<p>A fairly straightforward style lora trained on the work of Junpaku Karen. The lora should excel with anime-centered 2D digital art with detailed clothing, varied lighting, interplay between brighter colors and shades of blue, and dynamic but detailed backgrounds.</p><p></p><h1 id="prompting">Prompting</h1><p>You should use the tags "emjjunkarstyle, junpaku_karen" in the prompt. Tags work best for the Illustrious and Pony versions of this lora. The Anima, Chroma, and Z-Image versions should work fine with tags, descriptive prompts, or a combination of both. </p><p></p><p>For the Z-Image version I'd suggest guiding the style a bit more with a prompt of "emjjunkarstyle, junpaku_karen, The art style is high-fidelity modern digital anime illustration style characterized by crisp thin line art and a soft muted pastel color palette. The shading technique combines sharp, precise cel-shading for defining clothing folds and hair strands with gentle airbrushed gradients to add softness to the skin which creates a polished smooth finish. The lighting is diffused and flat but features distinct glossy white highlights on reflective surfaces." Conversely, I'd generally avoid doing so with the Anima version. You can get good results with that kind of additional style prompt but it tends to push things a bit too far in that direction. </p><p></p><h2 id="model-specific-tips">Model Specific Tips</h2><p><strong>Chroma</strong>: This style lora should stack fairly well with other loras. The gallery for the Chroma version should have some examples where I use a very low strength location focused lora, a medium strength character lora, and this lora at full strength. In those instances the location lora was usually at 0.1 strength, the character lora at a strength between 0.5 and 0.7 and this lora at 1.0.</p><p></p>

Junpaku Karen Style

anima_base1p0_JunpakuKaren_Style_32.safetensors

Files with SHA256: afb79685c56219ca0250d4315db29e088a421f969da2468bf80c7f0fb49456f3