Content
@
https://warpcast.com/~/channel/ai-art
0 reply
0 recast
0 reaction
Claus Wilke
@clauswilke
/ai-art WIP thread. Please reply with your works in progress. Only things that haven't been minted. If you reply with more than one work, please thread your replies. I will tip 200 DEGEN to each reply, once my allocation refills tomorrow and until I run out. Should have enough for at least 25 replies.
45 replies
11 recasts
54 reactions
Larry Scruff
@larryscruff
I've been trying to make an image of a man fighting a bear with a sword, which shouldn't be difficult for the AI, but for some reason it keeps making a bear in armour with a sword 😂 I've tried changing checkpoints, using controlnet, different LORA's etc. No luck! SDXL installed locally on my PC. I'm baffled!
2 replies
0 recast
0 reaction
daivd 🎩👽 ↑
@qt
Very baffling, sorry I don't have a tip but I've experienced similar issues and it can be so maddening. Cool look tho! 200 $degen
1 reply
0 recast
2 reactions
Claus Wilke
@clauswilke
Yeah, diffusion models do this. They have difficulty keeping concepts separate. That's why you need control nets and other techniques to ensure the right parts of the image get the right prompts. 200 $DEGEN
1 reply
0 recast
1 reaction
