MIXWAI ILLUSTRIOUS

• Version 1.0.0-Stable • Core Base Model: Wai-Illustrious-SDXL v1.7.0

🚀 THE CORE UPGRADE & BASE MODEL

MixWai Illustrious is a specialized, production-grade custom merge built to deliver exceptional image clarity, flawless facial anatomy, and pristine textures straight out of a single-pass generation loop.

🧬 OFFICIAL BASE ARCHITECTURE

This mix is built and branched directly on top of the elite [Wai-Illustrious-SDXL (v1.7.0)](https://civarchive.com/models/827184/wai-illustrious-sdxl) core checkpoint.

To elevate this foundation far beyond standard setups, two critical core components have been completely upgraded and recalibrated:

🧠 Next-Gen NoobAI CLIP Text Encoder

The vanilla system was replaced with the latest-generation NoobAI Anime CLIP (CLIP-L & OpenCLIP BigG variant). This drastically improves overall prompt accuracy, minimizes cross-token color bleeding, and handles long Danbooru tag sequences flawlessly.

🎨 Calibrated & Enhanced Anime VAE

Embedded a deeply fine-tuned custom VAE that prevents color-space deflation. It ensures deep, velvety blacks, rich dynamic contrast, and perfectly saturated vibrant colors without any artifacting.

---

🎛️ PROMPT MATRIX

This checkpoint is strictly optimized and calibrated to work with the specific positive and negative prompts specified below.
Positive promt
embedding:lazynsfw, embedding:lazypos, rating:general, source:anime, (vibrant color grading, warm low-key lighting:1.2), cinematic composition, crisp bold lineart, clean flat shading, detailed beautiful eyes, magnificent huge anime eyes, sharp focus, volumetric lighting, deep shadows, volumetric shadows, strong high contrast, depth of field, cowboy shot, (chiseled shadows:1.15), ambient
Negative promt
embedding:lazyneg, embedding:lazywet, bad quality, lowres, worst quality, worst detail, low quality, sketch, noise, blurry, distorted, bad anatomy, bad eyes, bad fingers, extra fingers, deformed, mutated, text, watermark, logo, freckles, age spots,facial scars, scratches, dirty skin, white pupils, long eyelashes, dust, particles,

RESULT WITHOUT RECOMMENDED PROMPTS:

⚠️ OFFICIAL RECOMMENDATION: For a streamlined workflow, it is highly recommended to use the official Lazy Embeddings pack by Lazy_Art

👉 [DOWNLOAD MANDATORY EMBEDDINGS HERE](https://civarchive.com/models/1302719/lazy-embeddings-for-all-illustrious-noobai-pony-sdxl-models-lazypositive-lazynegative-positive-and-negative-plus-more)

Download lazypositive, lazynegative, lazynsfw, and lazywet, drop them into your /embeddings/ folder, and choose your target style envelope below:

⚙️ AUTHOR'S VERIFIED RUNTIME SETTINGS

To get the exact high-end, razor-sharp results shown in our official showcase, please align your KSampler with these exact calibrated parameters from my personal workflow:

Base Generation Stage:

* Resolution: 896 x 1216 (Highly recommended for perfect portrait compositions)

* Generation Steps: 25

* CFG Scale: 5.5 (Strictly maintain this sweet spot to keep line-art clean)

* Sampler Name: euler_ancestral Euler a)

* Scheduler: beta (Mandatory for smooth, organic shadow gradients)

* Post-process: Sharpen Image sharpen_radius: 1, sigma: 0.20, alpha: 0.30)

Upscale Stage:

* Upscale Resolution: 1024 x 1344

* Sampler Name: euler_ancestral

* Scheduler: beta

* CFG Scale: 3.5

* Steps: 25 (with start_at_step: 20 and end_at_step: 25)

* Post-process: Contrast Fix blend_factor: 0.30, blend_mode: soft_light) + Sharpen Image sharpen_radius: 1, sigma: 0.30, alpha: 0.40)

* Clip Skip Config: 2 (Default for advanced Illustrious text interpretation across both stages)

---

💻 OFFICIAL WORKFLOW REQUIREMENT

As the creator, I personally work and build my projects inside ComfyUI. This model is finely calibrated to its node execution logic.

📦 Local Generation

I highly recommend downloading and using my official custom workflow file uploaded right alongside this model:

👉 [DOWNLOAD THE WORKFLOW HERE](https://civarchive.com/models/2945068/comfyuimixinstance) ComfyUIMixInstance

☁️ No powerful GPU at home? (Cloud Generation via Comfy on Civitai)

Good news! For those who can’t generate resource-intensive SDXL models locally due to hardware limitations, this ComfyUIMixInstance workflow can be run on Civitai’s cloud servers. You’ll be able to run this ecosystem in the cloud without needing a high-performance PC!

👉 [ACCESS THE CLOUD GENERATOR HERE](https://comfy.civitai.com/)

---

📢 A MESSAGE TO THE COMMUNITY & FEEDBACK LOOP

As the author, I want to be fully transparent: this model is currently fine-tuned and thoroughly tested specifically for Anime and Semi-Realism styles. I haven't done extensive benchmarking on other exotic categories yet.

I would be incredibly grateful for your active feedback! Please generate your images, test the model's limits, and upload your creations to the gallery below. If you notice any bugs, weird artifacts, or areas that need improvement, leave a detailed review.

I will look into every piece of feedback and fix/improve those aspects in the upcoming v2.0 update! 🛠️

Enjoy the model, load your mandatory embeddings, use the official *ComfyUIMixInstance*workflow, and let's create some beautiful art together!