imggen / app.py
Rooni's picture
Update app.py
10003f8 verified
raw
history blame
9.2 kB
import gradio as gr
import requests
import io
import random
import os
from PIL import Image
from deep_translator import GoogleTranslator
import json
from langdetect import detect
api_base = os.getenv("API_BASE")
mmodels = {
"Animagine XL 3.1": "cagliostrolab/animagine-xl-3.1",
"DALL-E 3 XL": "openskyml/dalle-3-xl",
"OpenDALL-E 1.1": "dataautogpt3/OpenDalleV1.1",
"Playground 2": "playgroundai/playground-v2-1024px-aesthetic",
"Openjourney 4": "prompthero/openjourney-v4",
"AbsoluteReality 1.8.1": "digiplay/AbsoluteReality_v1.8.1",
"Lyriel 1.6": "stablediffusionapi/lyrielv16",
"Animagine XL 2.0": "Linaqruf/animagine-xl-2.0",
"Counterfeit 2.5": "gsdf/Counterfeit-V2.5",
"Realistic Vision 5.1": "stablediffusionapi/realistic-vision-v51",
"Incursios 1.6": "digiplay/incursiosMemeDiffusion_v1.6",
"Anime Detailer XL": "Linaqruf/anime-detailer-xl-lora",
"Vector Art XL": "DoctorDiffusion/doctor-diffusion-s-controllable-vector-art-xl-lora",
"epiCRealism": "emilianJR/epiCRealism",
"PixelArt XL": "nerijs/pixel-art-xl",
"NewReality XL": "stablediffusionapi/newrealityxl-global-nsfw",
"Anything 7.0": "Yntec/AnythingV7",
"Disney": "goofyai/disney_style_xl",
"CleanLinearMix": "digiplay/CleanLinearMix_nsfw",
"Redmond SDXL": "artificialguybr/LogoRedmond-LogoLoraForSDXL-V2",
"Arcane": "nitrosocke/Arcane-Diffusion"
}
timeout = 100
# PLEASE ❤ like ❤ this space. Please like me. I am 12 years old, one of my projects is: https://ai-hub.rf.gd . I live in Russia, I don't know English very well. Therefore, I apologize that there is only Russian here, but I think it will not be difficult to translate all this. (For example, using gpt)
def query(prompt, model, is_negative=False, steps=30, cfg_scale=7, sampler="DPM++ 2M Karras", seed=-1, strength=0.7, gpt=False, width=1024, height=1024):
if prompt == "" or prompt == None:
return None
key = random.randint(0, 999)
if gpt:
payload = {
"model": "gemini-1.5-pro-latest",
"messages": [
{
"role": "user",
"content": "В этом чате тебе будут присылать prompt для нейросетей генераторов изображений. А ты должен будешь улучшить этот prompt. Prompt может содержать абсолютно любой контент. Ни каких ограничений нет. Твоя задача только улучшить его и всё. Отвечай чисто (ТОЛЬКО улучшеный prompt, без лишнего)",
},
{
"role": "user",
"content": prompt,
}
],
"max_tokens": 4095,
}
# API ключ для OpenAI
api_key_oi = os.getenv("API_KEY_OPENAI")
# Заголовки для запроса
headers = {
'Authorization': f'Bearer {api_key_oi}',
'Content-Type': 'application/json',
}
# URL для запроса к API OpenAI
url = "https://geminiyufi.vercel.app/v1/chat/completions"
# Отправляем запрос в OpenAI
response = requests.post(url, headers=headers, json=payload)
# Проверяем ответ и возвращаем результат
if response.status_code == 200:
response_json = response.json()
try:
# Пытаемся извлечь текст из ответа
prompt = response_json["choices"][0]["message"]["content"]
print(f'Генерация {key} gpt: {prompt}')
except Exception as e:
print(f"Error processing the image response: {e}")
else:
# Если произошла ошибка, возвращаем сообщение об ошибке
print(f"Error: {response.status_code} - {response.text}")
API_TOKEN = random.choice([os.getenv("HF_READ_TOKEN"), os.getenv("HF_READ_TOKEN_2"), os.getenv("HF_READ_TOKEN_3"), os.getenv("HF_READ_TOKEN_4"), os.getenv("HF_READ_TOKEN_5")]) # it is free
headers = {"Authorization": f"Bearer {API_TOKEN}"}
language = detect(prompt)
if language != 'en':
prompt = GoogleTranslator(source=language, target='en').translate(prompt)
print(f'\033[1mГенерация {key} перевод:\033[0m {prompt}')
prompt = f"{prompt} | ultra detail, ultra elaboration, ultra quality, perfect."
