diff --git a/README.md b/README.md
index 01ea26b..4bd7c78 100644
--- a/README.md
+++ b/README.md
@@ -19,6 +19,21 @@ Nodepack under development. Manual written by AI, please open issue if something
+## About This Nodepack
+
+Primere Nodes is not a collection of individual utility nodes. It's a **consistent full-system solution** architected around a central orchestrator (**Primiere Model Control**) that automatically adapts all generation parameters (sampling, CFG, VAE, CLIP, encoders, LoRAs, refiners) based on selected model type or individual checkpoint.
+
+**Why the complexity?** Supporting 25+ diverse model architectures (SD1, SDXL, Flux, Hunyuan, etc.) with conflicting requirements demands a system-level approach. Each model expects different schedulers, sigma ranges, CFG scales, CLIP handling, and attention mechanics. Manual parameter tweaking per model is unsustainable at scale. Primere solves this by:
+
+1. **Saving per-model settings** to JSON configs
+2. **Auto-loading** correct settings when checkpoint selected
+3. **Enforcing compatibility** between model type and network types (removing incompatible LoRAs, etc.)
+4. **Providing override capability** for testing/optimization, then re-saving for future use
+
+The workflow appears complex because it **handles real-world generation scenarios** (model switching, quality filtering, preview building, multi-stage refinement) that naive single-sampler approaches ignore.
+
+
+
> [!TIP]
> ## Something different: Universal API node 🚀
> Explore the Uniapi quick manual: **[PrimereApiProcessor (Uniapi) guide](Workflow/Manual/nodes/uniapi.md)**.
@@ -35,11 +50,15 @@ Nodepack under development. Manual written by AI, please open issue if something
Core generation pipeline. Single prompt input with full model concept support.
-**Supported model concepts:** SD1, SD2, SDXL, Illustrious, SD3, StableCascade, Chroma, Z-Image, Turbo, Flux, Nunchaku, QwenGen, QwenEdit, WanImg, KwaiKolors, Hunyuan, Playground, Pony, LCM, Lightning, Hyper, PixartSigma, SANA1024, SANA512, AuraFlow. Future support: HiDream, Mochi, WanT2V, WanI2V, Cosmos, Flux2, SSD, SegmindVega, KOALA, StableZero, SV3D, SD09, StableAudio, LTXV.
+**Central Orchestrator:** **Primiere Model Control** node reads selected checkpoint, auto-loads saved settings (sampler, CFG, steps, VAE, CLIP, encoders, LoRAs, refiners), and distributes `control_data` tuple to all downstream nodes for automatic parameter application.
+
+**Supported model concepts:** `SD1, SD2, SDXL, Illustrious, SD3, StableCascade, Chroma, Z-Image, Turbo, Flux, Nunchaku, QwenGen, QwenEdit, WanImg, KwaiKolors, Hunyuan, Playground, Pony, LCM, Lightning, Hyper, PixartSigma, SANA1024, SANA512, AuraFlow`.
+
+Future support: `HiDream, Mochi, WanT2V, WanI2V, Cosmos, Flux2, SSD, SegmindVega, KOALA, StableZero, SV3D, SD09, StableAudio, LTXV`.
### Minimal workflow features:
-- Central model concept selector node controls sampler, VAE, CLIP settings per model type
+- **Central model orchestrator:** Auto-configures `sampler`, `CFG`, `steps`, `VAE`, `CLIP`, `encoders`, `attention`, LCM, speed or SRPO `LoRAs` or refiners per `model type` or `model concept type`
- Automatic model keyword insertion to prompt
- Prompt selector to any prompt sources
- Prompt can be saved to `CSV` file directly from the prompt input nodes
@@ -47,12 +66,16 @@ Core generation pipeline. Single prompt input with full model concept support.
