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as An LG 65-inch Class C3 Series OLED evo 4K UHD Smart webOS TV with Dolby Vision and sleek bezel-less design..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/tiger/eyoDlyZQraW5siBxLvpMO.png
diff --git a/split_1000_1500/image_1246.json b/split_1000_1500/image_1246.json
new file mode 100644
index 0000000000000000000000000000000000000000..d829e6b452108da190c2e385b69d32654b2e8789
--- /dev/null
+++ b/split_1000_1500/image_1246.json
@@ -0,0 +1 @@
+{"index": 1246, "query": "A Fujifilm Instax Mini 12 instant camera in pastel blue, featuring automatic exposure and a selfie mirror.", "aspect_ratio": "3:2", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Fujifilm Instax Mini 12 instant camera in pastel blue, featuring automatic exposure and a selfie mirror..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:2", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/elephant/czmPxEwMPr5usLN8be3ce.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1247.json b/split_1000_1500/image_1247.json
new file mode 100644
index 0000000000000000000000000000000000000000..ad3afc29c3ff70cfbcf7120e7f0433b78db723d9
--- /dev/null
+++ b/split_1000_1500/image_1247.json
@@ -0,0 +1 @@
+{"index": 1247, "query": "A bottle of Olaplex No. 3 Hair Perfector, 100ml, a concentrated pre-shampoo treatment to repair damaged hair bonds.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A bottle of Olaplex No. 3 Hair Perfector, 100ml, a concentrated pre-shampoo treatment to repair damaged hair bonds..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/koala/uKy0dk38A8d2BjihWwR09.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1250.json b/split_1000_1500/image_1250.json
new file mode 100644
index 0000000000000000000000000000000000000000..e2ad3d78b9e845e97888595214a939b28c59579f
--- /dev/null
+++ b/split_1000_1500/image_1250.json
@@ -0,0 +1 @@
+{"index": 1250, "query": "A modern floor lamp with a tripod wooden base and a white linen drum shade, standing 60 inches tall.", "aspect_ratio": "3:2", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A modern floor lamp with a tripod wooden base and a white linen drum shade, standing 60 inches tall..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:2", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/koala/uslVjRQTXnVPc5sFUeYXe.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1251_5_twicemask.json b/split_1000_1500/image_1251_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..6138eca920c9d3bf1560420676d1f519e9d4701f
--- /dev/null
+++ b/split_1000_1500/image_1251_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1251,
+  "prompt": "query: A pair of Bose QuietComfort Ultra earbuds in black, featuring spatial audio and world-class noise cancellation.",
+  "mask_path": "objects/image_1251_5_twicemask.png",
+  "input_image": "objects/image_1251_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1254.json b/split_1000_1500/image_1254.json
new file mode 100644
index 0000000000000000000000000000000000000000..a0a77e937e1239e4f63341bce11046572f9fb1c6
--- /dev/null
+++ b/split_1000_1500/image_1254.json
@@ -0,0 +1 @@
+{"index": 1254, "query": "A woven rattan laundry hamper with a lid and cotton liner, natural finish, rectangular shape.", "aspect_ratio": "3:4", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A woven rattan laundry hamper with a lid and cotton liner, natural finish, rectangular shape..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:4", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/kangaroo/VBi_xc0gUx2bIUvQEe1MU.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1263_5_twicemask.json b/split_1000_1500/image_1263_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..17c56db59d920b29d9f1b7b60717d15757c4adb9
--- /dev/null
+++ b/split_1000_1500/image_1263_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1263,
+  "prompt": "query: A set of 400-thread count Egyptian cotton bed sheets, queen size in ivory, sateen weave for a silky feel.",
+  "mask_path": "objects/image_1263_5_twicemask.png",
+  "input_image": "objects/image_1263_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1279.json b/split_1000_1500/image_1279.json
new file mode 100644
index 0000000000000000000000000000000000000000..fa55f2adde75689776d18c9b57093aac39f86037
--- /dev/null
+++ b/split_1000_1500/image_1279.json
@@ -0,0 +1 @@
