test-100 / app.py
Kims12's picture
Update app.py
fd7e6af verified
raw
history blame
13.6 kB
import json
import os
import time
import tempfile
from PIL import Image
import gradio as gr
import logging
from io import BytesIO
from google import genai
from google.genai import types
# .env ํŒŒ์ผ์— ์ €์žฅ๋œ ํ™˜๊ฒฝ๋ณ€์ˆ˜ ๋กœ๋“œ (python-dotenv ์„ค์น˜ ํ•„์š”: pip install python-dotenv)
from dotenv import load_dotenv
load_dotenv()
# ๋กœ๊น… ์„ค์ • (๋กœ๊ทธ ๋ ˆ๋ฒจ: DEBUG)
logging.basicConfig(level=logging.DEBUG,
format='%(asctime)s - %(levelname)s - %(message)s')
logger = logging.getLogger(__name__)
def save_binary_file(file_name, data):
logger.debug(f"ํŒŒ์ผ์— ์ด์ง„ ๋ฐ์ดํ„ฐ ์ €์žฅ ์ค‘: {file_name}")
with open(file_name, "wb") as f:
f.write(data)
logger.debug(f"ํŒŒ์ผ ์ €์žฅ ์™„๋ฃŒ: {file_name}")
def merge_images(person_img_path, product_img_path, background_img_path, prompt, model="gemini-2.0-flash-exp-image-generation"):
logger.debug(f"merge_images ํ•จ์ˆ˜ ์‹œ์ž‘ - ํ”„๋กฌํ”„ํŠธ: '{prompt}'")
try:
# API ํ‚ค๋Š” ํ™˜๊ฒฝ๋ณ€์ˆ˜์—์„œ ๋ถˆ๋Ÿฌ์˜ด
effective_api_key = os.environ.get("GEMINI_API_KEY")
if effective_api_key:
logger.debug("ํ™˜๊ฒฝ๋ณ€์ˆ˜์—์„œ API ํ‚ค ๋ถˆ๋Ÿฌ์˜ด")
else:
logger.error("API ํ‚ค๊ฐ€ ํ™˜๊ฒฝ๋ณ€์ˆ˜์— ์„ค์ •๋˜์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค.")
raise ValueError("API ํ‚ค๊ฐ€ ํ•„์š”ํ•ฉ๋‹ˆ๋‹ค.")
client = genai.Client(api_key=effective_api_key)
logger.debug("Gemini ํด๋ผ์ด์–ธํŠธ ์ดˆ๊ธฐํ™” ์™„๋ฃŒ.")
# PIL ์ด๋ฏธ์ง€ ๊ฐ์ฒด๋กœ ๋ณ€ํ™˜
person_img = Image.open(person_img_path)
product_img = Image.open(product_img_path)
# ์ปจํ…์ธ  ๋ฆฌ์ŠคํŠธ ์ƒ์„ฑ
contents = []
# ์‚ฌ๋žŒ ์ด๋ฏธ์ง€ ์ถ”๊ฐ€
contents.append(
types.Part.from_data(data=person_img, mime_type="image/jpeg")
)
# ์ƒํ’ˆ ์ด๋ฏธ์ง€ ์ถ”๊ฐ€
contents.append(
types.Part.from_data(data=product_img, mime_type="image/jpeg")
)
# ๋ฐฐ๊ฒฝ ์ด๋ฏธ์ง€๊ฐ€ ์žˆ์œผ๋ฉด ์ถ”๊ฐ€
if background_img_path:
background_img = Image.open(background_img_path)
contents.append(
types.Part.from_data(data=background_img, mime_type="image/jpeg")
)
logger.debug("๋ฐฐ๊ฒฝ ์ด๋ฏธ์ง€ ์ถ”๊ฐ€๋จ")
# ๋งˆ์ง€๋ง‰์— ํ”„๋กฌํ”„ํŠธ ์ถ”๊ฐ€
contents.append(
types.Part.from_text(text=prompt)
)
logger.debug(f"์ปจํ…์ธ  ๊ฐ์ฒด ์ƒ์„ฑ ์™„๋ฃŒ: {len(contents)} ์•„์ดํ…œ")
# ์ƒ์„ฑ ์„ค์ •
generate_content_config = types.GenerateContentConfig(
temperature=1,
top_p=0.95,
top_k=40,
max_output_tokens=8192,
response_modalities=["image", "text"],
)
logger.debug(f"์ƒ์„ฑ ์„ค์ •: {generate_content_config}")
with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmp:
temp_path = tmp.name
logger.debug(f"์ž„์‹œ ํŒŒ์ผ ์ƒ์„ฑ๋จ: {temp_path}")
# ์ŠคํŠธ๋ฆฌ๋ฐ ๋ฐฉ์‹์œผ๋กœ ๋ฐ์ดํ„ฐ ์ˆ˜์‹  (์›๋ž˜ ์ฝ”๋“œ ๋ฐฉ์‹์œผ๋กœ ๋ณต์›)
response_stream = client.models.generate_content_stream(
model=model,
contents=[
types.Content(
role="user",
parts=contents,
),
],
config=generate_content_config,
)
logger.debug("์‘๋‹ต ์ŠคํŠธ๋ฆผ ์ฒ˜๋ฆฌ ์‹œ์ž‘...")
