Nendo is somewhere between a Merge and a Trained model.
I do not have the resources to train a whole model so I have trained severals LoRAs that were baked into my other popular model Brainime. Some weights from ToonYou Beta2 by the amazing Bradcatt are also present in this model so make sure you follow their license as well.
Most LoRAs were trained with anywhere between 20 and 200 pictures. All trained to 20 epochs at about ~1000 steps per epoch.
I'm very proud of this work and the feedback from the PixelWorld and CivitAI Discord has been monumental in some artistic decisions that were taken.
I'm looking forward to see what you will all make of it.
Description
You read that right, this is v1.0 and it's HOT.
What is new?
Some weights from the beautiful fcRadiance Apex and Resonance
Updated ToonYou Beta2 weights to Beta3 (Credit to the amazing Bradcatt)
Model is definitely closer to what I envisioned which is a model that provides pictures that are obviously not real but still feel somewhat touchable, like clay.
I hope you enjoy it! Your feedback is very valuable!
FAQ
Comments (2)
Your model looks really interesting and capable -- thanks for making it available. Not sure if the results from the versions differ substantially, but I wanted to offer a comment that your 0.3 beta images are much more interesting and dynamic than the 1.0 previews. Perhaps you could show some more variety with those. 👍
Point taken and feedback received @netdzynr !
I'm really grateful you shared your thoughts and I have updated the post, I hope it is a bit more interesting !