+{"index": 1279, "query": "A stainless steel toaster by Smeg, 2-slice retro 50s style in pastel green, with extra-wide slots.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A stainless steel toaster by Smeg, 2-slice retro 50s style in pastel green, with extra-wide slots..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/tiger/q_OyWUz-Wk_PfK2kg6_sf.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1280.json b/split_1000_1500/image_1280.json
new file mode 100644
index 0000000000000000000000000000000000000000..ced07c60857afe41d9335ceaf04c830f2e047422
--- /dev/null
+++ b/split_1000_1500/image_1280.json
@@ -0,0 +1 @@
+{"index": 1280, "query": "A bag of Royal Canin Feline Health Nutrition Indoor Adult dry cat food, 7 lb bag, formulated for hairball reduction.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A bag of Royal Canin Feline Health Nutrition Indoor Adult dry cat food, 7 lb bag, formulated for hairball reduction..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/panda/rG9MQwOdmYrV2JOzvWS3Y.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1283.txt b/split_1000_1500/image_1283.txt
new file mode 100644
index 0000000000000000000000000000000000000000..28b3a29d6a3099ead8d9b6716700123eb60c8bd5
--- /dev/null
+++ b/split_1000_1500/image_1283.txt
@@ -0,0 +1,21 @@
+query: A classic fedora hat by Brixton, made of wool felt in moss green with a leather band.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A classic fedora hat by Brixton, made of wool felt in moss green with a leather band..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A classic fedora hat by Brixton, made of wool felt in moss green with a leather band..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "4:5", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/lion/nsmkyiWriAU88l1EpnC_K.png
diff --git a/split_1000_1500/image_1283_5_twicemask.json b/split_1000_1500/image_1283_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..3c530df2ef20e9afd17516f1a6905ede7dcfc1c3
--- /dev/null
+++ b/split_1000_1500/image_1283_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1283,
+  "prompt": "query: A classic fedora hat by Brixton, made of wool felt in moss green with a leather band.",
+  "mask_path": "objects/image_1283_5_twicemask.png",
+  "input_image": "objects/image_1283_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1288.txt b/split_1000_1500/image_1288.txt
new file mode 100644
index 0000000000000000000000000000000000000000..5e1453f1f1d2c2ddb4e5e8500e279c78604acb84
--- /dev/null
+++ b/split_1000_1500/image_1288.txt
@@ -0,0 +1,21 @@
+query: A Microsoft Surface Laptop 5, 13.5-inch touchscreen, Intel Evo i5, 8GB RAM, 256GB SSD in platinum Alcantara.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Microsoft Surface Laptop 5, 13.5-inch touchscreen, Intel Evo i5, 8GB RAM, 256GB SSD in platinum Alcantara..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Microsoft Surface Laptop 5, 13.5-inch touchscreen, Intel Evo i5, 8GB RAM, 256GB SSD in platinum Alcantara..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/panda/a3r6Zb-ojy8kQBZa0Mw8K.png
diff --git a/split_1000_1500/image_1289.json b/split_1000_1500/image_1289.json
new file mode 100644
index 0000000000000000000000000000000000000000..5e12b4aeed7e1b5de916adc363c27db00ba2e7a5
--- /dev/null
+++ b/split_1000_1500/image_1289.json
@@ -0,0 +1 @@
+{"index": 1289, "query": "A set of acrylic paints by Liquitex, basics 48-tube set, 22ml each, suitable for canvas and wood.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A set of acrylic paints by Liquitex, basics 48-tube set, 22ml each, suitable for canvas and wood..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/panda/DOG3cm0YzSbGYgLiCUaDT.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1289.txt b/split_1000_1500/image_1289.txt
new file mode 100644
index 0000000000000000000000000000000000000000..f698f295fa2c48d9e15334530d490df1a56ad6b3
--- /dev/null
+++ b/split_1000_1500/image_1289.txt
@@ -0,0 +1,21 @@
+query: A set of acrylic paints by Liquitex, basics 48-tube set, 22ml each, suitable for canvas and wood.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A set of acrylic paints by Liquitex, basics 48-tube set, 22ml each, suitable for canvas and wood..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A set of acrylic paints by Liquitex, basics 48-tube set, 22ml each, suitable for canvas and wood..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/panda/DOG3cm0YzSbGYgLiCUaDT.png
diff --git a/split_1000_1500/image_1293.json b/split_1000_1500/image_1293.json
new file mode 100644
index 0000000000000000000000000000000000000000..6eae8df9e610f40482962d0f394f1ac8d5e3bfc6
--- /dev/null
+++ b/split_1000_1500/image_1293.json
@@ -0,0 +1 @@
