ALL THESE ARE FOR SDXL09
Lora-SDXL-FaceDetailer-2X:
you will need this to be installed for Face Detailer to work:
https://github.com/ltdrdata/ComfyUI-Impact-Pack
Lora-Remacri2X:
This is a simple preset for using the SDXL base with the SDXL refiner model and correct SDXL text encoders. This is pretty new so there might be better ways to do this, however this works well and we can stack Lora and Lycoris easily, then generate our text prompt at 1024x1024 and allow remacri to double the image size at the end.
You can change the size dimensions in the places near the prompt.
it will create 128x128 to your target 1280x768 for example, then double to 3560x1536.
I put this up to share my ComfyUI Presets
For SDXL you will need the SDXL model from hugging face ;)
consider pruning the model if the size gives you trouble
feel free to modify or improve any of these scripts! thanks!
Description
FAQ
Comments (9)
ComfyUI Q: any idea how I can select a lora/ckpt from the dropdown list by typing its name? It has become too cumbersome to work with ComfyUI with many lora/ckpts
You have to input the folder references in the extra_model_paths.yaml - ComfyUI folder.
@Piscabo I meant I have the list of checkpoints/safetensors;
If the list is very long I need to scroll till I find the one I wish to select and then click on it.
I wish to be able to start typing and filter out the list according to the string of what I am looking for, and then select it easily without having to scroll dozens of models.
in Load Checkpoint node there is the "Filter list" header but it does nothing.
same thing for the Load Lora node. Perhaps there's a different node system? or a way to config the UI?
@Natotela Can't show a screen grab, but search filtering works fine... maybe reinstall the WAS nodes.
Come chat here. https://discord.com/channels/752483298198028309/1084878026644393984
@Piscabo can't get to your link, discord's on "NO TEXT CHANNELS"
also: It seems that if I am very much zoomed-in, I automatically get the cursor in the Filter List header, but if I am zoomed out to see many nodes in the workflow, and then I press a key, the press is handled in the scope of the main comfyUI frame, not the dropdown menu I just clicked on
@Natotela you can use my example workflow, or with any workflow, simply add the lora loader after the model loader and connect the CLIP and MODEL to the next node as before.
@driftjohnson no prob with the lora node , just with the UI/UX of comfy, I need to zoom in real hard in order to be prompted for the name of the file/lora, rather than having to select it with mouse cursor. I would like to type lets say LOW.. and it will automatically filter anything that doesnt start with LOW
Thanks, I am enjoying this layout.
I generated 1 image and now suddenly I get this error. Where should I look for a resolution...
Error occurred when executing FaceDetailer: pic should be 2/3 dimensional. Got 4 dimensions
ah, because i did not use any custom nodes, you must change the resolution in both boxes near the prompt :) sorry for not explaining that in detail