I2V workflow to run on RunPod H100 – generating 6–10 second near-photorealistic videos from a photo with action guided by a text prompt.

- I2V workflow to run on RunPod H100 – generating 6–10 second near-photorealistic videos from a photo with action guided by a text prompt. - Reasonable render time for 10-second video outputs at 480p resolution. Requirements: - Input: 1 or 2 images (e.g., photos and/or a short reference video) + a text prompt to guide the action/motion. - Output: Short video (6–10 seconds) delivered as a direct, clean MP4 file. - Dialogue: - Converts dialogue written in tagged form (e.g., Bill: "Hello!") into spoken audio, with lines assigned to the tagged person and matching mouth movements (functional and convincing lip-sync is acceptable). Other methods are fine if they produce the same result. - Enables tone (via tags is fine) such as whisper, shout, softly, or harshly. - Accent selection not required. - Only one active speaker per continuous shot; cuts are allowed. - Near-Photorealistic Quality: Motion should appear near-photorealistic at normal viewing distance; still-frame perfection is not required. - Multi-Person Capability: Supports scenes with at least three people staying recognizable (not identical pixel-perfect faces) and correctly placed, with good face and character consistency across all frames. - Uncensored Workflow: No content filtering or moderation layers imposed by the workflow or models. - All download links provided. Deliverables: - Workflow JSON file optimized for a RunPod H100. - A working, tested RunPod H100 setup that runs the workflow as described (RunPod will be supplied). - Short operating guide covering settings and use. Budget & Timing: - $1,000 USD. - Soon – ready to begin.

Projekt-ID: 40085557