# ์‘๋‹ต์—์„œ ์ด๋ฏธ์ง€์™€ ํ…์ŠคํŠธ ์ถ”์ถœ
image_saved = False
response_text = ""
for chunk in response_stream:
logger.debug(f"chunk ์ˆ˜์‹ : {chunk}")
# ์‘๋‹ต ๊ฒ€์ฆ
if not hasattr(chunk, 'candidates') or not chunk.candidates:
logger.warning("chunk์— candidates๊ฐ€ ์—†์Šต๋‹ˆ๋‹ค. ๊ฑด๋„ˆ๋œ๋‹ˆ๋‹ค.")
continue
if not hasattr(chunk.candidates[0], 'content') or not chunk.candidates[0].content:
logger.warning("chunk.candidates[0]์— content๊ฐ€ ์—†์Šต๋‹ˆ๋‹ค. ๊ฑด๋„ˆ๋œ๋‹ˆ๋‹ค.")
continue
if not hasattr(chunk.candidates[0].content, 'parts') or not chunk.candidates[0].content.parts:
logger.warning("chunk.candidates[0].content์— parts๊ฐ€ ์—†์Šต๋‹ˆ๋‹ค. ๊ฑด๋„ˆ๋œ๋‹ˆ๋‹ค.")
continue
for part in chunk.candidates[0].content.parts:
if hasattr(part, 'text') and part.text:
response_text += part.text
logger.info(f"์ˆ˜์‹ ๋œ ํ…์ŠคํŠธ: {part.text}")
elif hasattr(part, 'inline_data') and part.inline_data:
save_binary_file(temp_path, part.inline_data.data)
logger.info(f"MIME ํƒ€์ž… {part.inline_data.mime_type}์˜ ํŒŒ์ผ์ด ์ €์žฅ๋จ: {temp_path}")
image_saved = True
if not image_saved:
logger.warning("์ด๋ฏธ์ง€๊ฐ€ ์ƒ์„ฑ๋˜์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค.")
return None, response_text or "์ด๋ฏธ์ง€๊ฐ€ ์ƒ์„ฑ๋˜์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค. ๋‹ค๋ฅธ ํ”„๋กฌํ”„ํŠธ๋‚˜ ์ด๋ฏธ์ง€๋กœ ์‹œ๋„ํ•ด๋ณด์„ธ์š”."
logger.debug("์ด๋ฏธ์ง€ ์ƒ์„ฑ ์™„๋ฃŒ.")
return temp_path, response_text
except Exception as e:
logger.exception("์ด๋ฏธ์ง€ ์ƒ์„ฑ ์ค‘ ์˜ค๋ฅ˜ ๋ฐœ์ƒ:")
return None, f"์˜ค๋ฅ˜ ๋ฐœ์ƒ: {str(e)}" # ์˜ค๋ฅ˜ ๋ฐœ์ƒ ์‹œ None๊ณผ ์˜ค๋ฅ˜ ๋ฉ”์‹œ์ง€ ๋ฐ˜ํ™˜
def process_images_and_prompt(person_pil, product_pil, background_pil, prompt):
logger.debug(f"process_images_and_prompt ํ•จ์ˆ˜ ์‹œ์ž‘ - ํ”„๋กฌํ”„ํŠธ: '{prompt}'")
try:
# ํ•„์ˆ˜ ์ด๋ฏธ์ง€ ํ™•์ธ
if person_pil is None:
return [], "์‚ฌ๋žŒ ์ด๋ฏธ์ง€๋ฅผ ์—…๋กœ๋“œํ•ด์ฃผ์„ธ์š”."
if product_pil is None:
return [], "์ƒํ’ˆ ์ด๋ฏธ์ง€๋ฅผ ์—…๋กœ๋“œํ•ด์ฃผ์„ธ์š”."
