anayachandra
Anaya Chandra
big difference actually.. more realistic... nice
[C Black TG Comics] Comics and Artwork
- Thread starter Sam Emberlyn
- Start date
- 3dcg
- feminization gender-change transformation
- trans-protagonist
Typee
New member
sigh is right...
mihaluck
Active member
yes I tried later it have different mode and presets to hand it. also good for cblack characters that proportions and style more realistic. Didnt like it on more styled characters by other authors. and it has problem people look older that I dont like.
mihaluck
Active member
have that problem, but will be better, still it much better lightning and realistic skin, hairs and clothes. for me originals looks too yellow/red or too tan, not natural.
santo
Member
mihaluck
Active member
Jackjons
Well-known member
Why it is looks like bloodlose from the skin . Lol
anayachandra
Anaya Chandra
wow.. interesting... which app is helping in doing this....
Abawell
Active member
It's great.
Now, double the resolution an do a realistic pass on it, like santo did. Problem, must be extremely long.
mihaluck
Active member
no its not. I already tried dlss5 upscale but it didnt works as in games, just resized images so need to fix it if upscale will work as in games would be amazing.
mihaluck
Active member
just installed dlss5 nodes in comfy ui.
@lexx$
Active member
Wow, it would be interesting to do a full remake of Superego or even include all its previous stories using DLSS 5. But I suppose that would be a bit complicated.
Nuvolari
Active member
I don't get all this feeding other people's hard work into some automated LLM thing trying to make it look more "realistic" while homogenising any soul out of it. "Realistic" is not necessarily (or even often) better. Feeding other people's hard work through Grok, or ChatGPT or whatever does not turn you into an artist.
Would feeding, say, Dave Gibbons' Watchmen through Grok to make it "realistic" improve it in any way? What about The Dark Knight Returns? Alex Raymond's Flash Gordon? Hugo Pratt's Corto Maltese? One Piece? Ranma 1/2? Tintin? AstƩrix? Calvin and Hobbes? Peanuts? Carl Barks' Donald Duck?
This is like turning perfectly good cheese into that processed yellow goop fast-food places put into hamburgers. It may look good, but there's no soul, no personality, no originality -- and no taste.
If you really like fiddling with image-generating LLMs, try using them to create something new, not as a defective Xerox machine. There are literally millions of stories all over the Internet which have no illustrations. I have seen people using LLMs for good effect as tools, generating interesting original illustrations. But then, that takes more than acquiring a bunch of LLM tokens and mindlessly feeding other people's work into an engine. It takes real effort.
Would feeding, say, Dave Gibbons' Watchmen through Grok to make it "realistic" improve it in any way? What about The Dark Knight Returns? Alex Raymond's Flash Gordon? Hugo Pratt's Corto Maltese? One Piece? Ranma 1/2? Tintin? AstƩrix? Calvin and Hobbes? Peanuts? Carl Barks' Donald Duck?
This is like turning perfectly good cheese into that processed yellow goop fast-food places put into hamburgers. It may look good, but there's no soul, no personality, no originality -- and no taste.
If you really like fiddling with image-generating LLMs, try using them to create something new, not as a defective Xerox machine. There are literally millions of stories all over the Internet which have no illustrations. I have seen people using LLMs for good effect as tools, generating interesting original illustrations. But then, that takes more than acquiring a bunch of LLM tokens and mindlessly feeding other people's work into an engine. It takes real effort.
mihaluck
Active member
no it quite easy now. would be better if upscaler add some details but I dont see difference now so will use without it.