+{"index": 1293, "query": "An organic matcha green tea powder by Jade Leaf, culinary grade, 3.53 oz pouch for lattes and baking.", "aspect_ratio": "4:3", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as An organic matcha green tea powder by Jade Leaf, culinary grade, 3.53 oz pouch for lattes and baking..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "4:3", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/penguin/scXOjrNo40_VCXuSY4RT6.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1300.txt b/split_1000_1500/image_1300.txt
new file mode 100644
index 0000000000000000000000000000000000000000..64c8b93f94a6321f1fc776375b5baab94db22a6a
--- /dev/null
+++ b/split_1000_1500/image_1300.txt
@@ -0,0 +1,21 @@
+query: A Brother HL-L2350DW monochrome laser printer, wireless compact design with duplex printing.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Brother HL-L2350DW monochrome laser printer, wireless compact design with duplex printing..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Brother HL-L2350DW monochrome laser printer, wireless compact design with duplex printing..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "16:9", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/penguin/xgnE9xBZfKyl5ciBC4dx9.png
diff --git a/split_1000_1500/image_1303.json b/split_1000_1500/image_1303.json
new file mode 100644
index 0000000000000000000000000000000000000000..8ea07bbdf9742bf17a9ce405080068cdffea2856
--- /dev/null
+++ b/split_1000_1500/image_1303.json
@@ -0,0 +1 @@
+{"index": 1303, "query": "A set of packing cubes by Shacke Pak, 4-piece set with laundry bag in orchid purple.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A set of packing cubes by Shacke Pak, 4-piece set with laundry bag in orchid purple..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/zebra/7thVeYP8MngPLXmffhHms.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1322.json b/split_1000_1500/image_1322.json
new file mode 100644
index 0000000000000000000000000000000000000000..e4ca8c91f5865024143d0486540e0530b382176c
--- /dev/null
+++ b/split_1000_1500/image_1322.json
@@ -0,0 +1 @@
+{"index": 1322, "query": "A pair of sterling silver hoop earrings, 50mm diameter, high polished finish with click-top closure.", "aspect_ratio": "3:4", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A pair of sterling silver hoop earrings, 50mm diameter, high polished finish with click-top closure..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:4", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/lion/EbDhbUJE8M2aoacKw6dlv.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1325.txt b/split_1000_1500/image_1325.txt
new file mode 100644
index 0000000000000000000000000000000000000000..ab402bbafe8fa6dfd9f356aa00f8dacab10b0b5b
--- /dev/null
+++ b/split_1000_1500/image_1325.txt
@@ -0,0 +1,21 @@
+query: A set of makeup brushes by Real Techniques, Everyday Essentials, including blush, crease, and setting brushes.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A set of makeup brushes by Real Techniques, Everyday Essentials, including blush, crease, and setting brushes..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A set of makeup brushes by Real Techniques, Everyday Essentials, including blush, crease, and setting brushes..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "5:4", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/koala/w_k_l9g1iCOndTODvJBhb.png
diff --git a/split_1000_1500/image_1331.txt b/split_1000_1500/image_1331.txt
new file mode 100644
index 0000000000000000000000000000000000000000..094bb5b4257aa07f2ad2e8db7e2e10b4fc07e09c
--- /dev/null
+++ b/split_1000_1500/image_1331.txt
@@ -0,0 +1,21 @@
+query: An adjustable laptop stand made of aluminum alloy, silver finish, foldable and ergonomic.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as An adjustable laptop stand made of aluminum alloy, silver finish, foldable and ergonomic..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as An adjustable laptop stand made of aluminum alloy, silver finish, foldable and ergonomic..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "4:5", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/tiger/rlMuP7_-bbOqEq7fTf8ks.png
diff --git a/split_1000_1500/image_1333_5_twicemask.json b/split_1000_1500/image_1333_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..4afaf9b50552bbe9f0919c2942e4491c978faf63
--- /dev/null
+++ b/split_1000_1500/image_1333_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1333,
+  "prompt": "query: A 24-piece makeup palette by Morphe, 24G Grand Glam Eyeshadow Palette, featuring oversized pans.",
+  "mask_path": "objects/image_1333_5_twicemask.png",
