Recommended: CFG: 4-6 minimum 20 StepsEuler a , karrasWith dmd 2: CFG: 1.3 ,6 steps(at least 4 steps), Euler a , karras V2 improves the overall style and fixes the eye issues from V1. I intentionally generated a few hundred high-quality images with good lighting and tags, used them to train two LoRAs, and baked those into V1 to create V2. After that, I blended a small amount of V1 back in by re-merging some of its blocks at a low percentage. <h2 id="i-wanted-to-share-my-first-&quot;finished&quot;-merged-model-here.-i-made-a-bid-so-you-should-be-able-to-try-it-on-site-after-the-auction-ends.">I wanted to share my first "finished" merged model here. I made a bid, so you should be able to try it on site after the auction ends.</h2><h3 id="this-model-tries-to-achieve-a-look-of-&quot;comicillustration-realism&quot;i-merged-it-for-myself-and-am-quite-happy-for-it.-ive-asked-the-creators-of-the-used-models-in-the-merging-process-for-permission-to-release-this-model.this-checkpoint-struggles-with-eyes-without-facedetailer-if-the-character-is-further-away.-i-hope-to-fix-that-in-the-future.this-model-is-good-at-illustration-realism-semi-realism-and-is-able-to-do-cartoontoon-images.ive-used-simple-and-block-merge-to-create-this-model.ive-used-dmd2(with-cfg-1.3-6-steps-euler-a-karras)-facedetailer-and-embedding:lazypos-for-the-sample-pictures.-thats-it.creation-processbase-merge:">This model tries to achieve a look of "comic/Illustration-realism" I merged it for myself and am quite happy for it. I´ve asked the Creators of the used models in the merging process for permission to release this model. This Checkpoint struggles with eyes without facedetailer if the Character is further away. I hope to fix that in the future. This model is good at illustration-realism, semi-realism and is able to do cartoon/toon images. I´ve used Simple and Block-Merge to create this model. I´ve used DmD2(With cfg 1.3 , 6 steps, euler a, karras) , FaceDetailer and embedding:lazypos for the sample pictures. Thats it. Creation Process Base Merge:</h3><ul><li><code>MistoonAnime</code></li><li><code>NovaCartoon</code></li><li><code>NovaComic</code></li><li><code>NovaMature</code></li><li>Some custom-trained style LoRAs</li></ul>→ Combined to achieve the general "realistic cartoon look".<ul><li>First Merge:<ul><li>Output Block 1 replaced with Juggernaut, to achieve more realistic facial structure</li></ul></li><li>Second Merge:<ul><li>Multiple intermediate ratios were merged and recombined to balance realism and stylization, added different style loras to them and recombined around 8 merged checkpoints with different ratios until i was satisfied</li></ul></li><li>Final Merge:<ul><li>Introduced Output Block 1 from NovaReality at ~50%.</li><li>Mid 00 and Mid 01 blocks blended from NovaCartoon &amp; NovaMature</li><li>Took the Base merge and replaced in4/in5 out1 with the Blocks of the final merge to keep the comic/cartoon look but have the more realistic proportions and change the prompt behaviour.</li><li>Introduced Output Block 1 from NovaReality again with 20% to add a bit more realism to the anatomy </li></ul></li></ul>Feel free to use this model for your own merges. I’d appreciate it if you mention or credit when you do.<h3 id="disclaimer">⚠️ Disclaimer</h3>This model was created purely for artistic and educational purposes. I do not take responsibility for any outputs, interpretations, or derivative works generated using this model. Users are solely responsible for ensuring that any generated content complies with ethical guidelines, community rules, and local laws.This model should not be used to produce, share, or promote harmful, or offensive material. Use responsibly, credit where appropriate.

IllustraCore

I´ve trained loras with very good quality images and baked the loras in to enhance the overall quality like eyes.

Files with SHA256: 1275eb33bc948fc01c9bf94b079e80fe50f50bdf041144bc93c7e0b7f2419f58