Update V3B
+
-
Usage recommendation
Sampling Method + Hiresfix
Euler Ancestral + Latent (nearest-exact)
image looks sharper
DPM++ SDE Karras + 4xUltraSharp
good lighting
DPM++ 2M Karras + 4xUltraSharp
Steps : 25+
CFG scale : 7 - 10
Resolution : 512x768 (Portrait), 768x512 (Landscape)
Quality Prompt : masterpiece, best quality
Negative Prompt : (worst quality, low quality:1.4), interlocked fingers, bad anatomy, disfigured
Is highly recommended to use Hiresfix or you will get bad results
Please review with images, it will be very useful for improving the next version, thank you
TensorArt : https://tensor.art/models/700274386156024519
Description
Update V3
+ Add more base model to merge
+ Improve face expressions more youth
+ Improve color sharpness and reduces blur
- Sometimes generate unwanted lines
- Unstable NSFW
Change VAE to WD
FAQ
Comments (8)
what's new in V3?
@sci not much, just important changes like improved color sharpness
Possibly dumb question, sorry lol. Is WD-2 the same as the WD-vae listed on hugging face and on civit? Maybe i'm tripping but I swear that those state they are WD v1 is why I ask?
@SirHempy My apologies if I confused you. this was my mistake for not doing research
As far as I remember, I downloaded a batch folder full of VAE and on my pc it said WD-V2. I rarely baked WD on my models, so I don't really understand some versions of WD and Thank you very much for telling me
Oh okay that makes sense. I was actually just asking cause if there was a v2 I wanted to add it my arsenal is all lol. Thanks for responding, and double thanks for the awesome model matey.
good series. thank you very much... it would be nice if you could reupload your old models though
Seriously, one of the greats! Amazing model and now my go-to for Anime.
I'm glad you like it, thank you ❤
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.