+  "input_image": "objects/image_1333_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1348.json b/split_1000_1500/image_1348.json
new file mode 100644
index 0000000000000000000000000000000000000000..529f4855e4c058203d184acf65af0e0221edb253
--- /dev/null
+++ b/split_1000_1500/image_1348.json
@@ -0,0 +1 @@
+{"index": 1348, "query": "A tennis racket by Babolat, Pure Drive 2021 model, unstrung, blue and black frame, grip size 4 3/8.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A tennis racket by Babolat, Pure Drive 2021 model, unstrung, blue and black frame, grip size 4 3/8..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/koala/pLFN5e15AI-t91LIduhjZ.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1350_5_twicemask.json b/split_1000_1500/image_1350_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..abbef8c7f07ec46867b0eefdaee323259cfd5ff3
--- /dev/null
+++ b/split_1000_1500/image_1350_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1350,
+  "prompt": "query: A bottle of extra virgin olive oil by California Olive Ranch, 16.9 oz, cold pressed everyday blend.",
+  "mask_path": "objects/image_1350_5_twicemask.png",
+  "input_image": "objects/image_1350_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1352.txt b/split_1000_1500/image_1352.txt
new file mode 100644
index 0000000000000000000000000000000000000000..73a078878ae0c4e7738a1923dabbc6086371ebf1
--- /dev/null
+++ b/split_1000_1500/image_1352.txt
@@ -0,0 +1,21 @@
+query: A gold-plated initial necklace by Pavoi, 14k gold plated letter 'A' pendant on a dainty chain.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A gold-plated initial necklace by Pavoi, 14k gold plated letter 'A' pendant on a dainty chain..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A gold-plated initial necklace by Pavoi, 14k gold plated letter 'A' pendant on a dainty chain..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/kangaroo/u2sB3HohVoATCVWJQKDr1.png
diff --git a/split_1000_1500/image_1355.txt b/split_1000_1500/image_1355.txt
new file mode 100644
index 0000000000000000000000000000000000000000..b4b44c127f4ac0ce41eef2f65bd752e2cacc86e9
--- /dev/null
+++ b/split_1000_1500/image_1355.txt
@@ -0,0 +1,21 @@
+query: A set of crystal wine glasses by Riedel, Vinum Cabernet/Merlot set of 2, machine-blown lead crystal.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A set of crystal wine glasses by Riedel, Vinum Cabernet/Merlot set of 2, machine-blown lead crystal..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A set of crystal wine glasses by Riedel, Vinum Cabernet/Merlot set of 2, machine-blown lead crystal..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/panda/T9mD3IvQOalxfpH1B81Bc.png
diff --git a/split_1000_1500/image_1363.json b/split_1000_1500/image_1363.json
new file mode 100644
index 0000000000000000000000000000000000000000..2d6f1f0e4175ac58e1a9a50f42dba2a82cf3d7f9
--- /dev/null
+++ b/split_1000_1500/image_1363.json
@@ -0,0 +1 @@
+{"index": 1363, "query": "A memory foam neck pillow for travel, grey velour cover, includes earplugs and eye mask.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A memory foam neck pillow for travel, grey velour cover, includes earplugs and eye mask..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: Introduce a new object of the same category (e.g., similar type but different style, color, brand, or material) and place it on the same background.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/penguin/o3e1wltsF6nOGN0b-0CU-.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1372_5_twicemask.json b/split_1000_1500/image_1372_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..c898ca06475a1176dbad2d1bb4612a966afc7467
--- /dev/null
+++ b/split_1000_1500/image_1372_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1372,
+  "prompt": "query: A smart door lock by August, Wi-Fi Smart Lock (4th Gen) in silver, fits existing deadbolts.",
+  "mask_path": "objects/image_1372_5_twicemask.png",
+  "input_image": "objects/image_1372_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1378.json b/split_1000_1500/image_1378.json
new file mode 100644
index 0000000000000000000000000000000000000000..bc66da62edf1926f9f3916d2b918df4af395305e
--- /dev/null
+++ b/split_1000_1500/image_1378.json
@@ -0,0 +1 @@
+{"index": 1378, "query": "A wireless charging pad by Belkin, MagSafe 3-in-1 wireless charger for iPhone, Apple Watch, and AirPods in white.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A wireless charging pad by Belkin, MagSafe 3-in-1 wireless charger for iPhone, Apple Watch, and AirPods in white..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/penguin/HH_mZZSE8CXv8dCLjmIZc.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1378_5_twicemask.json b/split_1000_1500/image_1378_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..822749d42e5cae6303ca709c11f4fbd5899bf80f
