Update app.py

app.py

@@ -7,6 +7,38 @@ from PIL import Image
 from deep_translator import GoogleTranslator
 from langdetect import detect
 
+import cv2
+import torch
+from basicsr.archs.srvgg_arch import SRVGGNetCompact
+from gfpgan.utils import GFPGANer
+from realesrgan.utils import RealESRGANer
+
+os.system("pip freeze")
+# download weights
+if not os.path.exists('realesr-general-x4v3.pth'):
+    os.system("wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-general-x4v3.pth -P .")
+if not os.path.exists('GFPGANv1.2.pth'):
+    os.system("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.2.pth -P .")
+if not os.path.exists('GFPGANv1.3.pth'):
+    os.system("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.3.pth -P .")
+if not os.path.exists('GFPGANv1.4.pth'):
+    os.system("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.4.pth -P .")
+if not os.path.exists('RestoreFormer.pth'):
+    os.system("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/RestoreFormer.pth -P .")
+if not os.path.exists('CodeFormer.pth'):
+    os.system("wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/CodeFormer.pth -P .")
+
+
+# background enhancer with RealESRGAN
+model_us = SRVGGNetCompact(num_in_ch=3, num_out_ch=3, num_feat=64, num_conv=32, upscale=4, act_type='prelu')
+model_us_path = 'realesr-general-x4v3.pth'
+half = True if torch.cuda.is_available() else False
+upsampler = RealESRGANer(scale=4, model_us_path=model_us_path, model_us=model_us, tile=0, tile_pad=10, pre_pad=0, half=half)
+
+os.makedirs('output', exist_ok=True)
+
+
+
 API_URL = "https://api-inference.huggingface.co/models/openskyml/dalle-3-xl"
 API_TOKEN = os.getenv("HF_READ_TOKEN") # it is free
 headers = {"Authorization": f"Bearer {API_TOKEN}"}
@@ -60,6 +92,64 @@ def query(prompt, model, is_negative=False, steps=20, cfg_scale=7, seed=None):
     return image
 
 
+
+def inference(img, version, scale, weight):
+    weight /= 100
+    print(img, version, scale, weight)
+    try:
+        extension = os.path.splitext(os.path.basename(str(img)))[1]
+        img = cv2.imread(img, cv2.IMREAD_UNCHANGED)
+        if len(img.shape) == 3 and img.shape[2] == 4:
+            img_mode = 'RGBA'
+        elif len(img.shape) == 2:  # for gray inputs
+            img_mode = None
+            img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
+        else:
+            img_mode = None
+
+        if version == 'v1.2':
+            face_enhancer = GFPGANer(
+            model_us_path='GFPGANv1.2.pth', upscale=2, arch='clean', channel_multiplier=2, bg_upsampler=upsampler)
+        elif version == 'v1.3':
+            face_enhancer = GFPGANer(
+            model_us_path='GFPGANv1.3.pth', upscale=2, arch='clean', channel_multiplier=2, bg_upsampler=upsampler)
+        elif version == 'v1.4':
+            face_enhancer = GFPGANer(
+            model_us_path='GFPGANv1.4.pth', upscale=2, arch='clean', channel_multiplier=2, bg_upsampler=upsampler)
+        elif version == 'RestoreFormer':
+            face_enhancer = GFPGANer(
+            model_us_path='RestoreFormer.pth', upscale=2, arch='RestoreFormer', channel_multiplier=2, bg_upsampler=upsampler)
+        elif version == 'CodeFormer':
+            face_enhancer = GFPGANer(
+            model_us_path='CodeFormer.pth', upscale=2, arch='CodeFormer', channel_multiplier=2, bg_upsampler=upsampler)
+
+        try:
+            _, _, output = face_enhancer.enhance(img, has_aligned=False, only_center_face=False, paste_back=True, weight=weight)
+        except RuntimeError as error:
+            print('Error', error)
+
+        try:
+            interpolation = cv2.INTER_AREA if scale < 2 else cv2.INTER_LANCZOS4
+            h, w = img.shape[0:2]
+            output = cv2.resize(output, (int(w * scale), int(h * scale)), interpolation=interpolation)
+        except Exception as error:
+            print('wrong scale input.', error)
+        if img_mode == 'RGBA':  # RGBA images should be saved in png format
+            extension = 'png'
+        else:
+            extension = 'jpg'
+        save_path = f'output/out.{extension}'
+        cv2.imwrite(save_path, output)
+
+        output = cv2.cvtColor(output, cv2.COLOR_BGR2RGB)
+        return output
+    except Exception as error:
+        print('global exception', error)
+        return None
+
+
+
+
 css = """
 footer {visibility: hidden !important;}
 """
@@ -75,11 +165,21 @@ with gr.Blocks(css=css) as dalle:
     with gr.Tab("Расширенные настройки"):
         negative_prompt = gr.Textbox(label="Negative Prompt", placeholder="Чего не должно быть на изображении", value="[deformed | disfigured], poorly drawn, [bad : wrong] anatomy, [extra | missing | floating | disconnected] limb, (mutated hands and fingers), blurry, text, fuzziness", lines=3, elem_id="negative-prompt-text-input")
 
+    with gr.Tab("Настройки апскейлинга"):
+        up_1 = gr.Radio(choices=['v1.2', 'v1.3', 'v1.4', 'RestoreFormer', 'CodeFormer'], value='v1.4', label='Версия'),
+        up_2 = gr.Slider(label="Коэффициент масштабирования", value=2, minimum=2, maximum=6),
+        up_3 = gr.Slider(0, 100, label='Weight, только для CodeFormer. 0 для лучшего качества, 100 для лучшей идентичности', value=50)
+    
     with gr.Row():
         text_button = gr.Button("Генерация", variant='primary', elem_id="gen-button")
     with gr.Row():
         image_output = gr.Image(type="pil", label="Изображение", elem_id="gallery")
+    with gr.Row():
+        up_button = gr.Button("Улучшить изображение", variant='primary', elem_id="gen-button")
+    with gr.Row():
+        up_output = gr.Image(type="pil", label="Улучшенное изображение", elem_id="gallery"),
         
     text_button.click(query, inputs=[text_prompt, model, negative_prompt], outputs=image_output)
+    up_button.click(query, inputs=[image_output, up_1, up_2, up_3], outputs=up_output)
 
 dalle.launch(show_api=False)