I get this error on 75 percent of my LTX workflows, I have checked my syntax, I have made sure all the correct models are in place. Anyone got a fix for this, I’m sure it’s something simple I’m missing.
I’ve been having this same issue and it’s taken me a while to work out what was wrong but I’ve now found out what was causing the problem for me. It was to do with my download urls. I was using wget to download my models from huggingface, but for some of them, the url was slightly wrong.
For example, when my script was downloading the wan_2.1_vae I was a hugging face url ending like this:
…/Wan_2.2_ComfyUI_Repackaged/blob/main/split_files/vae/wan_2.1_vae.safetensors
But this is the url to the download page and not the actual model. So my script was actually downloading the HTML of the download page and then the ComfyUI workflow was trying to parse that as a vae model, so of course json decoding failed.
I changed the download url to end like this:
…/Wan_2.2_ComfyUI_Repackaged/resolve/main/split_files/vae/wan_2.1_vae.safetensors
Essentially it just changed “blob” to “resolve” which, when navigated to, starts the download instead of taking the user to the download page.
This fixed my issue.
Whatever the cause of your issue it’s basically saying that your model is in an incorrect format so check the file sizes of the models it’s complaining about and make sure they are the size they should be. If the download didn’t complete for these models they won’t work and if you had a wrong url like I did, it looks like the model is downloaded but actually something else was.
If it’s failing for 75 percent of your workflows then those workflows have some models in common which are the problem ones. If necessary, delete those models and redownload them.
I hope this helps.
Mike