This LoRA was trained to reproduce a highly recognizable Western comic style. If you notice a familiar flair of a certain famous artist here—you're absolutely right! ;) This is exactly what I was looking for. <h3 id="what-you-need-to-know:">⚠️ What you need to know:</h3><ul><li>Focus: It is MOSTLY intended for NSFW generation. While it can produce beautiful SFW images, it naturally leans towards mature themes and works incredibly well with nudity. </li><li>Limitations: This LoRA shines its brightest on solo character images. Keep in mind that adding two or more people makes the anatomy unpredictable. Usually, generating a solo girl in a cowgirl pose is completely fine, but complex multi-character interactions are a known issue.</li></ul><h3></h3>This model is based on Pony Diffusion V6 XL. To get the best of both worlds, this version is a merge of my V1 model (which had a very wild, creative imagination) and V2 (which was much more straight and consistent to the prompt). <h3 id="recommended-settings:">Recommended Settings:</h3><ul><li>Trigger word: <code>inco_style</code></li><li>Base Model: Pony Diffusion V6 XL</li><li>CFG Scale: 5.0 - 7.0</li><li>Sampler: Euler or DPM++ 2M Karras</li><li>Resolution: 832x1216 (or other standard SDXL resolutions)</li></ul><h3 id="a-note-on-facedetailer">A Note on FaceDetailer</h3>The first few examples in the showcase were made using the FaceDetailer node. I highly recommend using it to keep the comic outlines sharp. Detailer Settings:<ul><li><code>denoise</code>: 0.32 - 0.35</li><li><code>bbox_crop_factor</code>: 2.0 - 3.0</li></ul>(Note: The last two images in the gallery showcase the raw output without any detailers, so you can check the initial, unfiltered result of the LoRA). This is my very first LoRA! I'm planning to improve and experiment with this style further on. Hopefully, the multi-character issue is something I can fix in a future V3 update. Full ComfyUI Guide &amp; Workflow:I just published an in-depth article breaking down my entire generation pipeline (including dual detailer setups, upscaler tricks, and tips for ComfyUI Desktop):👉 <a target="_blank" rel="ugc" href="https://civarchive.com/articles/35359/how-to-get-clean-comic-lineart-in-sdxl-my-complete-comfyui-workflow-sfw-and-nsfw">Read the Workflow Guide Here</a> I would be absolutely pleased if you leave a comment, a review, or share your generations. Enjoy!

Inco Style XL

Merged version (V1 Ep6 @ 80% + V2 Ep6 @ 20%) This is a merge of two different training runs (both at epoch 6). The V1 base (80% weight) serves as the core "imagination and style engine," providing the dynamic comic flair. The V2 component (20% weight) acts as a stabilizer, keeping the style consistent and reigning in the wilder artifacts of the first version.

Files with SHA256: 3bf283548d96ccb3240dd13277196e8b26acf919c0297fbe2a30b6396ed4132e