wehub/bytedance--lance

Files

T

wehub-resource-sync e710618da0 chore: import upstream snapshot with attribution

2026-07-13 13:16:54 +08:00

1.5 KiB

Raw Permalink Blame History

2026-05-29

Updates

Image-to-Video generation: We have added support for Image-to-Video generation.
Prompt-conditioned motion: The task uses an input first-frame image together with a text prompt, keeping the image content as the visual anchor while generating temporally coherent motion.
Optional prompt enhancement: ENHANCE_PROMPT can be enabled for T2V/I2V prompt rewrite. For I2V, the rewrite uses both the input text and the first-frame image.

bash inference_lance.sh \
  --TASK_NAME i2v \
  --MODEL_PATH downloads/Lance_3B_Video \
  --RESOLUTION video_480p \
  --NUM_FRAMES 61 \
  --VIDEO_HEIGHT 480 \
  --VIDEO_WIDTH 848 \
  --SAVE_PATH_GEN results/i2v

Examples

First Frame	Generated Video