(\ _ /)
( ･-･)
/っ 
用 NextChat 直接收发 Base64 图片的话 Tokens 会裂开，图床又懒得整，就把 Imagen 的 py 拿来改改用。
前置工作： Python 环境自己问大模型安装，GPT-3.5 都能胜任。
缺失的组件也是直接问 AI，基本都是 pip install 组件名 Done.
之前没考虑到 Gemini 还能 一次过吐多张图片 的，更新了下源码。
没啥好介绍的，直接放源码：
import requests
import json
import os
import sys
import time
import base64
import tkinter as tk
from tkinter import filedialog, messagebox
from pathlib import Path
# 设置代理（如需要）
os.environ.update({'http_proxy': 'http://127.0.0.1:1081', 'https_proxy': 'http://127.0.0.1:1081'})
# 请在此处填入您的 Gemini API 密钥
API_KEY = "*" # 替换为您的实际 Gemini API 密钥
def get_mime_type(file_path):
"""根据文件扩展名获取MIME类型"""
ext = os.path.splitext(file_path)[1].lower()
mime_types = {
'.jpg': 'image/jpeg',
'.jpeg': 'image/jpeg',
'.png': 'image/png',
'.gif': 'image/gif',
'.bmp': 'image/bmp',
'.webp': 'image/webp'
}
return mime_types.get(ext, 'image/jpeg')
def image_to_base64(image_path):
"""将图片文件转换为base64编码"""
with open(image_path, "rb") as image_file:
encoded_string = base64.b64encode(image_file.read()).decode('utf-8')
return encoded_string
def select_image_files():
"""打开文件选择对话框并返回选择的多个图片路径"""
root = tk.Tk()
root.withdraw() # 隐藏主窗口
root.attributes('-topmost', True) # 确保对话框在最前面
file_paths = filedialog.askopenfilenames(
title="选择参考图片（可多选）",
filetypes=[
("图片文件", "*.jpg *.jpeg *.png *.bmp *.gif *.webp"),
("所有文件", "*.*")
]
)
root.destroy()
return file_paths
def save_base64_image(base64_data, output_path):
"""将 base64 编码的图片保存到文件"""
# 从数据 URL 中提取 base64 部分
if "base64," in base64_data:
base64_data = base64_data.split("base64,")[1]
with open(output_path, "wb") as f:
f.write(base64.b64decode(base64_data))
return output_path
def ask_text_to_image_mode():
"""询问用户是否使用文生图模式"""
root = tk.Tk()
root.withdraw()
result = messagebox.askyesno("文生图模式", "未选择图片。是否使用文生图模式？")
root.destroy()
return result
def get_text_prompt(default_prompt="", is_text_to_image=False):
"""获取用户输入的文本提示词，优化UI样式"""
root = tk.Tk()
root.title("输入提示词" if not is_text_to_image else "输入文生图提示词")
# 设置窗口大小和位置
window_width = 500
window_height = 180
screen_width = root.winfo_screenwidth()
screen_height = root.winfo_screenheight()
center_x = int(screen_width/2 - window_width/2)
center_y = int(screen_height/2 - window_height/2)
root.geometry(f'{window_width}x{window_height}+{center_x}+{center_y}')
# 创建框架来包含所有元素
main_frame = tk.Frame(root, padx=10, pady=10)
main_frame.pack(fill=tk.BOTH, expand=True)
# 输入框，减少内边距
text_entry = tk.Text(main_frame, height=5, width=50, padx=5, pady=5)
text_entry.pack(fill=tk.BOTH, expand=True)
text_entry.insert("1.0", default_prompt)
# 创建底部按钮框架
button_frame = tk.Frame(main_frame)
button_frame.pack(fill=tk.X, pady=(5, 0))
prompt_result = {"text": ""}
def on_submit():
prompt_result["text"] = text_entry.get("1.0", "end-1c")
root.destroy()
# 按钮放在右侧
submit_button = tk.Button(button_frame, text="提交", command=on_submit, width=10)
submit_button.pack(side=tk.RIGHT)
# 绑定回车键提交（按Ctrl+Enter提交）
def on_ctrl_enter(event):
on_submit()
return "break" # 阻止默认行为
text_entry.bind("<Control-Return>", on_ctrl_enter)
# 确保窗口显示后聚焦到文本框
def set_focus():
text_entry.focus_force() # 强制聚焦
text_entry.mark_set("insert", "end") # 设置光标位置到末尾
# 使用after方法确保窗口完全显示后再聚焦
root.after(100, set_focus)
root.mainloop()
return prompt_result["text"]
def main():
# 弹出文件选择对话框
print("请选择一个或多个参考图片文件...")