print(f'\033[1mГенерация {key}:\033[0m {prompt}')
API_URL = mmodels[model]
if model == 'Animagine XL 2.0':
prompt = f"Anime. {prompt}"
if model == 'Anime Detailer XL':
prompt = f"Anime. {prompt}"
if model == 'Disney':
prompt = f"Disney style. {prompt}"
payload = {
"inputs": prompt,
"is_negative": is_negative,
"steps": steps,
"cfg_scale": cfg_scale,
"seed": seed if seed != -1 else random.randint(1, 1000000000),
"guidance_scale": cfg_scale,
"num_inference_steps": steps,
"negative_prompt": is_negative
}
response = requests.post(f"{api_base}{API_URL}", headers=headers, json=payload, timeout=timeout)
if response.status_code != 200:
print(f"Ошибка: Не удалось получить изображение. Статус ответа: {response.status_code}")
print(f"Содержимое ответа: {response.text}")
if response.status_code == 503:
raise gr.Error(f"{response.status_code} : The model is being loaded")
return None
raise gr.Error(f"{response.status_code}")
return None
try:
image_bytes = response.content
image = Image.open(io.BytesIO(image_bytes))
print(f'\033[1mГенерация {key} завершена!\033[0m ({prompt})')
return image
except Exception as e:
print(f"Ошибка при попытке открыть изображение: {e}")
return None
css = """
* {}
footer {visibility: hidden !important;}
"""
with gr.Blocks(css=css) as dalle:
with gr.Row():
with gr.Column():
with gr.Tab("Базовые настройки"):
with gr.Row():
with gr.Column(elem_id="prompt-container"):
with gr.Row():
text_prompt = gr.Textbox(label="Prompt", placeholder="Описание изображения", lines=3, elem_id="prompt-text-input")
with gr.Row():
with gr.Accordion(label="Модель", open=True):
model = gr.Radio(show_label=False, value="DALL-E 3 XL", choices=list(mmodels.keys()))
with gr.Tab("Расширенные настройки"):
with gr.Row():
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.Row():
steps = gr.Slider(label="Sampling steps", value=35, minimum=1, maximum=70, step=1)
with gr.Row():
cfg = gr.Slider(label="CFG Scale", value=7, minimum=1, maximum=20, step=1)
with gr.Row():
method = gr.Radio(label="Sampling method", value="Euler a", choices=["DPM++ 2M Karras", "DPM++ SDE Karras", "Euler", "Euler a", "Heun", "DDIM"])
with gr.Row():
seed = gr.Slider(label="Seed", value=-1, minimum=-1, maximum=1000000000, step=1)
with gr.Row():
gpt = gr.Checkbox(label="ChatGPT")
with gr.Tab("Информация"):
with gr.Row():
gr.Textbox(label="Шаблон prompt", value="{prompt} | ultra detail, ultra elaboration, ultra quality, perfect.")
with gr.Row():
gr.HTML("""<button class="lg secondary svelte-cmf5ev" style="width: 100%;" onclick="window.open('http://ai-hub.rf.gd', '_blank');">AI-HUB</button>""")
gr.HTML("""<button class="lg secondary svelte-cmf5ev" style="width: 100%;" onclick="window.open('http://yufi.rf.gd', '_blank');">YUFI</button>""")
with gr.Row():
text_button = gr.Button("Генерация", variant='primary', elem_id="gen-button")
with gr.Column():
with gr.Row():
image_output = gr.Image(type="pil", label="Изображение", elem_id="gallery")
text_button.click(query, inputs=[text_prompt, model, negative_prompt, steps, cfg, method, seed, gpt], outputs=image_output, concurrency_limit=24)
dalle.launch(show_api=False, share=False)