- Randomized latent noise for variations
- Prompt encoder with selectable custom clip model, long-clip mode with custom models, advanced encoding, injectable internal styles, last-layer options
- Sampler with `variation extender` and `Align Your Steps` features
-- A1111 style network injection supported by text prompt (LoRA, LyCoris, Hypernetwork, Embedding)
+- A1111 style network injection supported by text prompt `(LoRA, LyCoris, Hypernetwork, Embedding)`
+- Network tag cleaner: Auto-removes incompatible network tags per model architecture
- Automatized and manual image saver with optional **preview saver** for checkpoint selectors and saved .csv prompts
- Aesthetic scorer for final image quality assessment
- Upscaler (selectable Ultimate SD and hiresFix)
- Dynamic prompt support
-- Auto clean incompatible network tags from prompt by model architecture
+
+**by Central model orchestrator:**
+- Two-stage refiner support (quality refinement stage)
+- Speed LoRAs: LCM, Lightning, Hyper for fast generation
@@ -64,7 +87,7 @@ Core generation pipeline. Single prompt input with full model concept support.
Professional prompt development workflow. Extended prompt management for testing and iteration.
-**Same model support as Minimal workflow.**
+**Same central orchestration and model support as Minimal workflow.**
### Basic workflow features:
@@ -83,7 +106,7 @@ Professional prompt development workflow. Extended prompt management for testing
Full production pipeline with styling, refinement, and selective output.
-**Same model support as Minimal workflow.**
+**Same central orchestration and model support as Minimal workflow.**
### Basic Production workflow features:
@@ -98,20 +121,24 @@ Full production pipeline with styling, refinement, and selective output.
## Workflow Comparison
-| Feature | Minimal | Basic | Basic Production |
-|---------|-----------------|-------|------------------|
-| **Primary Use** | Core generation | Prompt development | Production pipeline |
-| **Prompt Inputs** | 4 | 12 | 19 |
-| **Prompt Selector** | 1-click | 1-click | 1-click |
-| **Model Concepts** | 25+ | 25+ | 25+ |
-| **Model Keyword Insertion** | ✓ | ✓ | ✓ |
-| **CSV/TOML Readers** | ✓ | ✓ | ✓ |
-| **Network Injection** | ✓ | ✓ | ✓ |
-| **Variation Extender** | ✓ | ✓ | ✓ |
-| **Image Saver** | ✓ | ✓ | ✓ |
-| **Aesthetic Scorer** | ✓ | ✓ | ✓ |
-| **Style Block** | — | — | ✓ |
-| **Refiner Blocks** | — | — | ✓ (Face, Eye, Mouth, Hands) |
-| **Selective Saver** | — | — | ✓ |
+| Feature | Minimal | Basic | Basic Production |
+|----------------------------------------|-----------------|-------|------------------|
+| **Primary Use** | Core generation | Prompt development | Production pipeline |
+| **Prompt Inputs** | 4 | 12 | 19 |
+| **Prompt Selector** | 1-click | 1-click | 1-click |
+| **Model Concepts** | 25+ | 25+ | 25+ |
+| **Central Model Orchestrator** | ✓ | ✓ | ✓ |
+| **Model Keyword Insertion** | ✓ | ✓ | ✓ |
+| **CSV/TOML Readers** | ✓ | ✓ | ✓ |
+| **Network Injection** | ✓ | ✓ | ✓ |
+| **Network Tag Cleaner** | ✓ | ✓ | ✓ |
+| **Variation Extender** | ✓ | ✓ | ✓ |
+| **Image Saver** | ✓ | ✓ | ✓ |
+| **Aesthetic Scorer** | ✓ | ✓ | ✓ |
+| **Speed LoRAs** | ✓ | ✓ | ✓ |
+| **Two-Stage Refiner** | ✓ | ✓ | ✓ |
+| **Style Block** | — | — | ✓ |
+| **Refiner Detailers** | — | — | ✓ (Face, Eye, Mouth, Hands) |
+| **Selective result and preview saver** | ✓ | ✓ | ✓ |
diff --git a/Workflow/Manual/nodes/model_control.jpg b/Workflow/Manual/nodes/model_control.jpg
new file mode 100644
index 0000000..8638722
Binary files /dev/null and b/Workflow/Manual/nodes/model_control.jpg differ