<h2 id="description">📋 Description</h2>A merge based on KREA2 Turbo, optimized for fast and high-quality generation. This model combines the strengths of the KREA2 architecture with custom tuning for improved results across various styles.<pre><code>Three versions have been posted: SAFE (clean art), BASE (nudity allowed), and NSFW (practically anything goes).</code></pre><hr /><h2 id="recommended-generation-settings">⚙️ Recommended Generation Settings</h2>ParameterValueSteps8SamplerEuler aSchedule typeBong TangentCFG scale1Seed3448636919Resolution1536×2048Modelfascium_SFWKREA2_fp8Model hash98d064dc8fModule 1qwen_image_vaeModule 2qwen3vl_4b_fp8_scaledDiffusion in Low BitsAutomatic (fp16 LoRA)RNGCPUStyle SelectorEnabled (Style: <code>base</code>)Krea2 Enhancer✅ Enabled<hr /><h2 id="tips-and-notes">💡 Tips &amp; Notes</h2><ul><li>Low CFG (1.0) — This model is designed to work best with a very low CFG scale. Higher values may cause artifacts or oversaturation.</li><li>Few steps needed — Only 8 steps are required thanks to the Turbo architecture. Going higher won't improve quality and will just waste time.</li><li>Krea2 Enhancer — Make sure to enable the Krea2 Enhancer for best results.</li><li>Style Selector — Set to <code>base</code> by default. You can experiment with other styles, but disable randomization for consistent output.</li><li>Resolution — Tested at 1536×2048 (portrait). The model supports various aspect ratios, but keep total pixel count around ~3MP for optimal performance.</li><li>fp8 quantized — This is an fp8 version for reduced VRAM usage while maintaining quality.</li></ul><hr /><h2 id="trigger-words-prompt-guide">🖼️ Trigger Words / Prompt Guide</h2>No specific trigger words required. Write natural, descriptive prompts. The model responds well to detailed style and composition descriptions.<hr /><h2 id="known-limitations">⚠️ Known Limitations</h2><ul><li>SFW-focused merge</li><li>Best results with the exact settings above — deviating from recommended CFG/sampler may degrade output quality</li></ul><hr />Feel free to adjust the model name, add your own example images section, or tweak any details to better match your vision! 🎨

FASCIUM KREA2

The bilingual base merge. 31 photographic voices × 99 painterly voices in one checkpoint, bench-validated 20/20 on the Showstopper Dataset. Transition frames (photo↔oil, engraving↔color, cel↔photo city), nested worlds, recursive witnesses and painted shadows — registers no single-parent model can reproduce.Inside the drop:<ul><li><code>fascium_masterpiece_int8_convrot.safetensors</code> — production edition</li><li><code>fascium_masterpiece_int4_convrot.safetensors</code> — fast draft edition</li><li><code>MASTERPIECE DATASET</code> — 20 showstopper prompts + canon variants born during the bench (monochrome-in-color, frontal regrowth, shadow choir with mics)</li></ul>BF16 note — read once: the full-precision master will not be released, now or ever. It remains the private archival source; every published edition is ConvRot-quantized from it and bench-validated against it. The master is the lineage of the series, not a product. Please do not request or re-upload it.

Files with SHA256: 1f6eb9da561694007681d34143cb603642533435e82b1cce7f2787782c48238d