For those more interested in the furry and scalie side of art, it’s time to celebrate!!
NovelAI Diffusion Furry V3 is a new diffusion model built on our improved SDXL architecture. The models training process used the same data curation and classification that yields SOTA performance for its task, and it was trained specifically to support our Vibe Transfer feature out of the box as well as trained on updated furry data and furry tags.
You will find that there has been an overall uplift in quality, much like our NovelAI Diffusion Anime V3 model. We’re excited to let you see the upgrade for yourself!
Besides the overall improvement in the quality of the images that the new model is capable of creating, you will also find that higher resolution images work much better now than they did before!
Here are a few general prompting tips to get you started, but please remember that these are not the only tags the model knows, so don’t limit yourself to them!
Character count tags: solo, duo, group (plus solo focus or duo focus)
Physical form tags: anthro, feral, humanoid, animal humanoid, human, taur
Background tags: detailed background, photo background, simple background, amazing background
Quality tags : worst quality, bad quality, normal quality, amazing quality, best quality
Aesthetics tags: very aesthetic, aesthetic, displeasing, very displeasing
Other notable tags: detailed
The images below illustrate how much the quality of the images created by the NovelAI Diffusion Furry V3 model has improved.
For some of the images, we will also show a comparison between NovelAI Diffusion v1.3 and NovelAI Diffusion v3 using the same prompt to illustrate the major difference in quality and composition!
The Furry V3 model can also be used to create traditional humanoids, as seen in the kitsune-themed beach girl example pictured above!
So many fluffy tails~!
But that’s not all there is to our Furry Diffusion V3 update…
NovelAI Diffusion Furry V3 is capable of creating the prettiest ponies!
Here are a few more comparisons between the old AI image generations from NAI Furry Diffusion V1.3 to todays V3 release!
We’re excited to see what amazing new things you will create with our image generation upgrade!
Go on and paw into the furry side of AI with NovelAI’s long-awaited Furry V3 image model now!
