Skip to content

feat(video): add SEF (--last-frame), S2V (--subject-image), model auto-switch#74

Merged
tars90percent merged 2 commits intoMiniMax-AI:mainfrom
raylanlin:pr/video
Apr 10, 2026
Merged

feat(video): add SEF (--last-frame), S2V (--subject-image), model auto-switch#74
tars90percent merged 2 commits intoMiniMax-AI:mainfrom
raylanlin:pr/video

Conversation

@raylanlin
Copy link
Copy Markdown

Changes

  • --last-frame: Add last frame image for SEF (start-end frame) interpolation mode
    • Auto-switches to Hailuo-02 model (unless --model explicitly set)
    • Requires --first-frame
  • --subject-image: Add subject reference for character consistency (S2V-01)
    • Auto-switches to S2V-01 model (unless --model explicitly set)
  • Model auto-switch: Explicit --model flag always takes priority
  • New examples: SEF and S2V usage

Files

  • src/commands/video/generate.ts
  • src/types/api.ts (add last_frame_image, subject_reference)
  • src/command.ts (apiDocs interface)

Raylan LIN and others added 2 commits April 10, 2026 16:41
…o-switch

- Add --last-frame flag for SEF (start-end frame) interpolation mode
- Add --subject-image flag for subject reference (S2V-01 character consistency)
- Auto-switch model: --last-frame -> Hailuo-02, --subject-image -> S2V-01
- Explicit --model flag always overrides auto-switch
- Add apiDocs field
- Update description with all available models (T2V/I2V/S2V)
- Add last_frame_image and subject_reference to VideoRequest type
- Add validation: --last-frame requires --first-frame
- Add SEF and S2V examples
…exclusion

- Change apiDocs from full URL to path (fixes broken URL in --help)
- Remove stray '+' in I2V description
- Error when --last-frame and --subject-image are combined (different modes)
@tars90percent tars90percent merged commit 9946f25 into MiniMax-AI:main Apr 10, 2026
1 check passed
@tars90percent
Copy link
Copy Markdown
Collaborator

Thanks! Pushed a fix commit (881d822) before merging:

  • apiDocs path fix (same as other PRs -- full URLs broke when concatenated with DOCS_HOSTS[region])
  • Fixed stray + in the description (I2V: + I2V-01I2V: I2V-01)
  • Added a check so --last-frame and --subject-image can't be used together since they're different modes (SEF vs S2V) targeting different models and request fields

Apologies for pushing directly rather than requesting changes -- should have let you address these yourself. Will do that next time.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants