0908 live stream version (modelscope#199)

* 0908_live_stream_version, High performance inpainting for single & double person, Simplify User Interface. Co-authored-by: 诚公 <[email protected]>
xlan-codes · Sep 9, 2023 · 1105d46 · 1105d46
1 parent 44669b3
commit 1105d46
Show file tree

Hide file tree

Showing 14 changed files with 1,092 additions and 26 deletions.
diff --git a/README.md b/README.md
@@ -20,10 +20,12 @@ FaceChain is powered by [ModelScope](https://github.com/modelscope/modelscope).
 
 
 ![image](resources/git_cover.jpg)
-![image](resources/git_cover_1.png)
+![image](resources/git_cover_1.jpg)
+![image](resources/git_cover_2.jpg)
 
 
 # News
+- High performance inpainting for single & double person, Simplify User Interface. (September 09th, 2023 UTC)
 - More Technology Details can be seen in [Paper](https://arxiv.org/abs/2308.14256). (August 30th, 2023 UTC)
 - Add validate & ensemble for Lora training, and InpaintTab(hide in gradio for now).  (August 28th, 2023 UTC)
 - Add pose control module.   (August 27th, 2023 UTC)

diff --git a/README_ZH.md b/README_ZH.md
@@ -18,10 +18,12 @@ FaceChain的模型由[ModelScope](https://github.com/modelscope/modelscope)开
 <br>
 
 ![image](resources/git_cover_CH.jpg)
-![image](resources/git_cover_1.png)
+![image](resources/git_cover_1.jpg)
+![image](resources/git_cover_2.jpg)
 
 
 # News
+- 高性能的(单人&双人)模版重绘功能，简化用户界面. (2023-09-09)
 - 更多技术细节可以在 [论文](https://arxiv.org/abs/2308.14256) 里查看. (2023-08-30)
 - 为Lora训练添加验证和根据face_id的融合，并添加InpaintTab（目前在Gradio界面上暂时默认隐藏）. (2023-08-28)
 - 增加姿势控制模块，可一键体验模版pose复刻. (2023-08-27)

diff --git a/app.py b/app.py
diff --git a/facechain/inference.py b/facechain/inference.py
@@ -506,8 +506,11 @@ def compress_image(input_path, target_size):
     image = cv2.imread(input_path)
 
     quality = 95
-    while cv2.imencode('.jpg', image, [cv2.IMWRITE_JPEG_QUALITY, quality])[1].size > target_size:
-        quality -= 5
+    try:
+        while cv2.imencode('.jpg', image, [cv2.IMWRITE_JPEG_QUALITY, quality])[1].size > target_size:
+            quality -= 5
+    except:
+        import pdb;pdb.set_trace()
 
     compressed_image = cv2.imencode('.jpg', image, [cv2.IMWRITE_JPEG_QUALITY, quality])[1].tostring()
 

diff --git a/facechain/inference_ipt.py b/facechain/inference_ipt.py
diff --git a/resources/git_cover_1.jpg b/resources/git_cover_1.jpg
diff --git a/resources/git_cover_1.png b/resources/git_cover_1.png
diff --git a/resources/git_cover_2.jpg b/resources/git_cover_2.jpg
diff --git a/resources/inpaint_template/0.jpg b/resources/inpaint_template/0.jpg
diff --git a/resources/inpaint_template/1.jpg b/resources/inpaint_template/1.jpg
diff --git a/resources/inpaint_template/2.jpg b/resources/inpaint_template/2.jpg
diff --git a/resources/inpaint_template/3.jpg b/resources/inpaint_template/3.jpg
diff --git a/resources/inpaint_template/4.jpg b/resources/inpaint_template/4.jpg
diff --git a/resources/inpaint_template/5.jpg b/resources/inpaint_template/5.jpg