Upload folder using huggingface_hub

.gitattributes

@@ -39,3 +39,4 @@ preview/result2.png filter=lfs diff=lfs merge=lfs -text
 preview/result3.png filter=lfs diff=lfs merge=lfs -text
 preview/result4.png filter=lfs diff=lfs merge=lfs -text
 preview/result5.png filter=lfs diff=lfs merge=lfs -text
+preview/input.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -14,19 +14,6 @@ tags:
 - DiT
 - Qwen-Image
 - ValiantCat
-widget:
-- text: >-
-    improve the composition and visual consistency of the image while maintaining style and realism.
-  output:
-    url: preview/sample1.png
-- text: >-
-    enhance aesthetic appeal and global color harmony of the photo.
-  output:
-    url: preview/sample2.png
-- text: >-
-    recompose the scene to improve perspective consistency and balance.
-  output:
-    url: preview/sample3.png
 ---
 
 <p align="center">
@@ -35,9 +22,9 @@ widget:
 
 ---
 
-# 🌈 starsfriday Qwen-Image-Edit-DiT-MeiTu
+# 🌈 starsfriday Qwen-Image-Edit-MeiTu
 
-This model — **Qwen-Image-Edit-DiT-MeiTu** — is an improved variant of [Qwen/Qwen-Image-Edit](https://huggingface.co/starsfriday/Qwen-Image-Edit-MeiTu), built with **DiT-based architecture fine-tuning** to enhance **visual consistency**, **aesthetic quality**, and **structural alignment** in complex edits.
+This model — **Qwen-Image-Edit-MeiTu** — is an improved variant of [Qwen/Qwen-Image-Edit](https://huggingface.co/starsfriday/Qwen-Image-Edit-MeiTu), built with **DiT-based architecture fine-tuning** to enhance **visual consistency**, **aesthetic quality**, and **structural alignment** in complex edits.
 
 Developed by **Valiant Cat AI Lab**, this version aims to further close the gap between high-fidelity semantic editing and coherent artistic rendering, achieving a more natural and professional output across a wide range of prompts and subjects.
 
@@ -75,7 +62,7 @@ from PIL import Image
 from diffusers import QwenImageEditPipeline
 
 # Load the enhanced pipeline
-pipeline = QwenImageEditPipeline.from_pretrained("starsfriday/Qwen-Image-Edit-DiT-Enhanced")
+pipeline = QwenImageEditPipeline.from_pretrained("starsfriday/Qwen-Image-Edit-MeiTu")
 pipeline.to(torch.bfloat16)
 pipeline.to("cuda")
 
@@ -104,11 +91,11 @@ with torch.inference_mode():
 
 Below are examples of **consistency and aesthetic improvement** in complex editing scenarios:
 
-| Task | Before | After |
-|------|---------|-------|
-| **Portrait lighting enhancement** | ![](result/sample1.png) | ![](result/sample1_out.png) |
-| **Scene recomposition with better perspective** | ![](result/sample2.png) | ![](result/sample2_out.png) |
-| **Global color harmony & fine detail restoration** | ![](result/sample3.png) | ![](result/sample3_out.png) |
+| Task | input & output
+|------|---------|
+| **Portrait lighting enhancement** | ![](preview/sample1.png)
+| **Scene recomposition with better perspective** | ![](preview/sample2.png)|
+| **Global color harmony & fine detail restoration** | ![](preview/sample3.png)
 
 ---