Feature Request: WAN 2.5 Fast Model Support in AI Studio (Outstanding for Singing Lip Sync)
Hi Reallusion team,
First off, I’m really impressed with AI Studio and the 3D‑guided workflow from iClone/Character Creator into models like Seedance 2.0, Wan 2.2 Fun Control, LTX 2/2.3, and Scail for 3D‑to‑video generation. The integration with Seedance 2.0 in particular is a strong step toward high‑quality, director‑controlled cinematic output.
For context, I’ve been actively testing many AI lip sync and performance models across platforms such as Higgsfield and OpenArt, specifically focused on singing performance: believable mouth shapes, expressive faces, and consistent characters over multiple shots. In my hands‑on testing, WAN 2.5 fast has delivered the best singing lip sync performance for character‑driven content:
- Mouth shapes track sung phonemes very convincingly, even on more complex vocal lines
- Facial expressions and micro‑expressions feel musically “aware” and emotionally coherent
- Body motion supports the performance instead of distracting from it
- Character identity remains acceptably consistent across takes and variations
In short, Wan 2.5 fast is currently my top choice when I need a singer that looks and feels like a real performer, not just a talking head.
Given that AI Studio is positioned as a multi‑model platform (Flux, Nano Banana, Veo 3, Kling AI, Wan, LTX, Scail, Seedance 2.0, etc.), I’d love to know:
- Are there any roadmap plans to add WAN 2.5 (Image‑to‑Video Fast) as a selectable model inside AI Studio?
- If yes, is there any rough timeline you can share (even something like “on the radar,” “under evaluation,” or “not planned at this time”)?
- If no current plans exist, are there particular technical, licensing, or cost constraints that make WAN 2.5 fast a poor fit for AI Studio’s architecture or pricing?
From a user perspective, WAN 2.5 fast would complement Seedance 2.0 and Wan 2.2 Fun Control very well, especially for:
- Music‑driven content where singing lip sync and expressive performance are critical
- Rapid iteration on performance passes for 3D‑blocked shots before sending hero shots to Seedance 2.0
- Creators who want multiple “performance engines” per shot within the same AI Studio pipeline
I’d be very interested in helping test WAN 2.5 fast in an early access or beta context, especially for singing and music‑driven character work.
Thanks for all the work you’re putting into AI Studio and for listening to community feedback.
Best regards,
nickArthur