LTXVNormalizingSampler error

#63
by arquiplay77 - opened

Please excuse my complete lack of knowledge in this area; I'm trying to learn with your workflow, however, I keep getting this error and no matter how hard I try to fix it, I can't, basically because I don't understand anything.

"LTXVNormalizingSampler
The size of tensor a (9360) must match the size of tensor b (1214336) at non-singleton dimension 2"

I understand is in the LTXV Normalizing Sampler node. I have everything as you workflow, not changing anything.

Any help will be great
Thank you

You might not have all of the model pieces selected and loaded correctly. I think the "gemma" models are often to blame. Check my original model and make sure all of the model parts (vae, gemma text encoder, connector etc.) all align.

Thank you for the fast response man.

I checked all and can't find the problem on the models paths,
Let me show you how i got them on my folders to double check please.

diffusion

text encoders

VAE

workflow

Is there something i'm missing??

Nothing is really jumping out at me here, unless something funky is going on in the I2V / first to last frame section at the top.

base on my expericence, it's because a another nodes something like "sMnodes" , if you remove this nodes from comfyUI, this issue will be resolved

Sign up or log in to comment