Hey guys and gals!
No EA this time, BUT 👇
I encourage you to post at least one image after downloading my model.
Just 150+ postings will compensate training costs at CivitAI. You can also collect this model or interact with my images. This is just a TIME for you but IS A GOLD to me.
BLUE BUZZ is the only thing required to make new trained model ;)
I don't ask you to donate YELLOW BUZZ, but will be very glad if you spare some. No matter what, all my BUZZ will be spent inside CivitAI on training new models.
TL;DR; Combination of creative visuals and sharpness of CinPony v2 with Booru / Natural language prompt adherence from Illustrious models. Feel free to post flawy examples and comparisons. I'm open for critics as usual.
[DISCLOSURE] I am using the Post-Maker (3-pass workflow) to produce HQ images using only the released model. You can download it and modify the Steps or resolutions. Steps 30 for 1st pass and 8-10 steps for 2nd and 3rd passes are enough to produce good results 90% of tries. 3rd pass is not so important.
V2 Highlights
RE-balnced colors toward true color natural palette
Sharper render - more small details
More natural expressions, poses and environment textures / geometry
Better lighting
Wider CFG range (CFG 7 == realism)
V1 Highlights
<TBD>
Description
HDR issue fixed.
Contrast increased.
Added tentacles and slime substance. Creativity increased.
Some cinematic LUTs applied.
Tested with CFG 3...9.
FAQ
Comments (8)
good for cyberpunk stuff, well done :)
You reminded me to add a trigger words "mechanical parts"
THX
BTW, thanks for BUZZ! Any donation helps as it goes on training the LoRa on CivitAI. This accelerates new developments a lot.
Did you reduce the realism?
Unfortunately yes. Cinematic LUTs and bunch of other additions (higher saturation, less sharpness, lower dynamic range) made a picture more fictional. It was quick release - few days. I need to recollect my thoughts and think more on how to hit into an optimal combination.
I wanted to add a cinematic sci-fi horror flavor. In some scenarios (some prompts) it shows realism and cinematicness, but in general this version might be less realistic than V2.
Girls are looking good, but undressing even when not asked.
Is it an Illustrious checkpoint or Pony checkpoint?
It is based on Illustrious with additions of Pony / SDXL resources.
I used the Illustrious model (probably v2 or some other NON-Original Illustrious), trained it with dataset containing the mixed captions (Natural + Pony style).