Stable diffusion sucks at logos.

I haven't looked into Vector Generation models in a while but that's what you'd need.

Reply to this note

Please Login to reply.

Discussion

Yeah actually just used Co-pilot (have access through work) which was dall-e under the hood I think. Images were OK but not simplistic enough for a logo and after one or two prompts it is hard to turn the image without the gen-ai over complicating or just screwing up the image.

*tune