feat: add sync-v3 lipsync model

[SecurityQQ]

Summary

• Add sync-v3 (fal-ai/sync-lipsync/v3) to the SDK's LIPSYNC_MODELS dictionary
• Add sync-v3 to VIDEO_ONLY_LIPSYNC_MODELS set in the TalkingHead renderer (video-only model, uses the two-step animate-then-lipsync pipeline)

Details

Sync-3 is Sync.so's most powerful lipsync model. It takes video_url + audio_url inputs and produces lipsynced video. Same input pattern as sync-v2 / sync-v2-pro.

Optional sync_mode parameter (cut_off, loop, bounce, silence, remap) is automatically supported via providerOptions.fal.

Depends on gateway PR for routing and pricing.

[coderabbitai]

Caution

Review failed

Pull request was closed or merged during review

📝 Walkthrough

Walkthrough

added git ignores for temp files and a scratch tsx file, registered a new sync-v3 lipsync model with the fal endpoint, and marked it as video-only in the talking-head renderer to ensure proper animation pipeline routing.

Changes

Cohort / File(s)	Summary
git config .gitignore	Added ignore patterns for tmp/ directory and hello.tsx scratch file.
lipsync model setup src/ai-sdk/providers/fal.ts, src/react/renderers/talking-head.ts	Registered new sync-v3 lipsync model mapping to fal-ai/sync-lipsync/v3 endpoint and marked it as video-only, triggering the image→video→lipsync pipeline instead of direct image lipsync.

Possibly related PRs

• feat: implement TalkingHead renderer with image/audio props #189: added talking-head renderer with lipsyncmodel handling that directly uses the constants modified here
• feat(fal): add OmniHuman v1.5 and VEED Fabric 1.0 #141: extends fal lipsync mappings and influences video-vs-image lipsync routing logic
• feat: add LTX-2 19B audio-to-video model (ltx-2-a2v) #193: adds new lipsync model-to-endpoint mappings in the same fal provider file

Poem

✨ sync-v3 arrives, video-first and bright 🎬
lipsync flows through the animated light ✨
git keeps the temps tucked away nice and clean
a sleek little model joins the lipsync machine 🎭

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	the title clearly summarizes the main change: adding the sync-v3 lipsync model across multiple files.
Description check	✅ Passed	the description directly relates to the changeset, covering the sync-v3 model addition, its configuration, and implementation details.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feature/sync-v3-lipsync

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}