Update README.md

README.md

@@ -10,7 +10,6 @@ base_model:
 - Qwen/Qwen-Image
 - black-forest-labs/FLUX.1-Krea-dev
 tags:
-- mode-zoo
 - generative-models
 - flux
 - sdxl
@@ -24,43 +23,84 @@ tags:
 - ray
 - ray-models
 ---
-# Ray’s Models Zoo
-<img src="https://huggingface.co/MutantSparrow/Ray/resolve/main/_IMG_/RAY_LOGOT.png" width="240">
+```
+---
+license: apache-2.0
+language:
+  - en
+base_model:
+  - Tongyi-MAI/Z-Image-Turbo
+  - black-forest-labs/FLUX.1-dev
+  - lodestones/Chroma1-Base
+  - stabilityai/stable-diffusion-xl-base-1.0
+  - Qwen/Qwen-Image
+  - black-forest-labs/FLUX.1-Krea-dev
+tags:
+  - model-zoo
+  - generative-models
+  - flux
+  - sdxl
+  - chroma
+  - qwen-image
+  - krea
+  - z-image
+  - photographic
+  - analog
+  - creative-ai
+  - ray
+  - ray-models
+library_name: diffusers
+pipeline_tag: image-generation
+---
+```
 
+---
+
+# Ray’s Models Zoo (README  is WIP)
+<img src="https://huggingface.co/MutantSparrow/Ray/resolve/main/_IMG_/RAY_LOGOT.png" width="240">
 
 **_An AI repository of curated visual styles — a personal generative model collection._**
 
+---
 
 ## TL;DR
+
 A curated collection of custom-trained generative models built on personal assets across Flux-Dev, Krea, SDXL, Chroma, Qwen-Image, Z-Image, and others — all designed to deliver my own distinctive visual style, mostly photographic, but not exclusively.
 
+---
+
 ## Model Philosophy
-All models in this zoo were trained on a unique, personally-curated dataset built around my own photography. Most of it originates from years of street photography, later expanded and refined with synthetic images generated from earlier generations of my models — creating a self-reinforcing aesthetic loop.
 
-The overarching goal is to capture a natural, analog-leaning photographic sensibility without sacrificing creative range. Many of the models inherit a subtle film-grain texture, balanced contrast, and an overall organic feel. Others explore alternate aesthetics — oil painting, digital illustration, stylized comic rendering — but always through the lens of my own visual taste.
+All models in this zoo were trained on a **unique, personally-curated dataset** built around my own photography. Most of it originates from years of street photography, later expanded and refined with synthetic images generated from earlier generations of my models — creating a self-reinforcing aesthetic loop.
 
-Across the different families (SDXL, Flux, Qwen-Image, Chroma, Z-Image), the intention remains the same:
-a deeply personal twist on photorealism and style-driven rendering, built from scratch and continuously refined.
+The overarching goal is to capture a **natural, analog-leaning photographic sensibility** without sacrificing creative range. Many of the models inherit a subtle film-grain texture, balanced contrast, and an overall organic feel. Others explore alternate aesthetics — oil painting, digital illustration, stylized comic rendering — but always through the lens of my own visual taste.
+
+Across the different families (SDXL, Flux, Qwen-Image, Chroma, Z-Image), the intention remains the same:  
+**a deeply personal twist on photorealism and style-driven rendering, built from scratch and continuously refined.**
+
+These models are designed for **everyone**:
 
-These models are designed for everyone:
 - Artists and photographers wanting high-fidelity realism  
 - Illustrators looking for stylistic control  
 - Prompt engineers who want predictable behavior  
-- Anyone exploring generative visuals with a distinctive mood
+- Anyone exploring generative visuals with a distinctive mood  
 
 Some families are more specialized — like the SDXL branch, with portrait-centric Raymnants, painterly Rayctifier, and comic-styled Rayburn — while newer architectures such as Flux and Qwen tend to be more powerful generalists with a strong stylistic backbone.
 