# ๊ธฐ๋ณธ ํ”„๋กฌํ”„ํŠธ ์„ค์ • (๋น„์–ด์žˆ๋Š” ๊ฒฝ์šฐ)
if not prompt or not prompt.strip():
if background_pil:
prompt = "์ด ๋ฐฐ๊ฒฝ์— ์ด ์‚ฌ๋žŒ์ด ์ด ์ƒํ’ˆ์„ ์‚ฌ์šฉํ•˜๋Š” ๋ชจ์Šต์„ ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ๋ณด์—ฌ์ฃผ์„ธ์š”. ์ƒํ’ˆ์„ ์ž˜ ๋ณด์ด๊ฒŒ ํ•ด์ฃผ์„ธ์š”. Create a natural composite image showing this person using this product in this background setting. Make sure the product is clearly visible."
else:
prompt = "์ด ์‚ฌ๋žŒ์ด ์ด ์ƒํ’ˆ์„ ์‚ฌ์šฉํ•˜๋Š” ๋ชจ์Šต์„ ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ๋ณด์—ฌ์ฃผ์„ธ์š”. ์ƒํ’ˆ์„ ์ž˜ ๋ณด์ด๊ฒŒ ํ•ด์ฃผ์„ธ์š”. Create a natural composite image showing this person using this product. Make sure the product is clearly visible."
# ํ”„๋กฌํ”„ํŠธ์— ์˜์–ด๊ฐ€ ์—†์œผ๋ฉด ์˜์–ด ํ”„๋กฌํ”„ํŠธ ์ถ”๊ฐ€ (๋” ๋‚˜์€ ๊ฒฐ๊ณผ๋ฅผ ์œ„ํ•ด)
if not any(ord(c) < 128 for c in prompt):
if background_pil:
prompt += " Create a realistic composite image of this person with this product in this background."
else:
prompt += " Create a realistic composite image of this person with this product."
# ์ด๋ฏธ์ง€ ์ €์žฅ
with tempfile.NamedTemporaryFile(suffix=".jpg", delete=False) as tmp_person:
person_path = tmp_person.name
# RGB๋กœ ๋ณ€ํ™˜ํ•˜์—ฌ ์ €์žฅ
person_img = person_pil
if person_img.mode != 'RGB':
person_img = person_img.convert('RGB')
person_img.save(person_path, 'JPEG')
logger.debug(f"์‚ฌ๋žŒ ์ด๋ฏธ์ง€ ์ €์žฅ ์™„๋ฃŒ: {person_path}")
with tempfile.NamedTemporaryFile(suffix=".jpg", delete=False) as tmp_product:
product_path = tmp_product.name
# RGB๋กœ ๋ณ€ํ™˜ํ•˜์—ฌ ์ €์žฅ
product_img = product_pil
if product_img.mode != 'RGB':
product_img = product_img.convert('RGB')
product_img.save(product_path, 'JPEG')
logger.debug(f"์ƒํ’ˆ ์ด๋ฏธ์ง€ ์ €์žฅ ์™„๋ฃŒ: {product_path}")
# ๋ฐฐ๊ฒฝ ์ด๋ฏธ์ง€ ์ €์žฅ (์žˆ๋Š” ๊ฒฝ์šฐ)
background_path = None
if background_pil is not None:
with tempfile.NamedTemporaryFile(suffix=".jpg", delete=False) as tmp_bg:
background_path = tmp_bg.name
# RGB๋กœ ๋ณ€ํ™˜ํ•˜์—ฌ ์ €์žฅ
background_img = background_pil
if background_img.mode != 'RGB':
background_img = background_img.convert('RGB')
background_img.save(background_path, 'JPEG')
logger.debug(f"๋ฐฐ๊ฒฝ ์ด๋ฏธ์ง€ ์ €์žฅ ์™„๋ฃŒ: {background_path}")
# ์ด๋ฏธ์ง€ ํ•ฉ์„ฑ ์‹คํ–‰
result_path, response_text = merge_images(
person_img_path=person_path,
product_img_path=product_path,
background_img_path=background_path,
prompt=prompt
)
# ์ด๋ฏธ์ง€ ๋ฐ˜ํ™˜ ๋ฐ ์ž„์‹œ ํŒŒ์ผ ์ •๋ฆฌ
if result_path:
logger.debug(f"์ด๋ฏธ์ง€ ์ƒ์„ฑ ์™„๋ฃŒ. ๊ฒฝ๋กœ: {result_path}")
try:
result_img = Image.open(result_path)
if result_img.mode == "RGBA":
result_img = result_img.convert("RGB")
# ์ž„์‹œ ํŒŒ์ผ ์ •๋ฆฌ
try:
os.unlink(person_path)
os.unlink(product_path)
if background_path:
os.unlink(background_path)
except Exception as e:
logger.warning(f"์ž„์‹œ ํŒŒ์ผ ์‚ญ์ œ ์ค‘ ์˜ค๋ฅ˜: {str(e)}")
return [result_img], response_text or "์ด๋ฏธ์ง€๊ฐ€ ์„ฑ๊ณต์ ์œผ๋กœ ์ƒ์„ฑ๋˜์—ˆ์Šต๋‹ˆ๋‹ค."
