<p>My first attempt at a Krea2 merge, based on <a target="_blank" rel="ugc" href="https://civarchive.com/models/2997793/kroma-krea-2">Kroma 0.3.1 turbo</a> with additional furry concept training, and LoRAs extracted from <a target="_blank" rel="ugc" href="https://civarchive.com/models/2918022/willyskrea2furrymix">Willy's Furry Mix</a> and <a target="_blank" rel="ugc" href="https://civarchive.com/models/2881657/wulver-krea-2">Wulver</a>. Good style flexibility, and excellent LoRA compatibility. LoRAs should be trained on Krea2 base.</p>

Furry e621 NSFW Kroma 0.3.1 Merge

<p>Version 2 recipe: <a target="_blank" rel="ugc" href="https://civitai.red/models/2997793/kroma-krea-2">Kroma 0.3.1 turbo</a> with additional 3000-image furry concept and high aesthetic LoRA trained at 1536 pixel resolution. 20% merge of LoRAs extracted from <a target="_blank" rel="ugc" href="https://civitai.red/models/2918022/willyskrea2furrymix">Willy's Furry Mix</a> and <a target="_blank" rel="ugc" href="https://civitai.red/models/2881657/wulver-krea-2">Wulver</a>. 30% merge of <a target="_blank" rel="ugc" href="https://civitai.red/models/2980065/krea-2-furryplus-general-nsfw-kroma-02">Krea2 FurryPlus</a> LoRA. 50% merge of <a target="_blank" rel="ugc" href="https://civitai.red/models/2993700/krea2-textfusion-refusal-reduction-loralowrank">Krea2 TextFusion Reduction</a> LoRA. Block merging of base Krea2 blocks 1, 3, 7, 9, 11, 13, 15, 17, 19 back into the final model at 40% weight to bring back stability to background elements. (Kroma tends to make crunchy, noisy solid colors and gradients, this is a band-aid for that)</p>

Files with SHA256: 2e37391788b5a53ba325c71fc5903ab767e93d4fe21676bb449be42fbedb04f2