-## Repository Structure
+---
+
+# Repository Structure
+
 ### `CHROMA/`
 Experiments built on Chroma’s compact and expressive architecture, focusing on stylized realism and fast inference.  
-Contains: Raydiant v1.0
+Contains: **Raydiant v1.0**
 
 ### `FLUX/`
 A broad suite of Flux-based models, ranging from generalist baselines to highly tuned photographic variants. These models offer strong detail rendering, stable composition, and a distinctive analog-tinted style.
 
 ### `FLUX_KREA/`
 Models trained on Flux-Krea, leveraging Krea’s training optimizations to push texture richness, skin rendering, and consistency.  
-Contains: Rayflux Krea v1.0
+Contains: **Rayflux Krea v1.0**
 
 ### `QWEN_IMAGE/`
 Models built on the Qwen-Image architecture, offering strong generalization, expressive color grading, and excellent multi-domain performance.
@@ -68,41 +108,52 @@ Models built on the Qwen-Image architecture, offering strong generalization, exp
 ### `SDXL/`
 A large family of specialized SDXL models, spanning portrait photography, oil painting, stylized illustration, and experimental aesthetics.
 
-### `Z_IMG/`
-Z-Image–based experiments focusing on fast, portable models with a clean photographic style.
+### `Z-IMAGE/`
+Z-Image based models with fast inference and clean photographic output.
+
+---
+
+# Model Families Overview
 
-## Model Families Overview
-### CHROMA — Raydiant
-Based on the open-source Chroma architecture, Raydiant transfers the photographic DNA of earlier Rayflux and Raymnants models into a lighter, faster framework.
+### CHROMA — Raydiant  
+Lightweight analog-photography oriented experiments.
 
-### FLUX — Rayflux Series
-Flagship line: photographic realism, analog grain, strong expressiveness.
+### FLUX — Rayflux  
+Flagship generalist line with analog grain and strong realism.
 
-### FLUX_KREA — Rayflux Krea
-An experimental refinement using Krea's FluxDev optimizations.
+### FLUX_KREA — Rayflux Krea  
+An experimental artistic refinement of Flux with strong skin rendering.
 
-### QWEN_IMAGE — RayQwest Series
-State-of-the-art, Apache-2 licensed, excellent prompt adhesion.
+### QWEN_IMAGE — RayQwest  
+Modern generalist models with great prompt adhesion and expressive range.
 
-### SDXL — Raymnants / Rayctifier / Rayburn
-Fast, low-VRAM models ranging from portrait realism to oil painting and comic styles.
+### SDXL — Raymnants / Rayctifier / Rayburn  
+Fast, strong stylistic models across portraits, painting, and comics.
 
-### Z_IMG — Rayzist
-Fast open-source architecture between SDXL-HCL and Flux-quality outputs.
+### Z-IMAGE-TURBO — Rayzist  
+Fast, low-VRAM model that performs above SDXL tier has a good prompt understanding.
 
-## Flagship Lines
-1. Rayflux (Flux)  
-2. RayQwest (Qwen-Image)  
-3. Raymnants (SDXL)  
-4. Rayzist (Z-Image)
+---
+
+# Flagship Lines
+
+1. **Rayflux** (Flux Dev)  
+2. **RayQwest** (Qwen-Image)  
+3. **Raymnants** (SDXL)  
+4. **Rayzist** (Z-Image Turbo)
+
+---
 
 # FLUX Family — Rayflux Models
 
-## Rayflux Horndog
-A model slightly overtuned on anatomy; it captures the photographic flair of the Rayflux line but tends to produce accidental NSFW content.
+> **Default settings:** Euler · Beta Guidance 2.5 · CFG 1 · Steps 16 · Shift/Max Shift 1/1.5. </br>
+Feel free to experiment with newer samplers/schedulers though! (ER_SDE, Res2S, Bong_tangent, Beta57....)
 