--- /dev/null
+++ b/split_1000_1500/image_1378_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1378,
+  "prompt": "query: A wireless charging pad by Belkin, MagSafe 3-in-1 wireless charger for iPhone, Apple Watch, and AirPods in white.",
+  "mask_path": "objects/image_1378_5_twicemask.png",
+  "input_image": "objects/image_1378_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1381.json b/split_1000_1500/image_1381.json
new file mode 100644
index 0000000000000000000000000000000000000000..9652747b7ecc411db5b3b5bfe62de0b54d53a955
--- /dev/null
+++ b/split_1000_1500/image_1381.json
@@ -0,0 +1 @@
+{"index": 1381, "query": "A tube of sunscreen by EltaMD, UV Clear Broad-Spectrum SPF 46, 1.7 oz, oil-free for acne-prone skin.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A tube of sunscreen by EltaMD, UV Clear Broad-Spectrum SPF 46, 1.7 oz, oil-free for acne-prone skin..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/monkey/RQM3XvjRBlcFEbmargxQs.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1383.json b/split_1000_1500/image_1383.json
new file mode 100644
index 0000000000000000000000000000000000000000..910a85c2f5b43d738325446575e6708a9b87f344
--- /dev/null
+++ b/split_1000_1500/image_1383.json
@@ -0,0 +1 @@
+{"index": 1383, "query": "A bag of potting mix by Miracle-Gro, Indoor Potting Mix, 6 qt bag, blended for house plants.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A bag of potting mix by Miracle-Gro, Indoor Potting Mix, 6 qt bag, blended for house plants..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/panda/-snc7QgEybV9IK7_ZBTUA.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1385.txt b/split_1000_1500/image_1385.txt
new file mode 100644
index 0000000000000000000000000000000000000000..78b80146550d30b839466ebed579b2de5a8a7083
--- /dev/null
+++ b/split_1000_1500/image_1385.txt
@@ -0,0 +1,21 @@
+query: A smart thermostat sensor by Ecobee, SmartSensor 2-pack, detects temperature and occupancy.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A smart thermostat sensor by Ecobee, SmartSensor 2-pack, detects temperature and occupancy..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A smart thermostat sensor by Ecobee, SmartSensor 2-pack, detects temperature and occupancy..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "9:16", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/elephant/ip8b6zOlSx5YPjFjtdckB.png
diff --git a/split_1000_1500/image_1396.txt b/split_1000_1500/image_1396.txt
new file mode 100644
index 0000000000000000000000000000000000000000..d4382947ae24f9b6de10b47343f1e58c61a958b7
--- /dev/null
+++ b/split_1000_1500/image_1396.txt
@@ -0,0 +1,21 @@
+query: A set of luggage tags, 2-pack leather privacy tags with flap cover and stainless steel loop in black.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A set of luggage tags, 2-pack leather privacy tags with flap cover and stainless steel loop in black..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A set of luggage tags, 2-pack leather privacy tags with flap cover and stainless steel loop in black..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "16:9", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/penguin/BZKoTE4_4ow5sIpGGx_4F.png
diff --git a/split_1000_1500/image_1396_5_twicemask.json b/split_1000_1500/image_1396_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..ac5b3003d8fa71fe9ef6fcdade0dfacbf6d53071
--- /dev/null
+++ b/split_1000_1500/image_1396_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1396,
+  "prompt": "query: A set of luggage tags, 2-pack leather privacy tags with flap cover and stainless steel loop in black.",
+  "mask_path": "objects/image_1396_5_twicemask.png",
+  "input_image": "objects/image_1396_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1402_5_twicemask.json b/split_1000_1500/image_1402_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..558526915780a6ab2d65562159bb7f18d719794f
--- /dev/null
+++ b/split_1000_1500/image_1402_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1402,
+  "prompt": "query: A KitchenAid Artisan Series 5-Quart Stand Mixer, empire red, stainless steel bowl, tilt-head design with 10 speed settings.",
+  "mask_path": "objects/image_1402_5_twicemask.png",
+  "input_image": "objects/image_1402_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1403.txt b/split_1000_1500/image_1403.txt