except Exception as e:
logger.exception(f"๊ฒฐ๊ณผ ์ด๋ฏธ์ง€ ๋กœ๋“œ ์ค‘ ์˜ค๋ฅ˜: {str(e)}")
return [], f"๊ฒฐ๊ณผ ์ด๋ฏธ์ง€ ์ฒ˜๋ฆฌ ์ค‘ ์˜ค๋ฅ˜: {str(e)}"
else:
logger.error("merge_images ํ•จ์ˆ˜์—์„œ None ๋ฐ˜ํ™˜๋จ.")
return [], response_text or "์ด๋ฏธ์ง€ ์ƒ์„ฑ์— ์‹คํŒจํ–ˆ์Šต๋‹ˆ๋‹ค. ๋‹ค๋ฅธ ์ด๋ฏธ์ง€๋‚˜ ํ”„๋กฌํ”„ํŠธ๋กœ ์‹œ๋„ํ•ด๋ณด์„ธ์š”."
except Exception as e:
logger.exception("process_images_and_prompt ํ•จ์ˆ˜์—์„œ ์˜ค๋ฅ˜ ๋ฐœ์ƒ:")
return [], f"์˜ค๋ฅ˜ ๋ฐœ์ƒ: {str(e)}" # ์˜ค๋ฅ˜ ์‹œ ๋นˆ ๋ฆฌ์ŠคํŠธ์™€ ์˜ค๋ฅ˜ ๋ฉ”์‹œ์ง€ ๋ฐ˜ํ™˜
# --- Gradio ์ธํ„ฐํŽ˜์ด์Šค ๊ตฌ์„ฑ ---
with gr.Blocks() as demo:
gr.HTML(
"""
<div style='display: flex; align-items: center; justify-content: center; gap: 20px'>
<div style="background-color: var(--block-background-fill); border-radius: 8px">
<img src="https://www.gstatic.com/lamda/images/gemini_favicon_f069958c85030456e93de685481c559f160ea06b.png" style="width: 100px; height: 100px;">
</div>
<div>
<h1>Gemini๋ฅผ ์ด์šฉํ•œ ์ด๋ฏธ์ง€ ํ•ฉ์„ฑ</h1>
<p>์‚ฌ๋žŒ, ์ƒํ’ˆ, ๋ฐฐ๊ฒฝ ์ด๋ฏธ์ง€๋ฅผ ํ•ฉ์„ฑํ•˜๋Š” ์• ํ”Œ๋ฆฌ์ผ€์ด์…˜์ž…๋‹ˆ๋‹ค.</p>
</div>
</div>
"""
)
gr.Markdown("์‚ฌ๋žŒ ์ด๋ฏธ์ง€, ์ƒํ’ˆ ์ด๋ฏธ์ง€, ๋ฐฐ๊ฒฝ ์ด๋ฏธ์ง€๋ฅผ ์—…๋กœ๋“œํ•˜๊ณ , ์–ด๋–ป๊ฒŒ ํ•ฉ์„ฑํ• ์ง€ ์„ค๋ช…ํ•ด์ฃผ์„ธ์š”.")
