My attempt to make a vpred Noob-based model that doesn't suck!
Recipe: StableMondAI-SDG + Noob vpred 1.0 + some LoRA magic
I am still trying to find a good sampler/scheduler and CFG that doesn't create blurry results with darn near every vpred checkpoint, if you find that magic combo, let me know.
The best results I have found so far are Euler A/Simple and CFG 3.0
Also, gen at the standard SDXL resolutions for best results:
1024x1024
1152x896
896x1152
1216x832
832x1216
1344x768
768x1344
640x1536
1536x640
Description
Version 2 Recipe:
StableMondAI-SDG + Noobai-Cyberfix + Indigo Furry Mix XL VPR_IV_4
LoRAs:
Extracted LoRA of Oops! All Furries v2 - Base Noob EPS 1.0
Aesthetic LoRA trained on 600 EasyFluff 11.2 generated images unet only
Aesthetic LoRA trained on 200 Furry tumblr images circa 2012 unet only
Euler A/Simple and CFG 3.0 reccommended.
Also, gen at the standard SDXL resolutions for best results:
1024x1024
1152x896
896x1152
1216x832
832x1216
1344x768
768x1344
640x1536
1536x640
FAQ
Comments (7)
First one - my favorite thoughout all Noob and V-preds. Can't wait to test 2-nd version
this model works extremely well with all the content civitai just banned i tried it..
Wait, what?
@prodriver8301991 it's true it does the best job.
I hadn't heard of anything being banned, but I can probably guess
@prodriver8301991 read the new terms on the announcement page
prodriver8301991 like everything lol
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.