new file mode 100644
index 0000000000000000000000000000000000000000..1fc77b13c1e907d55910e5fe09123d7e77f390cd
--- /dev/null
+++ b/split_1000_1500/image_1403.txt
@@ -0,0 +1,21 @@
+query: A pair of Ray-Ban Aviator Classic sunglasses, gold metal frame with crystal green G-15 lenses, standard size.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A pair of Ray-Ban Aviator Classic sunglasses, gold metal frame with crystal green G-15 lenses, standard size..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A pair of Ray-Ban Aviator Classic sunglasses, gold metal frame with crystal green G-15 lenses, standard size..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "5:4", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/rabbit/-tNmHuBKcG7ams_x_rpU8.png
diff --git a/split_1000_1500/image_1408_5_twicemask.json b/split_1000_1500/image_1408_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..edac1738dbf9fa1cf5269b99e2e3ee1fb4bd9691
--- /dev/null
+++ b/split_1000_1500/image_1408_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1408,
+  "prompt": "query: A Sony WH-1000XM5 wireless noise-canceling headphones, silver, soft fit leather, up to 30-hour battery life.",
+  "mask_path": "objects/image_1408_5_twicemask.png",
+  "input_image": "objects/image_1408_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1416_5_twicemask.json b/split_1000_1500/image_1416_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..053cb3208693f9793abd66580e295281ec4d57d3
--- /dev/null
+++ b/split_1000_1500/image_1416_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1416,
+  "prompt": "query: A Chanel No. 5 Eau de Parfum spray, 3.4 fl oz, floral aldehyde fragrance notes, iconic rectangular glass bottle.",
+  "mask_path": "objects/image_1416_5_twicemask.png",
+  "input_image": "objects/image_1416_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1423.json b/split_1000_1500/image_1423.json
new file mode 100644
index 0000000000000000000000000000000000000000..4b5ccea72bfd435ad3e1444d57e19efe2f265af5
--- /dev/null
+++ b/split_1000_1500/image_1423.json
@@ -0,0 +1 @@
+{"index": 1423, "query": "A Hydro Flask Wide Mouth water bottle with Straw Lid, 32 oz, pacific blue, stainless steel vacuum insulation.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Hydro Flask Wide Mouth water bottle with Straw Lid, 32 oz, pacific blue, stainless steel vacuum insulation..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/koala/28ULrGYaKHdakwoE765g_.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1425.txt b/split_1000_1500/image_1425.txt
new file mode 100644
index 0000000000000000000000000000000000000000..e30d09d36def8da5c60c7174ab44481afb6bcdb9
--- /dev/null
+++ b/split_1000_1500/image_1425.txt
@@ -0,0 +1,21 @@
+query: A Royal Canin Size Health Nutrition Medium Adult dry dog food, 30 lb bag, formulated for natural defenses and digestion.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Royal Canin Size Health Nutrition Medium Adult dry dog food, 30 lb bag, formulated for natural defenses and digestion..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Royal Canin Size Health Nutrition Medium Adult dry dog food, 30 lb bag, formulated for natural defenses and digestion..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/zebra/PETbS2-r0UQBInmqYl2MI.png
diff --git a/split_1000_1500/image_1425_5_twicemask.json b/split_1000_1500/image_1425_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..b11c9c1f0931790c7e2bcb0bf2473f73aa108bbf
--- /dev/null
+++ b/split_1000_1500/image_1425_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1425,
+  "prompt": "query: A Royal Canin Size Health Nutrition Medium Adult dry dog food, 30 lb bag, formulated for natural defenses and digestion.",
+  "mask_path": "objects/image_1425_5_twicemask.png",
+  "input_image": "objects/image_1425_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1429.json b/split_1000_1500/image_1429.json
new file mode 100644
index 0000000000000000000000000000000000000000..7ac204abed12c4f2fd133d55a22e324e2484cb0f
--- /dev/null
+++ b/split_1000_1500/image_1429.json
@@ -0,0 +1 @@
+{"index": 1429, "query": "A CeraVe Foaming Facial Cleanser, 16 oz pump bottle, for normal to oily skin, contains ceramides and hyaluronic acid.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A CeraVe Foaming Facial Cleanser, 16 oz pump bottle, for normal to oily skin, contains ceramides and hyaluronic acid..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/kangaroo/setIDW6G-J1rm1hZCu1jP.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1429_5_twicemask.json b/split_1000_1500/image_1429_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..167449114d23974c35a3d9bb36536a869fd89ccc
