a workflow with flux redux, fill and the new ace++ lora in a single workflow
Description
FAQ
Comments (11)
Good to play with, thanks!
u put wrong link for -flux fill
otherwise worked really well.
Q: is there a way to target a specific face if there are more than one face in an image?
@tazztone thanks for pointing that out, i'll correct it
currently there is none, as it automatically detects faces, but i'll make a V2 where you can paint a mask on a specific zone if you want to, effectively selecting one face
@Archi_teknik101 there may be other ways like targetting the LARGEST face or the "3rd from left" etc. at least that's how some extensions handled it in forge webui
It runs okay until: FaceBoundingBox
'NoneType' object is not iterable
any ideas??
i figured it out. it works! :-)
yes sometimes the dlib doesn't detect the face, in that case you can switch the model in the face detection node for the insightface, this should fix it, however the mask doesn't include the forehead which is a bit less precise, still good to play with.
Can never get any face swap to work. They have all these outdated nodes or files you can't find anymore.
nodes are not outdated, you need to have a correct installation of dlib with the facelandmarks models or insightface, both are notoriously terrible to install and make it work on comfy, i'd suggest you to do a separated install of comfy. I can confirm it still work as of yesterday 09/14/25
@Archi_teknik101 Thank you!