-**Base model:** Flux-dev  
-**Recommended settings:** Euler · Beta Guidance 2.5 · CFG 1 · Steps 16 · Shift/Max Shift 1/1.5
+---
+
+## **Rayflux Horndog**
+
+A model slightly overtuned on anatomy; it captures the photographic flair of the Rayflux line but tends to produce accidental NSFW content.
 
 <p align="center">
   <!-- IMAGE: rayflux-horndog-01.png -->
@@ -110,11 +161,11 @@ A model slightly overtuned on anatomy; it captures the photographic flair of the
   <!-- IMAGE: rayflux-horndog-03.png -->
 </p>
 
-## Rayflux Photoplus
-A best-in-class photorealistic model with excellent analog-style grain, stable anatomy, and broad creative range.
+---
+
+## **Rayflux Photoplus**
 
-**Base model:** Flux-dev  
-**Recommended settings:** Euler · Beta Guidance 2.5 · CFG 1 · Steps 16 · Shift/Max Shift 1/1.5
+Best-in-class photorealism with excellent analog grain and strong creative fidelity.
 
 <p align="center">
   <!-- IMAGE: rayflux-photoplus-01.png -->
@@ -122,11 +173,11 @@ A best-in-class photorealistic model with excellent analog-style grain, stable a
   <!-- IMAGE: rayflux-photoplus-03.png -->
 </p>
 
-## Rayflux v1.0
-Less photorealistic than Photoplus but more vibrant and creatively expressive.
+---
+
+## **Rayflux v1.0**
 
-**Base model:** Flux-dev  
-**Recommended settings:** Euler · Beta Guidance 2.5 · CFG 1 · Steps 16 · Shift/Max Shift 1/1.5
+Vibrant, creative, less strictly photorealistic than Photoplus; an excellent all-rounder.
 
 <p align="center">
   <!-- IMAGE: rayflux-v1-01.png -->
@@ -134,11 +185,10 @@ Less photorealistic than Photoplus but more vibrant and creatively expressive.
   <!-- IMAGE: rayflux-v1-03.png -->
 </p>
 
-## Rayflux v1.0 (UNet-only)
-Same aesthetic profile as Rayflux v1.0, in UNet-only format.
+---
 
-**Base model:** Flux-dev  
-**Recommended settings:** Euler · Beta Guidance 2.5 · CFG 1 · Steps 16 · Shift/Max Shift 1/1.5
+## **Rayflux v1.0 (UNet-only)**  
+Same aesthetic as v1.0, packaged as UNet-only.
 
 <p align="center">
   <!-- IMAGE: rayflux-v1-unet-01.png -->
@@ -146,11 +196,11 @@ Same aesthetic profile as Rayflux v1.0, in UNet-only format.
   <!-- IMAGE: rayflux-v1-unet-03.png -->
 </p>
 
-## Rayflux v3.0
-Evolution of the v1.0 line, trained on the same dataset as Photoplus.
+---
+
+## **Rayflux v3.0**
 
-**Base model:** Flux-dev  
-**Recommended settings:** Euler · Beta Guidance 2.5 · CFG 1 · Steps 16 · Shift/Max Shift 1/1.5
+Evolution of v1.0 using the Photoplus dataset — balanced fidelity + creativity.
 
 <p align="center">
   <!-- IMAGE: rayflux-v3-01.png -->
@@ -158,11 +208,10 @@ Evolution of the v1.0 line, trained on the same dataset as Photoplus.
   <!-- IMAGE: rayflux-v3-03.png -->
 </p>
 
-## Rayflux v3.x Series (v3.1 → v3.4)
-A consolidated series of refinement experiments designed to progressively reduce unintended biases and stabilize anatomy.
+---
 
-**Base model:** Flux-dev  
-**Recommended settings:** Euler · Beta Guidance 2.5 · CFG 1 · Steps 16 · Shift/Max Shift 1/1.5
+## **Rayflux v3.x (3.1–3.4)**  
+Refinement line aimed at de-biasing Horndog and stabilizing anatomy.
 
