ref2v lora blurs the hands and the overall quality is bad.
This lora messes the hands, they become blurry. its bad, the normal fl2v lightx2v didn't do this in a ref2v workflow. just swapping it in - same settings and this specific ref2v lora is bad. the quality is terrible for the overall ref2v video output. Maybe i am doing something wrong? but same settings work fine for the fl2v lightx2v lora in a ref2v workflow.
"blurs" hands for me as well, really just bad rendering of hands.
I think the best version is minimax_h3_fl2v_lightx2v_turbo_4step_v0.1_comfy.safetensors by Kijai. I use this for both fl2V and red2V. It's the most detailed in my opinion. All other versions seems blurry specially when there's a fast motion in the scene.
I think the best version is minimax_h3_fl2v_lightx2v_turbo_4step_v0.1_comfy.safetensors by Kijai. I use this for both fl2V and red2V. It's the most detailed in my opinion. All other versions seems blurry specially when there's a fast motion in the scene.
You are right. That's the one I use. That's the best one.
I found it better with er_sde and beta.
I had no idea the fl2V model works in the ref2V workflow. I tried it with the turbo LoRa, results look much clearer.