--- /dev/null
+++ b/split_1000_1500/image_1429_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1429,
+  "prompt": "query: A CeraVe Foaming Facial Cleanser, 16 oz pump bottle, for normal to oily skin, contains ceramides and hyaluronic acid.",
+  "mask_path": "objects/image_1429_5_twicemask.png",
+  "input_image": "objects/image_1429_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1463.json b/split_1000_1500/image_1463.json
new file mode 100644
index 0000000000000000000000000000000000000000..5747b24fb58598d9edbb8fb05799b84432621ef3
--- /dev/null
+++ b/split_1000_1500/image_1463.json
@@ -0,0 +1 @@
+{"index": 1463, "query": "A Black+Decker 20V MAX Cordless String Trimmer/Edger, orange and black, automatic feed spool, height adjustable.", "aspect_ratio": "3:4", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Black+Decker 20V MAX Cordless String Trimmer/Edger, orange and black, automatic feed spool, height adjustable..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:4", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/kangaroo/k0u9nOp7KUC-eROJn2tU-.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1464.json b/split_1000_1500/image_1464.json
new file mode 100644
index 0000000000000000000000000000000000000000..19056c4397dcafa8da5e08a09b1de8900e879b56
--- /dev/null
+++ b/split_1000_1500/image_1464.json
@@ -0,0 +1 @@
+{"index": 1464, "query": "A Muji Gel Ink Ballpoint Pen Set, 0.5mm, 10 colors, frosted barrel, smooth writing, cap style.", "aspect_ratio": "5:4", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Muji Gel Ink Ballpoint Pen Set, 0.5mm, 10 colors, frosted barrel, smooth writing, cap style..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "5:4", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/kangaroo/j8fLfvjVLL1QnoW_9du_-.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1468_5_twicemask.json b/split_1000_1500/image_1468_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..16cf870a2508999e39b4d34b9bbe04e213cb9f09
--- /dev/null
+++ b/split_1000_1500/image_1468_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1468,
+  "prompt": "query: A pair of Oakley Holbrook sunglasses, matte black frame, warm grey lenses, square shape, O-Matter stress-resistant frame.",
+  "mask_path": "objects/image_1468_5_twicemask.png",
+  "input_image": "objects/image_1468_2.png"
+}
\ No newline at end of file
diff --git a/split_1000_1500/image_1478.json b/split_1000_1500/image_1478.json
new file mode 100644
index 0000000000000000000000000000000000000000..e411dc37be6c78ca51a59b1614fbd33b160394f8
--- /dev/null
+++ b/split_1000_1500/image_1478.json
@@ -0,0 +1 @@
+{"index": 1478, "query": "A Tempur-Pedic TEMPUR-Cloud Pillow, standard size, extra soft memory foam, breathable knit cover.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Tempur-Pedic TEMPUR-Cloud Pillow, standard size, extra soft memory foam, breathable knit cover..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/elephant/DtW64RK0lbMLHqs9WpjyX.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1482.txt b/split_1000_1500/image_1482.txt
new file mode 100644
index 0000000000000000000000000000000000000000..a12fee97647d6881457554f3104ddfdbd6613ff9
--- /dev/null
+++ b/split_1000_1500/image_1482.txt
@@ -0,0 +1,21 @@
+query: A Kate Spade New York Spencer Saffiano Leather Dome Crossbody, black, gold spade logo, chain-link strap detail.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Kate Spade New York Spencer Saffiano Leather Dome Crossbody, black, gold spade logo, chain-link strap detail..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Kate Spade New York Spencer Saffiano Leather Dome Crossbody, black, gold spade logo, chain-link strap detail..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "5:4", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/monkey/jjNzyy1PkZaYypK-lTjb5.png
diff --git a/split_1000_1500/image_1483.txt b/split_1000_1500/image_1483.txt
new file mode 100644
index 0000000000000000000000000000000000000000..0d3cabe2c17329a0a793635dc8ca3041ab40ab78
--- /dev/null
+++ b/split_1000_1500/image_1483.txt
@@ -0,0 +1,21 @@
+query: A Dyson V15 Detect Cordless Vacuum, yellow/nickel, laser reveals microscopic dust, piezo sensor sizing.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Dyson V15 Detect Cordless Vacuum, yellow/nickel, laser reveals microscopic dust, piezo sensor sizing..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Dyson V15 Detect Cordless Vacuum, yellow/nickel, laser reveals microscopic dust, piezo sensor sizing..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "16:9", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/zebra/jhxGx0_kd73az9H8mS0tH.png
