Quoted By:
I'm starting to put a little more effort into making things look good, but I still feel like I don't know what I'm doing and have some nooby questions. I'm using SD 1.5 (only 8gb VRAM) and comfyui btw.
1. What checkpoint do you typically find yourself using?
2. If you generate something you mostly like but it has a part that's fucked up, do you:
• just keep trying different seeds until you get lucky and get something clean
• try tweaking the prompt/parameters using the same seed
• inpaint
• edit manually in an image editor, then inpaint
I find the last is often the most efficient, especially when the model repeatedly gets something wrong in the same way, but it feels like I'm doing something wrong if I have to resort to that.
3. I haven't experimented with different styles/artists yet really; when you want a different style, do you typically use different checkpoints, add style/artist tags to the prompt, use LORAs, or some combination?
4. I find my gens often lacking sharp details. Besides upscaling, what are some ways to improve this? Maybe this is where using a sharper style would help me.
Perhaps the answer to all these questions is "it depends, experiment and see what works". I just worry that I'll end up settling on a workflow that's really flawed/inefficient without knowing there's a better way.
