Read more about this topic on myanimelist, and join in the discussion on the largest online anime and manga database in the world Image search through danbooru (translated fan art), konachan, yand.re, gelbooru, and others listed on the site Join the online community, create your anime and manga list, read reviews, explore the forums, follow news, and so much more
Andrea (@andreahascannons) • Instagram photos and videos
Welcome to the unofficial comfyui subreddit
Please share your tips, tricks, and workflows for using this software to create your ai art
Please keep posted images sfw And above all, be nice A lot of people are just discovering this technology, and want to show off what they created Belittling their efforts will get you banned
Also, if this is new and exciting to you, feel free to post. Danbooru is a gallery for consumers, not for artists It's focused on the wants and needs of its audience, with the people who actually make the work just being convenient providers of content. Danbooru is the single most commonly used dataset for training ai on anime
Idk what you're going on about
I use this dataset and augment it with my own ai generated images (using real ones as input to img2img to increase sample size). Due to the minimal censorship in place on booru style image sharing websites, models trained on actual danbooru image sets usually have a strong nsfw bias However it is possible to use deepdanbooru captioning on other images sets, even sfw image sets. On danbooru, just put a space between the two tags you want (which is why single tags use _ instead of spaces)
Without a gold account, danbooru limits you to two tags, though you can add in a rating filter (e.g Rating:general) since this isn't technically a tag. Truehi everyone, i have trained and hosted a vision transformer on the danbooru dataset, as well as hosted a float16 optimized gpu version of blip2 on my website Danbooru tags generator image caption generator please feel free to upload your image2text images for prompt generation
I would appreciate any feedback on the vit model's performance (especially vs
The following are some resources that might help you find sources Great for finding an image source that might be fan art
