>>54398170>>54396648https://files.catbox.moe/a2lob9.pnghttps://files.catbox.moe/njeah7.pnghttps://files.catbox.moe/6i93v7.pnghttps://files.catbox.moe/ofumf7.pnghttps://files.catbox.moe/h90onj.pnghttps://files.catbox.moe/k4p978.pnghttps://files.catbox.moe/muhw5u.pnghttps://files.catbox.moe/5w2tsr.png>I'm really hoping for them to pull this off, though correct me if I'm wrong but can't they use individual fanarts of them as source?Yeah but I'm assuming people will be drawing them together a lot more then the other new girls.
Now I don't know anything about training LoRAs but the best images to train the AI are probably just the girl standing there looking cute or cool. But if artist are going to draw 2 girls they might be drawing them hugging, or crawling all over each other because... you know it's cuter. So a LoRA maker will probably need to photoshop one of them out and do it for a bunch images to make a training set to tell the AI "This girl is the pink doggo, she looks like this" and same for the the blue one.
>>54400027Cool animated stuff as always anon, I like that weird gun on the bottom right it's doing 'something'
What is your workflow? Do you make low res stuff in animateDiff then upscale in img2img? Are you also trimming down prompt tags to fit the prompt limit?