This is my first style lora for Flux Klein 9b so go easy with me on this one. I'm a bit late to the Klein party, but I was a bit disappointed to see a lack of furry loras and was curious if its possible. So I plugged my bara dataset into Civitai's trainer and this is the result. I'd say it blows Z-Image out the water, however, it can be hit or miss. Some prompts do amazing and some just never look right. I'm still new to Klein though, there might be a way I'm using it wrong or something. Btw the workflows are attached to the images, but its just the built-in Comfyui flow with loras attached. It can also help with edits if you want to make characters furry. Its not perfect but better at base Klein for that specifically. One step closer to locally generated furry memes. Also, it can't do peens. To achieve that I used another checkpoint found here <a target="_blank" rel="ugc" href="https://civarchive.com/models/2350642/klein-foreskin-full-checkpoint-undistilled-use-with-turbo-lora">https://civarchive.com/models/2350642/klein-foreskin-full-checkpoint-undistilled-use-with-turbo-lora</a>. I also used a turbo lora found here <a target="_blank" rel="ugc" href="https://civarchive.com/models/2324315/klein-4b9b-base-to-turbo-lora">https://civarchive.com/models/2324315/klein-4b9b-base-to-turbo-lora</a>. Not sure if its needed though. To save yourself a headache, I suggest passing a reference image to <a target="_blank" rel="ugc" href="https://huggingface.co/spaces/fancyfeast/joy-caption-beta-one">https://huggingface.co/spaces/fancyfeast/joy-caption-beta-one</a> and generating a "very long" caption, as that captioning model was what I used to train the data. Prefix it with "boerbara" as the trigger word. I find the following works quite well:"boerbara. This is a highly detailed, realistic digital painting of a muscular anthropomorphic...", followed by the joy-caption from the character species onwards. I find "soft fur" to also help.

Files with SHA256: 9f2289dd60df1a4dab6680d58808f162a0b11a9bad124354ebdd9d3f2d0c70c2