- This topic has 16 replies, 5 voices, and was last updated 1 year, 3 months ago by
.
-
Posts
-
@legolas18 @frank22 @fuzzy3000 @gaver and any others comfortable or curious with SD...
Please SD-ify (img2img) my recent "ZGFX and CHILLS" render:
No "rules", except perhaps stay as close to the subject render as is reasonable - including preservation of the ZoneGFX sign and announcement if possible. (Know that "watermark" and "signature" are often included in negative prompts, for example, and can erase or obliterate such things as signage or poster text.)
(Also know that text is a tough ask for AI engines, often coming out as gibberish. So if it's lost no big.)
Should be interesting to see comparative results.(I'd offer points, but it seems you people have too much fun with SD already. LOL. So maybe a contest later or something.)
P.S.: Please don't be shy. Post your result no matter your opinion of it. Everyone is learning here. I'm half asking/offering this idea as a purposeful comparison and contrast experiment to continue what I think is a really interesting input/output discussion about AI engines in general. There are no "bad entries", here. All contributes to the discussion in my opinion.
(Don't mean to hijack this thread in any way. If this should be a separate post please suggest as such and will relocate. Just seemed in-line with subject for here.)@spats The new gen of models can do some simple text, but it's still mostly 1-2 short words at most, and not perfect, so I don't think the sign in the image will be preserved. I'll give it a shot at this later because, as you said, it's fun 🙂
- You must be logged in to reply to this topic.
