Search Results
7/5/2025, 4:48:06 AM
>>96016554
Thanks for complimenting my MS Paint work. While it does have its own..."charm", if I were trying to produce a module for mass consumption, I'd want something that looks a little nicer.
If I really buckled down, I could do some Photoshop tricks, copy & paste stuff, and add in my own details as needed, but that's a whole lot of work for "Windows that show outer space" that will be used for one sequence of a session that will never be seen again. Instead of burning a lot of time on that, I could be working on balance, lore, and other writefag stuff for the module.
>>96016550
Some additional prompting is required to fine-tune stuff, but you usually don't need more than 2 or 3 tries at most if you want something simple on GPT. For everything else, local gen can cover a lot of bases. Here's a robot-faced shopkeeper for my si-fi setting I'd gen'd locally.
I've heard local is getting better image-to-image tech as of late, but I haven't had a chance to try it myself. I'm still using A1111. I need to upgrade to comfy UI or one of the newer APIs. I'll have to do some research into that because I'd love to do image-to-image locally.
Thanks for complimenting my MS Paint work. While it does have its own..."charm", if I were trying to produce a module for mass consumption, I'd want something that looks a little nicer.
If I really buckled down, I could do some Photoshop tricks, copy & paste stuff, and add in my own details as needed, but that's a whole lot of work for "Windows that show outer space" that will be used for one sequence of a session that will never be seen again. Instead of burning a lot of time on that, I could be working on balance, lore, and other writefag stuff for the module.
>>96016550
Some additional prompting is required to fine-tune stuff, but you usually don't need more than 2 or 3 tries at most if you want something simple on GPT. For everything else, local gen can cover a lot of bases. Here's a robot-faced shopkeeper for my si-fi setting I'd gen'd locally.
I've heard local is getting better image-to-image tech as of late, but I haven't had a chance to try it myself. I'm still using A1111. I need to upgrade to comfy UI or one of the newer APIs. I'll have to do some research into that because I'd love to do image-to-image locally.
Page 1