Update README for unified interface

README.md

@@ -15,24 +15,18 @@ startup_duration_timeout: 30m
 
 Efficient Native-Resolution Foundation Model for Image Generation and Editing.
 
-This Space demonstrates [Mage-Flow](https://huggingface.co/papers/2607.19064) from Microsoft, a 4B-scale image generation and editing foundation model. It supports:
+This Space demonstrates [Mage-Flow](https://huggingface.co/papers/2607.19064) from Microsoft, a 4B-scale image generation and editing foundation model through a single unified interface:
 
-- **Text → Image**: Generate images from text prompts at arbitrary resolutions.
-- **Image Edit**: Instruction-based editing of reference images.
+- **No image** → the prompt generates a new image (routes to [Mage-Flow-Turbo](https://huggingface.co/microsoft/Mage-Flow-Turbo)).
+- **With an uploaded image** → the prompt is treated as an edit instruction (routes to [Mage-Flow-Edit-Turbo](https://huggingface.co/microsoft/Mage-Flow-Edit-Turbo)).
 
-The demo uses the Turbo variants ([Mage-Flow-Turbo](https://huggingface.co/microsoft/Mage-Flow-Turbo) and [Mage-Flow-Edit-Turbo](https://huggingface.co/microsoft/Mage-Flow-Edit-Turbo)) for fast, few-step inference (4 steps, CFG 1.0).
+The demo uses the Turbo variants for fast, few-step inference (4 steps, CFG 1.0).
 
 ## Usage
 
-### Text → Image
-1. Enter a descriptive prompt.
-2. Adjust steps (Turbo default: 4) and CFG (Turbo default: 1.0).
-3. Click **Generate**.
-
-### Image Edit
-1. Upload a reference image.
-2. Enter an edit instruction (e.g., "change the background to a city street").
-3. Click **Edit**.
+1. Enter a prompt.
+2. *(Optional)* Upload an image to edit — the prompt becomes an edit instruction.
+3. Click **Run**.
 
 ## References