This LoRA is made to enhance the ability and quality to create anthro and furry characters with Ideogram 4. It also can generate other Fantasy Creatures , like werewolves
It improves consistency, fur detail, anatomy, and overall results.
This version is made for Ideogram4 other other model types the lora can be found here :
The Lora is trained my own database of the furry enhancer and Ratatoskr Qwen Image
its an fictional , fantasy database so it doesn't include real people, celebs etc.
this version is trained more on the realistic site of the database for comic i recommend using another lora version from above or my model ratatoskr qwen image.
for setting i will put an workflow example in made by
for settings use the strength between 0.6 and 0.8.
i use an modified version this workflow for loading loras
So why this version stands alone and is not in the Furry Enhancer lora collection?
it has to do with the Non-Commercial Model Agreement that Ideogram4 has put on the "open weight" model. that is also valid for loras trained on please read the License carefully!
In short:
Its an non commercial License so don't sell the Lora or any Output of it.
Please follow the original license terms. and keep it clean.
Description
What has changed?
second version retrained specific for realism
trained on much compact but realistic dataset from the Ratatoskr Qwen Image database
trained on 2048x2048 resolution
made for photos , high detailed digital art
for settings min 0.7 strength better 0.8 to 0.9
i recommend the lora when you wanna generate high resolution pictures at 2-2.5mp
you can combine both versions in one workflow
workflow is in the "training folder"
an comic version is also planned in the future .
have fun and please support my work with a like