Create app.py

app.py

@@ -0,0 +1,357 @@
+import gradio as gr
+import json
+
+def build_json(
+    api_key,
+    reference_image,
+    scene,
+    subject_type,
+    age_range,
+    hair,
+    makeup,
+    jewellery,
+    top,
+    bottom,
+    footwear,
+    wardrobe_notes,
+    pose_angle,
+    body_pose,
+    hands_pose,
+    framing,
+    camera_device,
+    flash,
+    orientation,
+    aspect_ratio,
+    distance,
+    focus,
+    texture,
+    sharpness,
+    color,
+    effects,
+    background_environment,
+    background_props,
+    style_genre,
+    authenticity,
+    use_original_structure,
+    face_description,
+    ban_mirror,
+    ban_phone,
+    ban_selfie,
+    ban_grainy,
+    ban_harsh_flash,
+    ban_logos,
+    ban_nsfw,
+    ban_cropped_feet,
+    output_count,
+    output_size,
+    safety,
+    variant_name,
+    variant_angle,
+):
+    banned_items = []
+    if ban_mirror:
+        banned_items.append("mirror")
+    if ban_phone:
+        banned_items.append("phone")
+    if ban_selfie:
+        banned_items.append("selfie look")
+    if ban_grainy:
+        banned_items.append("grainy noise")
+    if ban_harsh_flash:
+        banned_items.append("harsh LED flash")
+    if ban_logos:
+        banned_items.append("logos/brand text")
+    if ban_nsfw:
+        banned_items.append("nsfw")
+    if ban_cropped_feet:
+        banned_items.append("cropped feet")
+
+    output_json = {
+        "scene": scene,
+        "subject": {
+            "type": subject_type,
+            "age_range": age_range,
+            "hair": hair,
+            "makeup": makeup,
+            "jewellery": jewellery,
+        },
+        "wardrobe": {
+            "top": top,
+            "bottom": bottom,
+            "footwear": footwear,
+            "notes": wardrobe_notes,
+        },
+        "pose": {
+            "angle": pose_angle,
+            "body": body_pose,
+            "hands": hands_pose,
+            "framing": framing,
+        },
+        "camera": {
+            "device": camera_device,
+            "flash": flash,
+            "orientation": orientation,
+            "aspect_ratio": aspect_ratio,
+            "distance": distance,
+            "focus": focus,
+        },
+        "look": {
+            "texture": texture,
+            "sharpness": sharpness,
+            "color": color,
+            "effects": effects,
+        },
+        "background": {
+            "environment": background_environment,
+            "props": background_props,
+        },
+        "style": {"genre": style_genre, "authenticity": authenticity},
+        "reference_face": {
+            "use_original_structure": use_original_structure,
+            "description": face_description,
+        },
+        "ban": banned_items,
+        "output": {
+            "count": output_count,
+            "size": output_size,
+            "safety": safety
+        },
+        "variants": [{"name": variant_name, "angle": variant_angle}],
+    }
+
+    # The user's API key is available in the 'api_key' variable.
+    # You can now use this key and the reference_image path to call the
+    # gemini nano banana API with the generated 'output_json'.
+
+    # For this example, we will just return the generated JSON.
+    return json.dumps(output_json, indent=4)
+
+
+with gr.Blocks() as demo:
+    gr.Markdown("# Gemini Nano Banana Image Generator Interface")
+    gr.Markdown(
+        "Fill in the details below to generate the JSON input for the image generation model."
+    )
+
+    with gr.Tabs():
+        with gr.TabItem("Authentication and Image"):
+            api_key_input = gr.Textbox(
+                label="Gemini Nano Banana API Key", type="password"
+            )
+            reference_image_input = gr.Image(
+                label="Reference Image", type="filepath"
+            )
+
+        with gr.TabItem("Scene and Subject"):
+            with gr.Row():
+                scene_input = gr.Textbox(
+                    label="Scene", value="cinematic outdoor portrait; professional photography"
+                )
+            with gr.Row():
+                subject_type_input = gr.Textbox(
+                    label="Subject Type", value="adult woman (idol vibe)"
+                )
+                age_range_input = gr.Textbox(label="Age Range", value="20s")
+            with gr.Row():
+                hair_input = gr.Textbox(
+                    label="Hair",
+                    value="straight or styled natural open hair with natural shine",
+                )
+                makeup_input = gr.Textbox(
+                    label="Makeup", value="glossy lips, soft eyeliner, luminous skin"
+                )
+                jewellery_input = gr.Textbox(
+                    label="Jewellery", value="small hoops, thin chain, subtle bracelets"
+                )
+
+        with gr.TabItem("Wardrobe"):
+            with gr.Row():
+                top_input = gr.Textbox(label="Top", value="basic tee or camisole")
+                bottom_input = gr.Textbox(
+                    label="Bottom", value="denim shorts or mini skirt"
+                )
+                footwear_input = gr.Textbox(
+                    label="Footwear", value="sneakers or ankle boots"
+                )
+            with gr.Row():
+                wardrobe_notes_input = gr.Textbox(
+                    label="Wardrobe Notes",
+                    value="casual modern look, styled for natural setting",
+                )
+
+        with gr.TabItem("Pose and Framing"):
+            with gr.Row():
+                pose_angle_input = gr.Dropdown(
+                    label="Pose Angle",
+                    choices=["three-quarter", "full body"],
+                    value="three-quarter",
+                )
+                body_pose_input = gr.Textbox(
+                    label="Body Pose",
+                    value="standing or walking casually, relaxed natural posture",
+                )
+            with gr.Row():
+                hands_pose_input = gr.Textbox(
+                    label="Hands Pose",
+                    value="one resting by side or touching hair, the other relaxed",
+                )
+                framing_input = gr.Dropdown(
+                    label="Framing",
+                    choices=["head-to-toe", "waist-up"],
+                    value="waist-up",
