Make image-to-video example use the negative prompt by default

Replaces the plain 'Image to video' example with the negative-prompt variant so there is a single image-to-video section that passes the curated assets/negative_prompt.json as the per-sample negative prompt (JSON-encoded string). Drops the separate 'Image to video with negative prompt' section and points the example output at assets/edge_i2v_negprompt_output.mp4.

README.md

@@ -481,42 +481,7 @@ Cosmos3-Edge is a compact omni model that fits on a single GPU and is the recomm
 
 #### Image to video
 
-```shell
-BASE="https://huggingface.co/nvidia/Cosmos3-Edge/resolve/main/assets"
-python - "$BASE" <<'PY'
-import json, sys, urllib.request
-base = sys.argv[1]
-prompt = urllib.request.urlopen(f"{base}/example_i2v_prompt.json").read().decode()
-json.dump(
-    {"model_mode": "image2video", "prompt": prompt, "vision_path": f"{base}/example_i2v_input.jpg"},
-    open("i2v_edge.json", "w"), indent=2,
-)
-PY
-
-python -m cosmos_framework.scripts.inference \
-    --parallelism-preset=latency \
-    -i i2v_edge.json \
-    -o outputs/omni_edge \
-    --checkpoint-path Cosmos3-Edge \
-    --resolution 480 \
-    --num-frames 121 \
-    --fps 24 \
-    --num-steps 50 \
-    --guidance 5.0 \
-    --shift 3.0 \
-    --sampler unipc \
-    --seed 0
-```
-
-The generated video is written to `outputs/omni_edge/<sample_name>/vision.mp4`.
-
-Example output:
-
-<video controls width="832" height="480" src="https://huggingface.co/nvidia/Cosmos3-Edge/resolve/main/assets/edge_i2v_output.mp4"></video>
-
-#### Image to video with negative prompt
-
-The same image-to-video setup as above, but with a structured **negative prompt** to steer the model away from common failure modes (temporal flicker, morphing geometry, broken physics, unstable camera push-in). We reuse the same input image and prompt, and pass the curated `negative_prompt.json` from `assets/` as the per-sample `negative_prompt` (JSON-encoded as a string, mirroring the model's default negative caption format).
+Image-to-video generation steered by a structured **negative prompt** that keeps the model away from common failure modes (temporal flicker, morphing geometry, broken physics, unstable camera push-in). We pass the curated `negative_prompt.json` from `assets/` as the per-sample `negative_prompt` (JSON-encoded as a string, mirroring the model's default negative caption format).
 
 ```shell
 BASE="https://huggingface.co/nvidia/Cosmos3-Edge/resolve/main/assets"
@@ -532,14 +497,14 @@ json.dump(
         "vision_path": f"{base}/example_i2v_input.jpg",
         "negative_prompt": json.dumps(json.loads(negative)),
     },
-    open("i2v_edge_neg.json", "w"), indent=2,
+    open("i2v_edge.json", "w"), indent=2,
 )
 PY
 
 python -m cosmos_framework.scripts.inference \
     --parallelism-preset=latency \
-    -i i2v_edge_neg.json \
-    -o outputs/omni_edge_neg \
+    -i i2v_edge.json \
+    -o outputs/omni_edge \
     --checkpoint-path Cosmos3-Edge \
     --resolution 480 \
     --num-frames 121 \
@@ -551,7 +516,7 @@ python -m cosmos_framework.scripts.inference \
     --seed 0
 ```
 
-The generated video is written to `outputs/omni_edge_neg/<sample_name>/vision.mp4`.
+The generated video is written to `outputs/omni_edge/<sample_name>/vision.mp4`.
 
 Example output: