Discuss matters related to our favourite AI Art generation technology
Use ControlNet if you want to put something specific into the image.
Also, this Lora is very helpful, just adjust its weight.
Prompt:
(letter B) (favicon:1.3), circle, high contrast BREAK (sticker), 2d art, no humans, (vector art, illustration), vivid colors, simple, (black background) BREAK 4k, 8k, highres BREAK (masterpiece, best quality) <lora:anime_minimalist_v1:0.3>
Thank you for the suggestion, and the prompt I will try
I’ve experimented with this a little. In my experience, you just keep generating off of your prompts until you get lucky and it outputs something you will get good results with at smaller sizes. I can’t really say I have a workable prompt, but I try to use things like icon and clip art and highly stylized to try to get something that has thicker lines and fewer precise details.
Definitely post-process in an application like photoshop. You might be able to get away with as little as a crop and shrink, but I’d bet you’d have to erase some extraneous details to get a nice result in the end.
Adobe Illustrator would make a lot of sense in this case as part of the post-process as well. It's used to create vector graphics which can be scaled up or down easily. Steal it if you do of course because fuck adobe
I think keep regenerating is the best way. Thanks
If you're looking for a solid background I think I saw a tip where someone would load an image with a solid color and run img2img with a very high denoise strength.