 <p align="center">
   <!-- IMAGE: rayflux-v3x-01.png -->
@@ -170,14 +219,95 @@ A consolidated series of refinement experiments designed to progressively reduce
   <!-- IMAGE: rayflux-v3x-03.png -->
 </p>
 
-## Rayflux v4.0
-A better-tuned evolution of Horndog, retaining its strengths while offering more stable control.
+---
+
+## **Rayflux v4.0**
 
-**Base model:** Flux-dev  
-**Recommended settings:** Euler · Beta Guidance 2.5 · CFG 1 · Steps 16 · Shift/Max Shift 1/1.5
+A better-tuned evolution of Horndog, keeping detail but increasing control.
 
 <p align="center">
   <!-- IMAGE: rayflux-v4-01.png -->
   <!-- IMAGE: rayflux-v4-02.png -->
   <!-- IMAGE: rayflux-v4-03.png -->
-</p>
+</p>
+
+---
+
+# CHROMA Family — Raydiant
+
+## **Raydiant v1.0**
+
+Raydiant v1.0 is a Chroma-based model spun off from Chroma v48, tuned for uncensored analog photography with strong contrast and mild fuzziness. Performs well across bold illustrated styles too.
+
+**Settings:** Euler Beta · 16 steps · CFG 4.5 · min_length=3  
+**Caveat:** very colorful, fully uncensored.
+
+<p align="center">
+  <!-- IMAGE: raydiant-v1-01.png -->
+  <!-- IMAGE: raydiant-v1-02.png -->
+  <!-- IMAGE: raydiant-v1-03.png -->
+</p>
+
+---
+
+# FLUX_KREA Family — Rayflux Krea
+
+## **Rayflux Krea v1.0**
+
+Photorealistic checkpoint with excellent skin textures and artistic versatility.
+
+**Settings:** Euler Simple · 12–30+ steps · CFG 1 · low shift  
+**Note:** best at 30+ steps, upscaler recommended for lower steps.
+
+<p align="center">
+  <!-- IMAGE: rayflux-krea-v1-01.png -->
+  <!-- IMAGE: rayflux-krea-v1-02.png -->
+  <!-- IMAGE: rayflux-krea-v1-03.png -->
+</p>
+
+---
+
+# QWEN_IMAGE Family — RayQwest
+
+## **RayQwest v1.0**
+
+Removes Qwen's plasticness and restores analog grit + Ray-style vibe. Excellent prompt adhesion and range.
+
+**Settings:** Euler Simple · 16 steps · CFG 1  
+**Note:** LoRAs work but reduce strength.
+
+<p align="center">
+  <!-- IMAGE: rayqwest-v1-01.png -->
+  <!-- IMAGE: rayqwest-v1-02.png -->
+  <!-- IMAGE: rayqwest-v1-03.png -->
+</p>
+
+---
+
+## **RayQwest v0.7**
+
+Stronger analog feel, noisier, flatter dynamic range — but strong vibes and artistic cohesion.
+
+**Settings:** same as v1.0
+
+<p align="center">
+  <!-- IMAGE: rayqwest-v07-01.png -->
+  <!-- IMAGE: rayqwest-v07-02.png -->
+  <!-- IMAGE: rayqwest-v07-03.png -->
+</p>
+
+---
+
+## **RayQwest v0.65**
+
+Dreamy, saturated, softer — trained on a different dataset, atmospheric and expressive.
+
+**Settings:** same as v1.0
+
+<p align="center">
+  <!-- IMAGE: rayqwest-v065-01.png -->
+  <!-- IMAGE: rayqwest-v065-02.png -->
+  <!-- IMAGE: rayqwest-v065-03.png -->
+</p>
+
+---