This style is based on my own SFW sketches, digitized and prepared for LoRA training.Showcase images were created with <a target="_blank" rel="ugc" href="https://github.com/Acly/krita-ai-diffusion">krita-ai-diffusion</a> and are provided in PNG format.<h3 id="anima">Anima</h3>Base model used for the showcase images: <a target="_blank" rel="ugc" href="https://civarchive.com/models/2544636/wai-anima">WAI-ANIMA</a><pre><code>Trigger word: @br4cketgl7ph

For better results, try: @br4cketgl7ph, lineart, sketch, black and red, ink</code></pre><h3 id="krea-2">Krea 2</h3>Base model used for the showcase images: <a target="_blank" rel="ugc" href="https://civarchive.com/models/2726029/krea-2-turbo-official-comfy-org-checkpoints-krea2?modelVersionId=3147780">krea2_raw_int8_convrot</a>LoRA's file used for the showcase images: krea2_br4cketgl7ph_v01-1100, with strength 0.8<pre><code>Trigger word: br4cketgl7ph

For better results, try: br4cketgl7ph, lineart, sketch, black and red, ink</code></pre>At its core, this is not NSFW LoRA. For anything the base Krea 2 can't do, you'll need some additional LoRA (at your own discretion).<hr /><h3 id="notes">Notes</h3>From my observations, the shorter the overall positive prompt, the closer the result tends to stay to the original ink-and-pencil lineart style on paper.On the other hand, more detailed prompts or i2i tend to produce results closer to the full-color showcase images. Have fun.

Files with SHA256: 31b4e477064bdd7b6626c926ef9656591f189989b337a4856a29212142f25452