73 min, 16 fps 81 frames, 1024 ref max size, and a black output video lot of things that aren't good
73 min, 16 fps 81 frames, 1024 ref max size, and a black output video lot of things that aren't good
That’s bad. usually, black video is caused by env. misconfiguration sometimes enabling Sage Attention or setting FP16 accumulation to true can also cause black video.
I generated a 5 sec video using the NVFP4 models and no black video.
As of now, ComfyUI team hasn’t implemented support for Bernini v2, Let’s hope they do soon
That’s bad. usually, black video is caused by env. misconfiguration sometimes enabling Sage Attention or setting FP16 accumulation to true can also cause black video.
I generated a 5 sec video using the NVFP4 models and no black video.As of now, ComfyUI team hasn’t implemented support for Bernini v2, Let’s hope they do soon
but still 73 minutes? where does it come from this slowness?
but still 73 minutes? where does it come from this slowness?
One thing’s for sure you’re a very patient man. for me, if a video is around 2 to 3 minutes or so, that’s fine otherwise I wll immediately delete the model and start a new life all over again 😃
I tried the old LightX2V LoRA to see if there was something we could do about it, but those high and low noises are different from V1. When it comes to using fewer steps, we’re at the mercy of the LightX2V team or another organization with the resources else, it is what it is 😃