Nvidia Cosmos 3 Super Image-to-Video API

NVIDIA Cosmos 3 Super Image-to-Video

NVIDIA Cosmos 3 Super Image-to-Video is NVIDIA's flagship image animation model built on the Cosmos 3 architecture. It transforms a single still image into a smooth, physically grounded video clip guided by a text prompt — delivering cinematic motion, accurate temporal coherence, and fine-grained control over frame count and pacing.

🌟 Key Features

🎬 Image-Driven Animation

Animates any still image into a fluid video sequence, preserving the original composition, subject identity, and visual style of the reference image.

📝 Prompt-Guided Motion

Accepts a text prompt to direct the type, direction, and style of motion — from subtle ambient movement to dynamic camera trajectories and subject actions.

🚫 Negative Prompting

Accepts a negative_prompt to explicitly suppress unwanted motion artifacts, distortions, or visual elements in the output video.

⏱ Flexible Duration Control

Set the output duration in whole seconds (1–7). Each second renders at 24 fps — pay only for the duration you generate.

🔁 Reproducible Results

Pin a seed value to reproduce or iterate on a specific generation, enabling systematic prompt and parameter refinement.

⚙️ Parameters

💲 Pricing

Billing is based on the duration parameter in seconds (e.g. 5s → $0.275, 7s →$0.385).

💡 Prompt Tips

The elephant slowly raises its trunk, gentle breeze moves the surrounding grass, cinematic slow push-in, golden hour lighting.

• Describe the motion explicitly: subject action, camera movement, and environmental dynamics.
• Use negative_prompt to suppress common issues: flickering, blur, distortion, sudden cuts, static.
• Keep guidance_scale between 5–7 for natural motion; increase to 8–10 for strict prompt adherence.
• Use shorter duration (1–2s) for quick previews before committing to a full-length generation.
• Fix seed when iterating on the prompt to isolate the effect of each wording change.

🎯 Use Cases

• Product Animation — Bring static product photos to life with subtle motion and environmental effects.
• Cinematic Previsualization — Quickly animate storyboard frames or concept art to preview camera moves and scene timing.
• Social Media Content — Generate eye-catching looping video clips from still images for any platform.
• Creative Storytelling — Animate illustrations, paintings, or photographs into dynamic narrative sequences.
• Virtual Production — Produce physically grounded environment and character animations for pre-production workflows.
• Marketing & Advertising — Transform campaign visuals into dynamic video assets without full video production.

📝 Notes

• Both prompt and image_url are required fields.
• image_url must be a publicly accessible URL or a Base64-encoded image string.
• Each second of video renders at 24 fps. Max duration is 7 seconds (168 frames).
• Task status values: created, processing, completed, failed.
• Generated video URLs are returned in outputs once status is completed.

🔗 Related Models

• NVIDIA Cosmos 3 Super Text-to-Image — Generate photorealistic images from text prompts using the Cosmos 3 architecture.