Building characters straight from MiniMax H3 T2V + R2VA, no more SDXL anchors
SIR_NVAX_A_LOT · reddit · 2026-08-17
The author ditched the old SDXL+KREA2 anchor-image pipeline and went straight to MiniMax H3's T2V plus R2VA for character generation: rooftop actors were incepted via T2V, cleaned up with V2V, then detailed with an anime reference for the pilot's plugsuit. The cockpit scene was generated purely from a prompt, with the HUD/mecha-kaiju referenced from a 15s battle render. They also experimented with transferring a Kaiju's look onto a woman's armor. One caveat: R2VA sometimes makes actors wider/squished, so FL2VA is safer if you don't want re-envisioning.
More from Multimodal
- Underwater brand films built from one jellyfish with an open-source Minimax template — w1kke · 2026-10-16
- Community speculation: Flux 3 Klein 5.6b is an upgrade of the 4b and a precursor to a 10b model — cradledust · 2026-10-16
- Higgsfield reneged on open-sourcing, so Comfy is offering $25K to rebuild its apps in the open — crystal_alpine · 2026-10-16
- HyperFrames launches Reference Any Video and auto-applied Brand Kits for AI video — Aiden_Tech_Ai · 2026-10-16
- T=16 flow model generates 16 frames per inference for smoother motion — pixlpa · 2026-10-16
- "AI can't make art"? Creator fires back with a 3-minute, $4 AI-generated drawing tutorial — flowersslop · 2026-10-16