Trained on the art of Jim Unger
Description
FAQ
Comments (7)
đź––
Can you share the details about the training?
I extracted the panels from the books, picked out 1000 representative images, ran them through a captioner and then ran them through the Civitai Lora training.
captions were in the following format
"a black and white newspaper comic illustration of Herman, <image caption> by Jim Unger"
Tks， I wonder how you process the images with dialogue which influence the training. Is it that you remove them with some related model?
@MCcocktail None of the Herman comics have dialog, they have some text below but that gets cut off when I extract the panels. I used this to cut up the panels https://github.com/njean42/kumiko
@MCcocktail This can also be used https://github.com/thieupepijn/ComicPanelSplitter
@TallAndGreen thanks very much
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.