Update and try!
This sounds amazing. Any hope for MAC users?
There are two critical mistakes in that T2V workflow if you use the second set of nodes without lightx2v LoRA:
- There's a missing link the between the second KSampler's Latent output and VAE Decode's samples input.
- The first KSampler's "end at step" should be 10 rather than 20.
I have a question :)
I have upgraded comfyUI but I can't see the template available in the list
is it normal ? thanks a lot
Reinstall the pip requirements
you genius! it worked! thanks again
Excellent. It was really necessary.
Can I run 14B T2V model on 12gb VRAM card? (RTX 4070 Super) and 64 GB RAM?
If I can run the example T2V model at 384x256 on 4gb VRAM card, you can definitely do that!
This sounds amazing. Any hope for MAC users?
There are two critical mistakes in that T2V workflow if you use the second set of nodes without lightx2v LoRA:
- There's a missing link the between the second KSampler's Latent output and VAE Decode's samples input.
- The first KSampler's "end at step" should be 10 rather than 20.
I have a question :)
I have upgraded comfyUI but I can't see the template available in the list
is it normal ? thanks a lot
Reinstall the pip requirements
you genius! it worked! thanks again
Excellent. It was really necessary.
Can I run 14B T2V model on 12gb VRAM card? (RTX 4070 Super) and 64 GB RAM?
If I can run the example T2V model at 384x256 on 4gb VRAM card, you can definitely do that!