kinda scary ngl
Description
Trained on Anima Base v1.0
FAQ
Comments (7)
Oh damn, you beat me to it. Been working on this on and off for a year now. Tried to combine style + the nose hook stuff and so on
Yeah but mine is kind of plain, i only used the trigger word in my tag editing, so i didn't do anything specific
@Sfdwackys Maybe so (even though it looks good), but seeing those preview images gives me a bit of motivation to work on it again. It's a real damn pain to tag them all right. The dataset is gigantic and I got so tired of making sure everything is tagged right since I want to be able to trigger every unique detail from Lestart
@Lopodelas Do you not have a workflow for tagging the images? I could show you how i tag specific images with a certain concepts out of a dataset if you'd like.
@Sfdwackys Tbh not really. I usually let an autotagger do it and then i prune/add tags as I see fit. I don't have a reliable automation for this and I don't trust it to be good when I don't edit it myself. What's your workflow?
@Lopodelas Well ofc an autotagger is non negotiable, but for the tag editing. There is a dataset tag editor for the original A1111 fork that you can select to replace tags, batch add tags to selected images, remove unwanted tags, etc.
https://github.com/toshiaki1729/stable-diffusion-webui-dataset-tag-editor
@Sfdwackys I mainly use this GitHub - starik222/BooruDatasetTagManager · GitHub
and sometimes this GitHub - jhc13/taggui: Tag manager and captioner for image datasets · GitHub