docs: remove control_context_scale parameter and refine model features

README.md

@@ -41,11 +41,9 @@ This repository hosts the **ConvRot INT8** quantized weights of **`Qwen-Image-25
 
 ## 🛠️ Model Features
 
-- ControlNet added on 5 layer blocks of Qwen-Image-2512.
-- Supports multiple control conditions: Canny, HED, Depth, Pose, MLSD, Scribble, Gray.
-- Inpainting mode is supported.
-- Acquiring control images in a multi-resolution manner results in better generalization.
-- Optimal range for `control_context_scale` is from 0.70 to 0.95.
+- Multi-condition ControlNet Union model (added on 5 layer blocks of Qwen-Image-2512).
+- Supported control conditions: Canny, HED, Depth, Pose, MLSD, Scribble, Gray, and Inpainting.
+- Quantized to **ConvRot INT8** to reduce VRAM and disk footprint while maintaining structural control fidelity.
 
 ---