Update app.py

app.py

@@ -5,157 +5,78 @@ import torch
 from diffusers import DiffusionPipeline
 from datasets import load_dataset
 
-# Configurações do dispositivo e modelo
-device = "cuda" if torch.cuda.is_available() else "cpu"
-model_repo_id = "stabilityai/sdxl-turbo"  # Substitua pelo modelo desejado
 
-torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
+# Configurações do dispositivo para uso apenas da CPU
+device = "cpu"
+model_repo_id = "stabilityai/sdxl-turbo"  # Continuando com o modelo especificado
 
-pipe = DiffusionPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
+# Carregar o pipeline configurado para CPU
+pipe = DiffusionPipeline.from_pretrained(model_repo_id)
 pipe = pipe.to(device)
 
-# Definições de parâmetros gerais
-MAX_SEED = np.iinfo(np.int32).max
-MAX_IMAGE_SIZE = 752
-
 # Carregando o dataset do Hugging Face
 dataset = load_dataset("LEIDIA/Data_Womleimg")
 
-# Adicionando descrições ao dataset
-descriptions = [
+
+# Parâmetros para carregar o dataset personalizado
+dataset_descriptions = [
     "A woman wearing a full blue leather catsuit",
-    "A woman in a black leather pants",
-    "A legs woman in tigh high blue leather boots",
-    "A woman in long red leather jacket, red leather shorts and a tigh high red leather boots",
+    "A woman in black leather pants",
+    "A legs woman in tight high blue leather boots",
+    "A woman in a long red leather jacket, red leather shorts, and tight high red leather boots",
     "A legs woman in cream color leather pants",
-    "A woman in purple leather leggings with tigh high black leather boots",
-    # Adicione mais descrições conforme necessário
+    "A woman in purple leather leggings with tight high black leather boots",
+    "A woman in black leather top and a long black leather skirt",
+    "A blonde woman with long curly hair wearing a yellow mini tight leather skirt",
+    "A thin Asian woman wearing a thigh-long black leather dress",
+    "Simple high brown leather boots",
+    "A beautiful brunette woman wearing leather clothes",
+    "A beautiful brunette woman in a sleeveless black dress seated at a bar holding a glass of champagne, with a cozy and elegant atmosphere in the background.",
+    "A curly blonde woman wearing a bold red leather jacket paired with black leather tight pants and red high-heeled leather boots, creating a modern and edgy vibe.",
+    "An ebony woman standing outdoors against a backdrop of rolling hills and a cloudy sky, wearing a striking outfit of a red leather shirt, black leather mini corset, red plaid skirt, and knee-high red lace-up leather boots.",
+    "A blonde curly woman wearing a fitted, shiny blue leather outfit including a jacket and pants with metallic buttons, paired with knee-high boots, in a neutral-colored room.",
+    "A girl in a black leather outfit with a heart-shaped cutout top, high-waisted leggings, and a purple cape, giving a superhero vibe.",
+    "A girl in a sleek black leather cropped top with a zip closure and high-waisted bottom, paired with long black gloves and pink hair styled in a ponytail, creating a bold and fashion-forward look.",
+    "A girl wearing a form-fitting black leather top, with long pink hair cascading down, creating a striking contrast in a neutral background."
 ]
 
-def infer(prompt, num_inference_steps):
-    """Função para gerar a imagem baseada no prompt."""
-    image = pipe(
-        prompt=prompt,
-        num_inference_steps=num_inference_steps,
-        height=MAX_IMAGE_SIZE,
-        width=MAX_IMAGE_SIZE,
-    ).images[0]
-    return image
-
-# Função completa para inferência com mais parâmetros
-
-def advanced_infer(
-    prompt,
-    negative_prompt,
-    seed,
-    randomize_seed,
-    width,
-    height,
-    guidance_scale,
-    num_inference_steps,
-):
-    if randomize_seed:
-        seed = random.randint(0, MAX_SEED)
-
-    generator = torch.Generator().manual_seed(seed)
+# Definir parâmetros padrão para geração rápida
+DEFAULT_PROMPT = "A beautiful brunette woman wearing a leather outfit"
+DEFAULT_INFERENCE_STEPS = 6
+IMAGE_WIDTH = 512
+IMAGE_HEIGHT = 512
+GUIDANCE_SCALE = 1.5
 
+# Função simples para gerar imagem
+def infer_simple(prompt):
     image = pipe(
         prompt=prompt,
-        negative_prompt=negative_prompt,
-        guidance_scale=guidance_scale,
-        num_inference_steps=num_inference_steps,
-        width=width,
-        height=height,
-        generator=generator,
+        num_inference_steps=DEFAULT_INFERENCE_STEPS,
+        guidance_scale=GUIDANCE_SCALE,
+        height=IMAGE_HEIGHT,
+        width=IMAGE_WIDTH,
     ).images[0]
-
-    return image, seed
+    return image
 
 # Interface Gradio
 with gr.Blocks() as demo:
-    with gr.Column(elem_id="col-container"):
-        gr.Markdown("## Text-to-Image Optimized for CPU")
-
     with gr.Row():
-        prompt = gr.Textbox(
-            label="Prompt",
-            show_label=False,
-            max_lines=1,
-            placeholder="Enter your prompt",
-            container=False,
-        )
+        gr.Markdown("## Text-to-Image Wom Test - Quick CPU Version")
 
-    # Slider para definir o número de passos de inferência
-    num_inference_steps = gr.Slider(
-        label="Number of inference steps",
-        minimum=1,
-        maximum=7,
-        step=1,
-        value=15,
+    prompt = gr.Textbox(
+        label="Prompt",
+        value=DEFAULT_PROMPT,
+        placeholder="Describe the image you want to generate",
     )
 
-    # Botão para gerar a imagem
     generate_button = gr.Button("Generate")
     result = gr.Image(label="Generated Image")
 
-    # Clique no botão para gerar a imagem
     generate_button.click(
-        infer,
-        inputs=[prompt, num_inference_steps],
+        fn=infer_simple,
+        inputs=prompt,
         outputs=result,
     )
 
-    # Configurações avançadas
-    with gr.Accordion("Advanced Settings", open=False):
-        negative_prompt = gr.Textbox(
-            label="Negative prompt",
-            max_lines=1,
-            placeholder="Enter a negative prompt",
-        )
-
-        seed = gr.Slider(
-            label="Seed",
-            minimum=0,
-            maximum=MAX_SEED,
-            step=1,
-            value=0,
-        )
-
-        randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
-
-        with gr.Row():
-            width = gr.Slider(
-                label="Width",
-                minimum=64,
-                maximum=MAX_IMAGE_SIZE,
-                step=8,
-                value=512,
-            )
-
-            height = gr.Slider(
-                label="Height",
-                minimum=64,
-                maximum=MAX_IMAGE_SIZE,
-                step=8,
-                value=512,
-            )
-
-        guidance_scale = gr.Slider(
-            label="Guidance scale",
-            minimum=0.0,
-            maximum=20.0,
-            step=0.5,
-            value=1.0,
-        )
-
-    # Exemplos de prompts
-    gr.Examples(
-        examples=[
-            "A woman wearing leather pants",
-            "A woman in a red leather jacket",
-        ],
-        inputs=[prompt],
-    )
-
 if __name__ == "__main__":
     demo.launch()