Simple txt2img workflow fails to parse about cushystudio

[rvion]

want to join today's stream on https://discord.gg/GfAN6hF2ad ?
I'm going to fix this one live right now

[rvion]

Can't seem to find your lcm-optimizer-sd15.safetensors on the net.
any link to share (didn't found the file from https://latent-consistency-models.github.io/ )

without it, not quite sure I'm debugging the right thing because of this error

[oxysoft]

These weights https://huggingface.co/latent-consistency/lcm-lora-sdv1-5/blob/main/pytorch_lora_weights.safetensors I think you can just put in any LoRA tho, the output won't look right but that's fine for this repro

[oxysoft]

Yeah so actually I think literally every single image I ever rendered in ComfyUI doesn't open in CushyStudio with this same bug, so this is critical to figure out. Here is a simplified burger to help you

burger.json

[rvion]

Thanks ! I'll check this one out tomorrow

[rvion]

I think I'm onto something 🤣

[rvion]

I guess I just have to prioritize adding validation everywhere so I don't send time debugging things anymore, and just wait for people to send me nice screenshots of what error message they see 🙃

[oxysoft]

Haha yes indeed, game development taught me you can never have too much printing. I didn't like that error message very much myself, you should know exactly which node crashed it, or if it was outside of that, etc.

[rvion]

@oxysoft

honestly, I have no idea if I have fixed your issue or not, your burger.json never ran on my end.
Your "random prompt" node simply don't work on my ComfyUI install, with everything up-to-date.

Can you update your nodes and send me some other example with both Comfy and your node pack up to date ?

when I replace this node by a text primitive, it works.