r3f24rwegwerg

r3f24rwegwerg_v10ewgrwegrwerg.safetensors

v1.0ewgrwegrwerg

Just a model I use trying to move it around

Stupid model

stupidModel_v10.safetensors

v1.0

Rainmaker Illutrious — Booru-tag-Based Model Instruction0) Overview<ul><li>Type: High-quality anime/stylized image generator.</li><li>Strengths: Works best with structured tag prompts. Preloaded with 1600+ anime characters (callable by name tags). Supports artist/style tags for stylistic control. Extremely responsive to negative prompts for quality filtering.</li></ul>1) Input / Output FormatInput<ul><li>Positive Prompt (pos): Comma-separated tag list with optional weight (tag:weight).</li><li>Negative Prompt (neg): Comma-separated tags for suppression.</li><li>Params: resolution, steps, guidance, sampler, seed.</li></ul>Output<ul><li>Model returns image(s) faithful to tags, weights, and parameters.</li></ul>2) Prompt Construction Rules2.1 Positive Prompt (pos)<ul><li>Always start with quality tags: best quality, amazing quality, very aesthetic, absurdres, {{{masterpiece}}}</li><li>Add style tags with weight if needed: (chiaroscuro:1.2), (ciloranko:0.8), (ke-ta:0.9), (ningen mame:1.1)</li><li>Add subject + details: 1girl, pink eyes, white long hair, black_skin, black shirt, purple tie, black shorts, very loose black coat with purple decorations</li><li>Add scene/environment tags: apocalypse background, holding rapier, cool mood</li><li>Separate with commas, avoid natural language sentences.</li></ul>2.2 Negative Prompt (neg)Use a concise, powerful set:text, watermark, censored, bar_censor, split, 3d, lowres, (worst quality:1.2), (bad quality:1.2), bad anatomy, bad hands, sketch, jpeg artifacts, signature, old, extra head, conjoined, multiple views<ul><li>Deduplicate tokens.</li><li>Use weight on worst/bad quality to strongly suppress them.</li><li>Add specific negatives (extra fingers, deformed hands) if needed.</li></ul>2.3 ParametersRecommended defaults:resolution = 1536x2048 steps = 28~32 guidance = 5.5~6.5 sampler = Euler a / DPM++ 2M Karras (preferred) seed = auto (or fixed for reproducibility)3) Examplepos:best quality, amazing quality, very aesthetic, absurdres, (ciloranko:0.78), (rhasta:0.9), tidsean, (ke-ta:0.9), (chiaroscuro:1.27), (ningen mame:1.1), {{{masterpiece}}}, 1girl, pink eyes, black_skin, white long hair, black shirt, purple tie, black shorts, very loose black coat with purple decorations, apocalypse background, holding rapier, cool mood, masterpieceneg:text, watermark, censored, bar_censor, split, 3d, lowres, (worst quality:1.2), (bad quality:1.2), bad anatomy, bad hands, sketch, jpeg artifacts, signature, old, extra head, conjoined, multiple viewsparams:resolution = 1536x2048 steps = 30 guidance = 5.5 sampler = auto seed = auto✅ Key Points<ul><li>Keep tag format, no full sentences.</li><li>Use weights (tag:1.x) to boost style elements.</li><li>Always include quality stack and negative prompt cleanup.</li><li>Characters from built-in DB can be called simply by name tag, e.g. asuka_langley, (asuka_langley:1.1)</li><li>Artist styles only if explicitly requested: (by kei:1.0), (by mikoto_shinkai:1.0)</li></ul>

Rainmaker Illustrious

🌧️ Rainmaker Illustrious 2.0
Where Aesthetic Mastery Meets Technical Excellence
Rainmaker Illustrious 2.0 marks a significant evolution in anime AI art generation, representing the pinnacle of aesthetic-driven checkpoint development. Built upon the robust Illustrious XL architecture, this checkpoint delivers exceptional visual quality with enhanced control over artistic expression.

✨ What Makes V2.0 Revolutionary
🎨 Superior Aesthetic Control

Fine-tuned for harmonious color palettes and balanced compositions
Excels at creating visually stunning scenes with natural flow
Enhanced understanding of atmospheric lighting and mood

👥 Extensive Character Knowledge

Pre-trained recognition of 1000+ anime & game characters
Generate iconic characters without additional LoRAs
Accurate character designs from: Genshin Impact, Blue Archive, Honkai, Fate, Hololive, Azur Lane, popular seasonal anime, and more
Simply prompt the character name - the model knows their canonical appearance

🖌️ Enhanced Style Versatility

Seamless blending of multiple artist styles with precise weighting
Better interpretation of both Danbooru tags and natural language prompts
Improved consistency across varied art styles (from soft watercolor to sharp anime cel)

👤 Refined Anatomy & Detail

Advanced hand and facial feature accuracy
Superior clothing physics and fabric rendering
Enhanced fine details (hair strands, jewelry, intricate patterns)

📐 Optimized Prompting Efficiency

Responds exceptionally well to the Golden Formula structure
Stronger recognition of emoticon expressions (;), ^_^, :d)
Better handling of complex multi-character scenes


🎯 Recommended Settings
CFG Scale: 5.0-6.5 (sweet spot: 5.5-6.0)
Steps: 24-32 (optimal: 26-28)
Resolution: 1024×1024 to 1536×1536
Sampler: DPM++ 2M Karras / Euler A

💡 Ideal For

Character illustrations with rich emotional expression (no LoRAs needed!)
Fanart & crossover art featuring popular anime/game characters
Scenic compositions with atmospheric depth
Style experimentation (artist blending, hybrid aesthetics)
High-resolution outputs requiring minimal post-processing
Creative projects demanding consistent aesthetic quality


🔑 Key Improvements from V1.0
✅ 40% better color harmony and saturation balance
✅ Enhanced detail retention at high resolutions
✅ Improved style consistency across generations
✅ Better natural language understanding for scene descriptions
✅ Refined negative prompt response for cleaner outputs
✅ Expanded character recognition - draw your favorites instantly

🎮 Character Prompting Examples
Simple character generation:
"raiden_shogun, 1girl, masterpiece, best_quality, purple_eyes, 
long_hair, purple_hair, japanese_clothes, outdoors, newest"

Multi-character scene:
"2girls, mona_(genshin_impact), fischl_(genshin_impact), 
sitting, cafe, chatting, masterpiece, best_quality"
No LoRAs, no extra training - just pure checkpoint power!

