Could you share the expected timeline for releasing the 8‑step turbo LoRA for Ref2VA?
Could you share the expected timeline for releasing the 8‑step turbo LoRA for Ref2VA?
We are using Ref2VA for comic‑drama video generation on RTX 5060Ti 16G, and really looking forward to this acceleration LoRA.We also hope the 8-step LoRA can maintain stable character identity and reduce motion artifacts in action scenes, especially for fast sword fights and combat sequences.
one small tip i read somewhere. you can use the fl2va model in ref2va workflow with no problem. try this out 😄
The fl2va turbo lora model works in ref2v.
Try this : Main diffusion model connected to minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors lora and minimax_h3_ref_lora_rank_256_bf16.safetensors lora by Kijai. Works just fine at 8 - 10 steps.
The fl2va turbo lora model works in ref2v.
Try this : Main diffusion model connected to minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors lora and minimax_h3_ref_lora_rank_256_bf16.safetensors lora by Kijai. Works just fine at 8 - 10 steps.
It does not work for me, It reports a problem with Kijai lora