with gr.Row():
with gr.Column():
person_input = gr.Image(type="pil", label="์‚ฌ๋žŒ ์ด๋ฏธ์ง€ (ํ•„์ˆ˜)", image_mode="RGB")
product_input = gr.Image(type="pil", label="์ƒํ’ˆ ์ด๋ฏธ์ง€ (ํ•„์ˆ˜)", image_mode="RGB")
background_input = gr.Image(type="pil", label="๋ฐฐ๊ฒฝ ์ด๋ฏธ์ง€ (์„ ํƒ ์‚ฌํ•ญ)", image_mode="RGB")
prompt_input = gr.Textbox(
lines=2,
placeholder="ํ•ฉ์„ฑ ๋ฐฉ๋ฒ•์„ ์„ค๋ช…ํ•ด์ฃผ์„ธ์š”. (์˜ˆ: '์ด ๋ฐฐ๊ฒฝ์—์„œ ์ด ์‚ฌ๋žŒ์ด ์ƒํ’ˆ์„ ์‚ฌ์šฉํ•˜๋Š” ๋ชจ์Šต' ๋˜๋Š” '์ด ์‚ฌ๋žŒ์ด ์ด ์ƒํ’ˆ์„ ๋“ค๊ณ  ์ด ๋ฐฐ๊ฒฝ์— ์„œ ์žˆ๋Š” ๋ชจ์Šต')",
label="ํ•ฉ์„ฑ ๋ฐฉ๋ฒ• ์„ค๋ช…"
)
submit_btn = gr.Button("์ด๋ฏธ์ง€ ํ•ฉ์„ฑ ์‹คํ–‰")
with gr.Column():
output_gallery = gr.Gallery(label="ํ•ฉ์„ฑ ๊ฒฐ๊ณผ")
output_text = gr.Textbox(label="AI ์‘๋‹ต ํ…์ŠคํŠธ", visible=True)
submit_btn.click(
fn=process_images_and_prompt,
inputs=[person_input, product_input, background_input, prompt_input],
outputs=[output_gallery, output_text],
)
gr.HTML("""
<div style="margin-top: 20px; padding: 10px; background-color: #f8f9fa; border-radius: 8px;">
<h3>์‚ฌ์šฉ ํŒ:</h3>
<ul>
<li><strong>๋ฐฐ๊ฒฝ ํ™œ์šฉ:</strong> "์ด ๋ฐฐ๊ฒฝ์—์„œ ์ด ์‚ฌ๋žŒ์ด ์ด ์ƒํ’ˆ์„ ์‚ฌ์šฉํ•˜๋Š” ์ž์—ฐ์Šค๋Ÿฌ์šด ๋ชจ์Šต์„ ๋งŒ๋“ค์–ด์ฃผ์„ธ์š”."</li>
<li><strong>ํŠน์ • ์œ„์น˜ ์ง€์ •:</strong> "์ด ์‚ฌ๋žŒ์ด ์ด ์ƒํ’ˆ์„ ์†์— ๋“ค๊ณ  ์ด ๋ฐฐ๊ฒฝ ์•ž์— ์„œ ์žˆ๋Š” ๋ชจ์Šต์„ ๋ณด์—ฌ์ฃผ์„ธ์š”."</li>
<li><strong>์ƒํ’ˆ ๊ฐ•์กฐ:</strong> "์ด ์‚ฌ๋žŒ์ด ์ด ๋ฐฐ๊ฒฝ์—์„œ ์ด ์ƒํ’ˆ์„ ์‚ฌ์šฉํ•˜๋Š” ๋ชจ์Šต์„ ๋ณด์—ฌ์ฃผ๋˜, ์ƒํ’ˆ์ด ์ž˜ ๋ณด์ด๋„๋ก ํ•ด์ฃผ์„ธ์š”."</li>
<li><strong>์žฅ๋ฉด ์„ค์ •:</strong> "์ด ์‚ฌ๋žŒ์ด ์ด ์ƒํ’ˆ์œผ๋กœ ์š”๋ฆฌํ•˜๋Š” ๋ชจ์Šต์„ ์ด ์ฃผ๋ฐฉ ๋ฐฐ๊ฒฝ์—์„œ ๋ณด์—ฌ์ฃผ์„ธ์š”."</li>
<li><strong>์˜์–ด ํ”„๋กฌํ”„ํŠธ:</strong> ๋” ๋‚˜์€ ๊ฒฐ๊ณผ๋ฅผ ์œ„ํ•ด ์˜์–ด์™€ ํ•œ๊ตญ์–ด๋ฅผ ํ•จ๊ป˜ ์‚ฌ์šฉํ•ด ๋ณด์„ธ์š”.</li>
<li><strong>๋ฐฐ๊ฒฝ ์„ ํƒ์‚ฌํ•ญ:</strong> ๋ฐฐ๊ฒฝ ์ด๋ฏธ์ง€๋Š” ์„ ํƒ์‚ฌํ•ญ์ž…๋‹ˆ๋‹ค. ๋ฐฐ๊ฒฝ ์—†์ด ์‚ฌ๋žŒ๊ณผ ์ƒํ’ˆ๋งŒ ํ•ฉ์„ฑํ•  ์ˆ˜๋„ ์žˆ์Šต๋‹ˆ๋‹ค.</li>
</ul>
</div>
""")
# --- ์‹คํ–‰ ---
if __name__ == "__main__":
demo.launch(share=True)