xAI · External model reference
Grok Imagine Video 1.5
See how Grok Imagine Video 1.5 handles voice, start images and references, then evaluate an audio-led scene.
This model is covered for comparison. If you want to start in Oakgen, use Seedance 2.5 for the workflow below.
Explore Seedance 2.5 on OakgenInputs and controls
- Text or opening still
- fal documents 1–15 seconds and up to 1080p for text-to-video and image-to-video. The image workflow inherits the image's aspect ratio.
- Reference video workflow
- The reference endpoint uses image references without fixing the start frame. Check its resolution options separately from text and image generation.
Run a side-by-side test
Use a recording-free visual brief of a person at a ticket counter who says one short line before a mechanical stamp lands. Generate a version from a starting still and a version with the same person as a reference image.
Transcribe the spoken line before judging the picture. Check the last mouth movement against the end of the line and the stamp sound against contact on the desk. Score the two audio events independently.
This is a proposed test. Keep the settings and each attempt, including failures. Compare spend per usable output using total spend divided by outputs that pass your checks.
Compare with Seedance 2.5
Run the same dialogue-and-action brief in its native-audio workflow. Confirm the voice and sound can both be heard before scoring sync.
Sources
Reviewed .