search

Ling-APE / ComfyUI-All-in-One-FluxDev-Workflow

An All-in-One FluxDev workflow in ComfyUI that combines various techniques for generating images with the FluxDev model, including img-to-img and text-to-img. This workflow can use LoRAs, ControlNets, enabling negative prompting with Ksampler, dynamic thresholding, inpainting, and more.
132 stars 3 forks source link

dual clip and vae node question #5

Open joshgura opened 4 weeks ago

joshgura commented 4 weeks ago

hi, thanks for this workflow. haven't gotten it to work yet because i'm just starting out with flux. I will continue to study the info page.

was wondering about the dual clip and vae nodes in your workflow, they just look like little link nodes, don't see the loaders for them. are they remotely loaded somewhere else in the workflow?

as a side note, i'm attempting to use the GGUF custom loader instead of the checkpoint loader provided.

edit:

put ae.safetensors in vae folder. but i'm getting 'vae has no dtype' errors.

OperationTwelve commented 2 weeks ago

The Vae error might be because this workflow has ae.sft as default as that is what it used to be named... check and make sure you select all the models manually. Also as for using GGUF you can make sure you get this custom node image

and then you can use the UNET Loader (GGUF) node instead like this image

SwiftEdge7 commented 1 week ago

I'm a ComfyUI beginner and this solved my first problem. Didn't realise some of the nodes were closed and could be expanded by clicking the circle.

The workflow says you can switch between txt-to-img and img-to-img, how do you go about making this choice? I can't see the starting point or which nodes to bypass. Thanks