diff --git a/split_1000_1500/image_1485.json b/split_1000_1500/image_1485.json
new file mode 100644
index 0000000000000000000000000000000000000000..f900e1becf1cb4f38964ef22249f50696d5d825d
--- /dev/null
+++ b/split_1000_1500/image_1485.json
@@ -0,0 +1 @@
+{"index": 1485, "query": "A Coway Airmega AP-1512HH True HEPA Air Purifier, white, eco mode, air quality indicator, covers 361 sq. ft.", "aspect_ratio": "1:1", "arguments": {"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Coway Airmega AP-1512HH True HEPA Air Purifier, white, eco mode, air quality indicator, covers 361 sq. ft..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "1:1", "resolution": "4K", "output_format": "png"}, "image_url": "https://v3b.fal.media/files/b/zebra/3wZ4eDlvsuPv8Dxw2RyjN.png"}
\ No newline at end of file
diff --git a/split_1000_1500/image_1491.txt b/split_1000_1500/image_1491.txt
new file mode 100644
index 0000000000000000000000000000000000000000..13c3b0de6bfcce96c240f7ab609999a5418a03a3
--- /dev/null
+++ b/split_1000_1500/image_1491.txt
@@ -0,0 +1,21 @@
+query: A Brooklinen Luxe Sateen Core Sheet Set, queen size, graphite, 480-thread count, long-staple cotton.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Brooklinen Luxe Sateen Core Sheet Set, queen size, graphite, 480-thread count, long-staple cotton..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Brooklinen Luxe Sateen Core Sheet Set, queen size, graphite, 480-thread count, long-staple cotton..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:2", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/koala/Xxi5H1zhi2f9o9myoFSEt.png
diff --git a/split_1000_1500/image_1494.txt b/split_1000_1500/image_1494.txt
new file mode 100644
index 0000000000000000000000000000000000000000..490522b1b18fd9e64e431d0e34e459fa2656c11a
--- /dev/null
+++ b/split_1000_1500/image_1494.txt
@@ -0,0 +1,21 @@
+query: A Samsung 65-Inch Class OLED S90C Series TV, 4K UHD, Quantum HDR, LaserSlim design, solar cell remote.
+
+prompt:
+Generate a 2×2 image grid illustrating variations of an object of interest, described as A Samsung 65-Inch Class OLED S90C Series TV, 4K UHD, Quantum HDR, LaserSlim design, solar cell remote..
+Images in each grid should show the following:
+
+Top-left: The object displayed on a clean, white background.
+
+Top-right: The same object placed on a background that complements its appearance and make it attractive.
+
+Bottom-left: The same background as the top-right image, but without the object present.
+
+Bottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.
+
+Ensure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.
+Don't add any frames or bounders between the images.
+
+arguments:
+{"prompt": "\nGenerate a 2\u00d72 image grid illustrating variations of an object of interest, described as A Samsung 65-Inch Class OLED S90C Series TV, 4K UHD, Quantum HDR, LaserSlim design, solar cell remote..\nImages in each grid should show the following:\n\nTop-left: The object displayed on a clean, white background.\n\nTop-right: The same object placed on a background that complements its appearance and make it attractive.\n\nBottom-left: The same background as the top-right image, but without the object present.\n\nBottom-right: The segmentation mask of the object from the top-right image, shown as a solid white silhouette against a black background.If there is any shadow/reflection cast by the object onto other surfaces in the top-right image, label in green.\n\nEnsure consistent lighting, camera angle, and composition across all four images for a coherent and professional layout.\nDon't add any frames or bounders between the images.\n", "num_images": 1, "aspect_ratio": "3:2", "resolution": "4K", "output_format": "png"}
+
+image_url: https://v3b.fal.media/files/b/kangaroo/9mIJe9rIfW2OYjV6Fch-Y.png
diff --git a/split_1000_1500/image_1495_5_twicemask.json b/split_1000_1500/image_1495_5_twicemask.json
new file mode 100644
index 0000000000000000000000000000000000000000..028a1c3b206daeedea62b9a3679f82e50b982f40
--- /dev/null
+++ b/split_1000_1500/image_1495_5_twicemask.json
@@ -0,0 +1,6 @@
+{
+  "index": 1495,
+  "prompt": "query: A Yves Saint Laurent Rouge Pur Couture Lipstick, shade 01 Le Rouge, satin finish, gold metallic tube.",
+  "mask_path": "objects/image_1495_5_twicemask.png",
+  "input_image": "objects/image_1495_2.png"
+}
\ No newline at end of file