Comments by "Kasumi Rina" (@KasumiRINA) on "Man Carrying Thing"
channel.
-
1
-
1
-
1
-
1
-
1
-
To generate AI art, people render dozens, sometimes hundreds of pictures, then select a few for upscaling, then keep fixing those with inpainting... AI art right now has so many issues I give up and use Photoshop to finish my renders, and those are meme-grade things. The biggest issues in Stable Diffusion isn't even hands, and those can be fixed manually (hehe), it's inability to tell apart relation between items and draw separate distinct people. For example, you write "red shirt, long hair" and end up with red hair, and not ginger but actually RGB red.
AI is VERY helpful for a lot of things (healing brush and auto-fill in Photoshop are AI too), but honestly, fine-tuning, for example, a voice synth, takes more effort than just speaking into microphone. Generated art IS faster than drawing, even if its bad, but it is MUCH better if you get involved, i.e. make a sketch to colorize, filter an existing image, or use control net, blender render etc. to begin. Yes, some people actually model things in a 3D program to screenshot them and put into AI art generator to make it look like drawing or photo.
Companies saying it was made by computer not a person are manipulative, they say that to rid people of copyright. AI isn't thinking or acting on its own, saying person didn't make the pic when they inputted data and tuned the details is as ridiculous as saying you didn't make a picture when you just clicked the shutter. The anti-AI sentiment is really like anti-photo one from two centuries ago. "It's not real art, the camera made it", come on. And again, if I kept drawing more I would totally use AI to help with coloring, details, and stuff like backgrounds, instead of trying to campaign against a boogeyman like a horse salesman when cars appeared.
1
-
1
-
1
-
1
-
1
-
1
-
1
-
1