<p>Flux 2 was not what we wanted or needed, but Klein appears to be what we deserved! Let's unlock the full NSFW potential of this model!</p><p>Miraclein focuses on photorealism and more grounded/grungy images. Not amateur style, but not fantastical either.</p><h2 id="generation-settings"><span style="color:rgb(230, 73, 128)">Generation Settings</span></h2><p>Most settings work, but I use:</p><ul><li><p><strong>12 steps</strong></p></li><li><p><strong>Euler</strong></p></li><li><p><strong>CFG 1.1</strong></p></li></ul><h3 id="fp8-download">FP8 Download</h3><p>For some reason, when you upload multiple formats (BF16 + FP8) on one model, which they now support, they hide the FP8 under the BF16 model group in the side menu...</p><p><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/9cc7e493-abe7-4852-b947-8c55793f4ca6/original=true/9cc7e493-abe7-4852-b947-8c55793f4ca6.jpeg" /></p><h2 id="editing">Editing</h2><p>It works fantastically! Try it out as an edit model!</p><h3 id="ggufs">GGUFs</h3><p>Do it yourself:</p><p><a target="_blank" rel="ugc" href="https://github.com/qskousen/ggufy">https://github.com/qskousen/ggufy</a></p><h3 id="forge-neo">Forge Neo</h3><p>Should work fine using qwen3_8b <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/vae-text-encorder-for-flux-klein-9b/blob/main/split_files/text_encoders/qwen_3_8b_fp8mixed.safetensors">qwen_3_8b_fp8mixed.safetensors </a>and for VAE <a target="_blank" rel="ugc" href="https://huggingface.co/black-forest-labs/FLUX.2-small-decoder">FLUX.2-small-decoder</a>. Make sure to be on the correct preset.</p>

Miraclein NSFW [Generation & Edit] [Flux2Klein]

miracleinNSFWGeneration_30Bf16Fp8.safetensors

3.0(bf16+fp8)

2.0_FP8

2.0_bf16

1.0_bf16

1.0_nvfp4

1.0_fp8

<p>Flux 2 was not what we wanted or needed, but Klein appears to be what we deserved! Let's unlock the full NSFW potential of this model!</p><p>Kleinova focuses on professional photography and beautiful shots. It's vibrant and colorful and over the top.</p><h2 id="generation-settings"><span style="color:rgb(130, 201, 30)">Generation Settings</span></h2><p>Most settings work, but I use:</p><ul><li><p><strong>12 steps</strong></p></li><li><p><strong>Euler</strong></p></li><li><p><strong>CFG 1.1</strong></p></li></ul><h2 id="editing">Editing</h2><p>It works fantastically! Try it out as an edit model!</p><h3 id="fp8-download">FP8 Download</h3><p>For some reason, when you upload multiple formats (BF16 + FP8) on one model, which they now support, they hide the FP8 under the BF16 model group in the side menu...</p><p><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/9cc7e493-abe7-4852-b947-8c55793f4ca6/original=true/9cc7e493-abe7-4852-b947-8c55793f4ca6.jpeg" /></p><h3 id="ggufs">GGUFs</h3><p>Do it yourself:</p><p><a target="_blank" rel="ugc" href="https://github.com/qskousen/ggufy">https://github.com/qskousen/ggufy</a></p><h3 id="forge-neo">Forge Neo</h3><p>Should work fine using qwen3_8b <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/vae-text-encorder-for-flux-klein-9b/blob/main/split_files/text_encoders/qwen_3_8b_fp8mixed.safetensors">qwen_3_8b_fp8mixed.safetensors </a>and for VAE <a target="_blank" rel="ugc" href="https://huggingface.co/black-forest-labs/FLUX.2-small-decoder">FLUX.2-small-decoder</a>. Make sure to be on the correct preset.</p>

Files with SHA256: 0dd0a286430d270411f937bfb774b29642b3c82f82c729918ee5806ada3485b2