Check out my other sibling model for highly stylized semi-realistic anime illustration style art: ImmortalEternalMix

Check out my other sibling model for semi-realistic anime style art: MaiMixSR

MaiMix is an anime illustration style model which aims to produce high quality art with little prompting at low steps but even greater results with more complex prompts.

All sample images have been generated without the use of textual inversion/negative embeddings and VAE (except δ Variant which has VAE baked) at 512x768 resolution (no upscaling) with no extra features to display the strength of the model. For real world use it is recommended that ClearVAE and EasyNegative is used for even higher quality results. Hires.fix and/or Adetailer is recommended for images where the character is more distant for better results.

MaiMix δ Variant (VAE baked) is the latest version using a different model GhostNotesDelta based on a combination of line art and brush stroke illustration style with a hint of semirealism.

MaiMixHard is an updated version based on a line art illustration style.

MaiMixSoft is an updated version based on a brush stroke illustration style.

MaiMixBeta is the original generalist MaiMix with an illustration style closer to the MareColoris base model used.

NSFW is supported but NSFW LoRA's may be needed for better results.

Recommended Settings (for versions with VAE not baked):

Sampler: Euler or DPM++ 2M Karras (25-50 steps)
CFG Scale: 4 to 11 (7.5 preferred for balanced outputs)
Clip Skip: 2 for more variation but model works well with 1 as well.
Negatives (if not using EasyNegative): ((low quality!!, worst quality!!, monochrome noir)), AND bad anatomy, AND hideous extra bad hands, AND auxiliary missing malformed fingers, AND unrefined duplicate body parts, AND fused overlap unseparated conjoined, AND (strabismus, (red eyes)), AND fused merged additional arms, AND baby face, AND child, AND loli, AND childlike, AND young, AND (((corner text!, watermark!, logo!, signature!, words!, letters!, content!, artist name!, trademark!, inscription!, symbol!, hieroglyph!, glyphs!, type!, scribble!))), face tattoos, face paint, scars, misplaced multiple boobs, inaccurate nipples, Asian face, transsexual, hermaphrodite, deformed malformed, twisted, mutilated, blur, haze, censor

Medium Length Prompting Example:

wide telephoto waist up half portrait, stereoscopic dimensions, layered artistic depth|color shading, beautiful|charming stoic|casual developed mature adult|mother 1girl fashionista goth|student executive|soldier, eye reflections, complex dynamic colorful hair, choker, toned feminine muscles, natural promiscuous|lavish|seductive pose, tattoos, schematics graffiti, dark natural lighting, transcendent mech punk rock, cool focused energetic vibes, studio catchlight, outdoors street slums, CG visual effects background, vivid complementary|contemporary scenery, SLE, ultra detailed textures, masterpiece composition, professional|original illustration, close up, dutch angle shot, vibrant palette, semirealistic, digital full frame, sharp Photoshop|Lightroom brush stroke techniques, watercolor pencil paint style filter edit

Extreme Wildcard Prompt Example (Over-Prompting):

professional|commercial|finished colorful|pristine|dark|saintly reflective|iridescent|crystalline|textured polychromatic|immersive|transcendent|cinematic|surreal|phantasmagoria intricate|delicate|exquisite detail floral|snowflake|sunset tattooed fashionista|executive|goth|student|swordswoman|punk|seraph angled|dynamic|focused full portrait masterpiece CGI by wlop|WENJR|ufotable AND attractive|seductive|trendy|beautiful|mature and womanly|alluring|proportional|motherly|subservient|homely|casual|erotic with jewel-like color contacts and ((((salon dyed complex hair wearing a sleeveless|simple low-cut sportswear belt kimono|dress uniform jacket)))) AND contemporary|Desi|Arabian|Oriental|European|multinational|cybernetic traditional|elegant|sophisticated fractal|endless|entrancing|infinite|perpetual|abstract mural natural|ambient|moody ancient|opulent|fancy cityscape|dusk fantasy beauty showpiece outline BREAK aesthetic|pleasing|artistic|tasteful official 8K HDR anime|photorealistic outdoors|oceanic|scenic|picteresque depth|layered|dimensional landscape|background|theme|scenery|silhouette original film grain polaroid style AND enhanced|featured|complete|final|enriched rainbow|pastel|neon|psychedelic continuous|flowing|curvaceous|diffuse|eternal encased|juxtaposed|overlap underwater bubble jellyfish butterfly bloom concept AND absurdres ultra high quality sharp|extravagant|defined|precise|crisp resolution|clarity magical|cryptic|arcane mystic|majestic|dramatic|action cosmic energy modern|excessivism|analytical perfect|fantastic|breathtaking award|prize winning digital movie|blockbuster game release coverart wallpaper poster illustration image with hand-drawn thick|distinct brush|ink|watercolor stroke|impasto edge line shift|shimmering volumetric lighting using Unity|Unreal|Octane Engine|Render trending on Pixiv|Instagram|Artstation

Model Design Aim:

MaiMix is designed as a proof-of-concept weight merge model using LoRA's only to further train and fine-tune models to produce new model characteristics and improvements over base models. This newer model is also likely to produce better results when checkpoint merged with other models - although this was not tested.

Special Thanks To The Creators of:

MareColoris and GhostNotesDelta (now removed) models which were used as the base models and the creators of the following LoRA that are part of the Civitai community:

Japanese scene_20230609163753-000018.safetensors, [LoRa] Shoal 錦鱗 Concept (With multires noise version).safetensors, BG2_dpep4+silicon29dim64.safetensors, butterfly.safetensors, Concept_scenery_background.safetensors, crystal_v1.safetensors, improve_backgrounds.safetensors, Pyramid lora_Ghibli_n3.safetensors, AMechaSSS[color_theme,mecha musume, mechanical parts,robot joints,headgear].safetensors, bigeye.safetensors, Cube-10.safetensors, detailed_eye-10.safetensors, fate_kyrielight-10.safetensors, FBH3-000005.safetensors, kitagawa_marin_v1-1.safetensors, Loraeyes_V1.safetensors, LR-000007BB.safetensors, mathayom_uniform16.safetensors, matWomen_v2.safetensors, minamoto_no_raikou_v1.safetensors, QuickHandV2.safetensors, tattoo_lora01.safetensors, 眼睛双eyes.safetensors, add_detail.safetensors, bad_pic.safetensors, loraH(DiffLoRA)_FaceShadowTweaker_v1_dim4.safetensors, more_details.safetensors, SilvermoonMix2.safetensors, tangbohu-detailmaker_v2.5.safetensors, XenoDetailer.safetensors, AI_Lora-05.safetensors, colorful world_20230606112956-000017.safetensors, Cutesy4-10.safetensors, doukou.safetensors, midjourney_20230624181825.safetensors, style_Aobeni.safetensors, style_TOKKYU.safetensors, tachi-e.safetensors, wlop_offset.safetensors, Yoneyama Mai_v2.safetensors, 3DMM_V11.safetensors, add_sharpness.safetensors, epiNoiseoffset_v2-pynoise.safetensors, FilmVelvia2.safetensors, flat2.safetensors, hipoly_3dcg_v7-epoch-000012.safetensors, saturation.safetensors, tangbohu-line_1.0.safetensors, thickline_fp16.safetensors, LAS.safetensors, GachaSplash4.safetensors, vibrantProj-07.safetensors

Disclaimer:

• You cannot use MaiMix models to deliberately produce nor share illegal or harmful outputs or content

• The author claims no rights on the outputs you generate, you are free to use them and are accountable for their use which must not go against the provisions set in the CreativeML Open RAIL-M license