+                )
+
+        with gr.TabItem("Camera and Look"):
+            with gr.Row():
+                camera_device_input = gr.Textbox(
+                    label="Camera Device",
+                    value="professional cinema camera / DSLR with prime lens",
+                )
+                flash_input = gr.Textbox(
+                    label="Flash", value="none; natural golden hour light or soft reflectors"
+                )
+            with gr.Row():
+                orientation_input = gr.Dropdown(
+                    label="Orientation", choices=["vertical", "horizontal"], value="vertical"
+                )
+                aspect_ratio_input = gr.Dropdown(
+                    label="Aspect Ratio", choices=["16:9", "3:2", "4:3", "1:1"], value="16:9"
+                )
+            with gr.Row():
+                distance_input = gr.Textbox(
+                    label="Distance", value="cinematic portrait distance with shallow depth"
+                )
+                focus_input = gr.Textbox(
+                    label="Focus", value="sharp on subject; soft bokeh background"
+                )
+            with gr.Row():
+                texture_input = gr.Textbox(
+                    label="Texture", value="smooth high-resolution detail"
+                )
+                sharpness_input = gr.Textbox(
+                    label="Sharpness", value="very high; crisp cinematic clarity"
+                )
+            with gr.Row():
+                color_input = gr.Textbox(
+                    label="Color", value="warm cinematic grading; golden tones and soft contrast"
+                )
+                effects_input = gr.Textbox(
+                    label="Effects",
+                    value="subtle film grain; natural light flares, depth of field",
+                )
+
+        with gr.TabItem("Background and Style"):
+            with gr.Row():
+                background_environment_input = gr.Textbox(
+                    label="Background Environment",
+                    value="nature setting — forest, park, or meadow with soft light",
+                )
+                background_props_input = gr.Textbox(
+                    label="Background Props", value="none; focus on subject against natural backdrop"
+                )
+            with gr.Row():
+                style_genre_input = gr.Textbox(
+                    label="Style Genre", value="cinematic portrait photography"
+                )
+                authenticity_input = gr.Textbox(
+                    label="Authenticity", value="natural, elegant, polished"
+                )
+
+        with gr.TabItem("Reference Face and Bans"):
+            with gr.Row():
+                use_original_structure_input = gr.Checkbox(
+                    label="Use Original Face Structure", value=True
+                )
+            with gr.Row():
+                face_description_input = gr.Textbox(
+                    label="Face Description",
+                    value="maintain the same face shape, features, and proportions as in the provided reference image",
+                )
+            with gr.Row():
+                gr.Markdown("#### Banned Items")
+            with gr.Row():
+                ban_mirror_input = gr.Checkbox(label="Mirror")
+                ban_phone_input = gr.Checkbox(label="Phone")
+                ban_selfie_input = gr.Checkbox(label="Selfie Look")
+                ban_grainy_input = gr.Checkbox(label="Grainy Noise")
+            with gr.Row():
+                ban_harsh_flash_input = gr.Checkbox(label="Harsh LED Flash")
+                ban_logos_input = gr.Checkbox(label="Logos/Brand Text")
+                ban_nsfw_input = gr.Checkbox(label="NSFW")
+                ban_cropped_feet_input = gr.Checkbox(label="Cropped Feet")
+
+        with gr.TabItem("Output and Variants"):
+            with gr.Row():
+                output_count_input = gr.Slider(
+                    label="Output Count", minimum=1, maximum=10, step=1, value=1
+                )
+                output_size_input = gr.Textbox(label="Output Size", value="1920x1080")
+                safety_input = gr.Dropdown(
+                    label="Safety", choices=["strict", "moderate", "none"], value="strict"
+                )
+            with gr.Row():
+                variant_name_input = gr.Textbox(
+                    label="Variant Name", value="cinematic_nature_fullbody"
+                )
+                variant_angle_input = gr.Textbox(
+                    label="Variant Angle",
+                    value="full-body shot in meadow or forest path, subject centered with depth of field",
+                )
+
+    generate_button = gr.Button("Generate JSON")
+    json_output = gr.JSON(label="Generated JSON")
+
+    generate_button.click(
+        fn=build_json,
+        inputs=[
+            api_key_input,
+            reference_image_input,
+            scene_input,
+            subject_type_input,
+            age_range_input,
+            hair_input,
+            makeup_input,
+            jewellery_input,
+            top_input,
+            bottom_input,
+            footwear_input,
+            wardrobe_notes_input,
+            pose_angle_input,
+            body_pose_input,
+            hands_pose_input,
+            framing_input,
+            camera_device_input,
+            flash_input,
+            orientation_input,
+            aspect_ratio_input,
+            distance_input,
+            focus_input,
+            texture_input,
+            sharpness_input,
+            color_input,
+            effects_input,
+            background_environment_input,
+            background_props_input,
+            style_genre_input,
+            authenticity_input,
+            use_original_structure_input,
+            face_description_input,
+            ban_mirror_input,
+            ban_phone_input,
+            ban_selfie_input,
+            ban_grainy_input,
+            ban_harsh_flash_input,
+            ban_logos_input,
+            ban_nsfw_input,
+            ban_cropped_feet_input,
+            output_count_input,
+            output_size_input,
+            safety_input,
+            variant_name_input,
+            variant_angle_input,
+        ],
+        outputs=json_output,
+    )
+
+if __name__ == "__main__":
+    demo.launch()