import gradio as gr import requests import time import requests import base64 import os token = os.getenv("runpod_key") ############################################################# ################################################# def SD_call(prompt, image_prompt, age, weight, w_w, color, hair_color,h_c_w, hair_length, h_l_w,hair_texture, h_t_w,skin_details, eye_colors, e_c_w,NSFW): positive = "clothes" negative = "(((naked))), (((nsfw))), porn" serverless_api_id = 'zcparyadhjnmg7' # Define the URL you want to send the request to url = f"https://api.runpod.ai/v2/{serverless_api_id}/runsync" # Define your custom headers headers = { "Authorization": f"Bearer {token}", "Accept": "application/json", "Content-Type": "application/json" } weight = f"{'(' * w_w}{weight} woman{')' * w_w}" hair_color=f"{'(' * h_c_w}{hair_color} hair{')' * h_c_w}" hair_length=f"{'(' * h_l_w}{hair_length} hair{')' * h_l_w}" hair_texture =f"{'(' * h_t_w}{hair_texture} hair{')' * h_t_w}" eye_colors=f"{'(' * e_c_w}{eye_colors} eyes{')' * e_c_w}" if NSFW == True: positive = "((naked)), ((nsfw))" negative = "((clothes))" if prompt.strip(): total_prompt = prompt else: color = ", ".join(color) skin_details = ", ".join(skin_details) total_prompt = f"masterpiece, best quality, 8k, (looking at viewer:1.1), gorgeous, hot, seductive, {age} years old american {color} woman, {weight}, (eye contact:1.1), beautiful face, hyper detailed, best quality, ultra high res, {hair_length}, {hair_color}, {hair_texture},{eye_colors}, {skin_details}, photorealistic, high resolution, detailed, raw photo, 1girl,{image_prompt}, amateur cellphone photography. f8.0, samsung galaxy, noise, jpeg artefacts, poor lighting, low light, underexposed, high contrast " # Define your data (this could also be a JSON payload) print("SD_processing") data = { "input": { "api": { "method": "POST", "endpoint": "/sdapi/v1/txt2img" }, "payload": { "override_settings": { "sd_model_checkpoint": "cyberrealistic_classicV21", "sd_vae": "vae-ft-mse-840000-ema-pruned.safetensors" }, "override_settings_restore_afterwards": True, "refiner_checkpoint": "", "refiner_switch_at": 0.8, "prompt": f"{total_prompt}, {positive}", "negative_prompt": f"EasyNegative, fat, paintings, sketches, lowres, ((monochrome)), ((grayscale)), bad anatomy, text, error, cropped, signature, watermark, username, blurry, bad feet, poorly drawn face, bad proportions, gross proportions, ng_deepnegative_v1_75t, badhandsv5-neg, {negative}", "seed": -1, "batch_size": 1, "steps": 25, "cfg_scale": 7, "width": 520, "height": 520, "sampler_name": "DPM++ SDE Karras", "sampler_index": "DPM++ SDE Karras", "restore_faces": False } } } # data = { # "input": { # "prompt": f"{total_prompt}, {positive}", # "negative_prompt": f"paintings, sketches, lowres, ((monochrome)), ((grayscale)), bad anatomy, text, error, cropped, signature, watermark, username, blurry, bad feet, poorly drawn face, bad proportions, gross proportions, badhandsv5-neg, {negative}", # "width": 512, # "height": 720, # "guidance_scale": 7.5, # "num_inference_steps": 30, # "num_outputs": 1, # "prompt_strength": 0.8, # "scheduler": "K-LMS" # } # } # Send the POST request with headers and data response = requests.post(url, headers=headers, json=data) # Check the response if response.status_code == 200: response_data = response.json() msg_id = response_data['id'] print("Message ID:", msg_id) # Poll the status until it's not 'IN_QUEUE' while response_data['status'] == 'IN_QUEUE': time.sleep(5) # Wait for 5 seconds before checking again print("1") response = requests.get(f"{url}/{msg_id}", headers=headers) try: response_data = response.json() except Exception as e: print("Error decoding JSON:", e) print("Response content:", response.text) break # Exit the loop on JSON decoding error # Check if the response contains images if 'images' in response_data.get('output', {}): print("image") base64_image = response_data['output']['images'][0] image_bytes = base64.b64decode(base64_image) # Save the image to a file image_path = f"output_image_{msg_id}.png" with open(image_path, "wb") as img_file: img_file.write(image_bytes) print(f"Image downloaded successfully: {image_path}") return image_path else: return "No images found in the response." else: # Print error message return f"Error: {response.status_code} - {response.text}" def greet(prompt, image_prompt, age, weight, w_w, color, hair_color, h_c_w, hair_length, h_l_w,hair_texture, h_t_w,skin_details, eye_colors, e_c_w,NSFW): image_path = SD_call(prompt, image_prompt, age, weight, w_w, color, hair_color,h_c_w, hair_length, h_l_w,hair_texture, h_t_w,skin_details, eye_colors, e_c_w,NSFW) if image_path is not None: return image_path else: return None demo = gr.Interface( fn=greet, inputs=[ gr.Textbox(label="Personal prompt", lines=3), gr.Textbox(label="Girl_prompt", lines=3), gr.Slider(label="Age", value=22, minimum=18, maximum=75), gr.Radio(["skinny", "slim", "athletic", "muscular", "average", "curvy", "chubby", "overweight", "obese"],label="Body Type",type="value"), gr.Slider(label="Body Type weight", value=2, minimum=1, maximum=4, step=1), gr.CheckboxGroup(choices=["asian", "white", "black", "latina", "middle eastern","indigenous", "Mixed"],label="Color",type="value"), gr.Radio(["black", "brown", "brunette", "dark brown", "light brown", "blonde", "dirty blonde", "platinum blonde", "red", "auburn", "ginger", "strawberry blonde", "gray", "silver", "white", "blue", "green", "purple", "pink", "rainbow", "multicolored"],label="Hair Color",type="value"), gr.Slider(label="Hair Color weight", value=2, minimum=1, maximum=4, step=1), gr.Radio(["short", "long", "mi-long"],label="Hair length", type="value"), gr.Slider(label="Hair length weight", value=2, minimum=1, maximum=4, step=1), gr.Radio(["straight", "curvy", "wavy"],label="Hair texture", type="value"), gr.Slider(label="Hair texture weight", value=2, minimum=1, maximum=4, step=1), gr.CheckboxGroup(choices=["((tattoos))", "((birthmark))", "freckles", "((scars))"],label="Skin details", type="value"), gr.Radio(["brown", "hazel", "green", "blue", "gray", "amber", "black", "red", "violet"],label="Eyes Color", type="value"), gr.Slider(label="Eyes color weight", value=2, minimum=1, maximum=4, step=1), gr.Checkbox(label="NSFW", info="👀👀👀") ], flagging_options=["blurry", "incorrect", "other"], outputs=[gr.Image(label="Generated Image", type="filepath")], ) demo.launch(share=True)