📝 Pro Tips

Artist weights work best between (0.6-0.9) for optimal style blending
Use quality tags hierarchy: masterpiece, best_quality, amazing_quality, newest
Leverage atmospheric keywords: soft_lighting, cinematic_composition, depth_of_field
Character names use underscores: character_name_(series) format
For NSFW content: maintain aesthetic focus with proper anatomical tags


🚀 Quick Start Prompt Template
1girl, [character_name], masterpiece, best_quality, amazing_quality,
[physical_features], [outfit_details], [pose_action],
[environment], [lighting], (artist_name:0.8), absurdres, newest
Negative:
worst_quality, low_quality, bad_anatomy, watermark, 
signature, text, jpeg_artifacts, blurry

🌟 Community Note
Rainmaker Illustrious 2.0 is designed to push creative boundaries while maintaining production-ready quality. Whether you're crafting character art from your favorite series, exploring unique styles, or producing professional illustrations, this checkpoint adapts to your artistic vision with remarkable consistency.
The extensive built-in character library means you can focus on creativity rather than hunting for LoRAs!
Every download supports ongoing development and helps refine future iterations! 🎨

Compatible with: Illustrious XL pipeline, ComfyUI, Automatic1111
Training: Custom dataset optimized for aesthetic excellence + character recognition
Base: Illustrious XL architecture

Rainmaker_Illutrious_V2.0.safetensors

V2.0

NEW REALISTIC IL CHECKPOINT FOR COSPLAY: <a target="_blank" rel="noopener noreferrer nofollow" href="https://tensor.art/models/931644552488475250/Rainmaker-Realistic-Illustrious-V1.0">LINK HERE</a>Rainmaker Illutrious — Booru-tag-Based Model Instruction0) Overview<ul><li>Type: High-quality anime/stylized image generator.</li><li>Strengths: Works best with structured tag prompts. Preloaded with 1600+ anime characters (callable by name tags). Supports artist/style tags for stylistic control. Extremely responsive to negative prompts for quality filtering.</li></ul>1) Input / Output FormatInput<ul><li>Positive Prompt (pos): Comma-separated tag list with optional weight (tag:weight).</li><li>Negative Prompt (neg): Comma-separated tags for suppression.</li><li>Params: resolution, steps, guidance, sampler, seed.</li></ul>Output<ul><li>Model returns image(s) faithful to tags, weights, and parameters.</li></ul>2) Prompt Construction Rules2.1 Positive Prompt (pos)<ul><li>Always start with quality tags: best quality, amazing quality, very aesthetic, absurdres, {{{masterpiece}}}</li><li>Add style tags with weight if needed: (chiaroscuro:1.2), (ciloranko:0.8), (ke-ta:0.9), (ningen mame:1.1)</li><li>Add subject + details: 1girl, pink eyes, white long hair, black_skin, black shirt, purple tie, black shorts, very loose black coat with purple decorations</li><li>Add scene/environment tags: apocalypse background, holding rapier, cool mood</li><li>Separate with commas, avoid natural language sentences.</li></ul>2.2 Negative Prompt (neg)Use a concise, powerful set:text, watermark, censored, bar_censor, split, 3d, lowres, (worst quality:1.2), (bad quality:1.2), bad anatomy, bad hands, sketch, jpeg artifacts, signature, old, extra head, conjoined, multiple views<ul><li>Deduplicate tokens.</li><li>Use weight on worst/bad quality to strongly suppress them.</li><li>Add specific negatives (extra fingers, deformed hands) if needed.</li></ul>2.3 ParametersRecommended defaults:resolution = 1536x2048 steps = 28~32 guidance = 5.5~6.5 sampler = Euler a / DPM++ 2M Karras (preferred) seed = auto (or fixed for reproducibility)3) Examplepos:best quality, amazing quality, very aesthetic, absurdres, (ciloranko:0.78), (rhasta:0.9), tidsean, (ke-ta:0.9), (chiaroscuro:1.27), (ningen mame:1.1), {{{masterpiece}}}, 1girl, pink eyes, black_skin, white long hair, black shirt, purple tie, black shorts, very loose black coat with purple decorations, apocalypse background, holding rapier, cool mood, masterpieceneg:text, watermark, censored, bar_censor, split, 3d, lowres, (worst quality:1.2), (bad quality:1.2), bad anatomy, bad hands, sketch, jpeg artifacts, signature, old, extra head, conjoined, multiple viewsparams:resolution = 1536x2048 steps = 30 guidance = 5.5 sampler = auto seed = auto✅ Key Points<ul><li>Keep tag format, no full sentences.</li><li>Use weights (tag:1.x) to boost style elements.</li><li>Always include quality stack and negative prompt cleanup.</li><li>Characters from built-in DB can be called simply by name tag, e.g. asuka_langley, (asuka_langley:1.1)</li><li>Artist styles only if explicitly requested: (by kei:1.0), (by mikoto_shinkai:1.0)</li></ul>

Files with SHA256: 3428a37cf14f02c032e8bd334c17c97e2f89806dc5a24030e901e265826061ed