cartoony but detailed style focused on hyper.
The dataset used was made up of 600~ images all tagged. It should be quite versatile for poses. It can also do comics and multiple angles as well as internal views, if you don't want those be sure to include those in the negative prompt.
The tagging and large amount of images means that there are a lot of characters that can be recognised, although some may need accompanying loras for greater detail.
Use with a weighting of 0.8-1.0 and a cfg of 5.5-8 for Illustrious.
Use a cfg of 4-5.5 for NoobAI.
Description
For use with the vpred version of noobai and it's submodels.
If having issues with proportions or prompt accuracy try using the cyberfix checkpoint or lower the weighting.
Use a cfg of 4-5.5
FAQ
Comments (5)
Is Noob better than Illustrious is that what everyone is switching to now?
I'd say its on par or slightly better. It recognises a lot more furry characters and has some built in artist styles. It also has an understanding on angles and lighting which makes it nice to use
I will have to give it a go! I was surprised at how amazing and flexible Illustrious was so its intriguing that it could be better!
im wondering, can gabe newell do anything about this lmao
Imagine if people stopped posting images made with pony models so we could actually see how the lora performs.. good job nontheless.
Details
Files
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.