image_paths = select_image_files()
# 获取当前脚本所在目录
script_dir = os.path.dirname(os.path.abspath(__file__))
# 创建输出目录
output_dir = os.path.join(script_dir, "Gemini Images")
os.makedirs(output_dir, exist_ok=True)
timestamp = int(time.time())
# 检查是否选择了图片
if not image_paths:
# 询问是否使用文生图模式
use_text_to_image = ask_text_to_image_mode()
if not use_text_to_image:
print("未选择图片且不使用文生图模式，退出程序。")
return
# 获取用户输入的提示词
prompt_text = get_text_prompt("", True)
if not prompt_text.strip():
print("未输入提示词，退出程序。")
return
print(f"使用文生图模式，提示词: {prompt_text}")
# 构建文生图请求数据
request_data = {
"contents": [{
"parts":[
{"text": prompt_text}
]
}],
"generationConfig": {
"maxOutputTokens": 8192,
"responseModalities": ["Text", "Image"]
},
"safetySettings": [{
"category": "HARM_CATEGORY_HARASSMENT",
"threshold": "BLOCK_NONE"
}, {
"category": "HARM_CATEGORY_HATE_SPEECH",
"threshold": "BLOCK_NONE"
}, {
"category": "HARM_CATEGORY_SEXUALLY_EXPLICIT",
"threshold": "BLOCK_NONE"
}, {
"category": "HARM_CATEGORY_DANGEROUS_CONTENT",
"threshold": "BLOCK_NONE"
}, {
"category": "HARM_CATEGORY_CIVIC_INTEGRITY",
"threshold": "BLOCK_NONE"
}]
}
# 生成文件名
request_id = f"text_to_image_{timestamp}"
else:
print(f"已选择 {len(image_paths)} 张图片")
# 获取第一个文件的名称作为请求标识
first_file_name = os.path.splitext(os.path.basename(image_paths[0]))[0]
if len(image_paths) > 1:
request_id = f"{first_file_name}_and_{len(image_paths)-1}_more_{timestamp}"
else:
request_id = f"{first_file_name}_{timestamp}"
# 获取用户输入的提示词
default_prompt = ""
prompt_text = get_text_prompt(default_prompt)
# 打印提示词 - 添加这一行以解决问题②
print(f"使用的提示词: {prompt_text}")
# 构建多图像请求数据
parts = [{"text": prompt_text}]
# 添加所有图像到请求
for image_path in image_paths:
try:
base64_image = image_to_base64(image_path)
mime_type = get_mime_type(image_path)
# 添加图像部分
parts.append({
"inline_data": {
"mime_type": mime_type,
"data": base64_image
}
})
except Exception as e:
print(f"处理图片 {image_path} 时出错: {e}")
# 构建完整请求
request_data = {
"contents": [{
"parts": parts
}],
"generationConfig": {
"maxOutputTokens": 8192,
"responseModalities": ["Text", "Image"]
},
"safetySettings": [{
"category": "HARM_CATEGORY_HARASSMENT",
"threshold": "BLOCK_NONE"
}, {
"category": "HARM_CATEGORY_HATE_SPEECH",
"threshold": "BLOCK_NONE"
}, {
"category": "HARM_CATEGORY_SEXUALLY_EXPLICIT",
"threshold": "BLOCK_NONE"
}, {
"category": "HARM_CATEGORY_DANGEROUS_CONTENT",
"threshold": "BLOCK_NONE"
}, {
"category": "HARM_CATEGORY_CIVIC_INTEGRITY",
"threshold": "BLOCK_NONE"
}]
}
# 发送请求
print("正在发送请求到 Gemini API...")
try:
api_url = f'https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash-exp-image-generation:generateContent?key={API_KEY}'
response = requests.post(
api_url,
headers={
'Content-Type': 'application/json'
},
json=request_data,
proxies={
'http': 'http://127.0.0.1:1081',
'https': 'http://127.0.0.1:1081'
}
)
# 保存请求数据用于调试
debug_request_file = os.path.join(output_dir, f"{request_id}_request.json")
with open(debug_request_file, 'w', encoding='utf-8') as f:
# 移除base64数据以避免文件过大，仅用于调试
debug_data = request_data.copy()
if image_paths: # 只有在有图片时才需要移除base64数据
for i in range(1, len(debug_data["contents"][0]["parts"])):
if "inline_data" in debug_data["contents"][0]["parts"][i]:
debug_data["contents"][0]["parts"][i]["inline_data"]["data"] = "[BASE64_DATA_REMOVED_FOR_DEBUGGING]"
json.dump(debug_data, f, ensure_ascii=False, indent=2)
# 检查响应状态
if response.status_code != 200:
print(f"API返回错误: {response.status_code}")
print(f"错误详情: {response.text}")
# 保存错误响应
error_file = os.path.join(output_dir, f"{request_id}_error.json")
with open(error_file, 'w', encoding='utf-8') as f:
try:
error_json = response.json()
json.dump(error_json, f, ensure_ascii=False, indent=2)
except:
f.write(response.text)
print(f"错误信息已保存到: {error_file}")
sys.exit(1)
# 解析响应
result = response.json()
# 保存完整响应到 JSON 文件
response_file = os.path.join(output_dir, f"{request_id}_response.json")
with open(response_file, 'w', encoding='utf-8') as f:
json.dump(result, f, ensure_ascii=False, indent=2)
# 尝试提取和保存图片
try:
# 根据Gemini API响应结构，提取图像数据
if "candidates" in result and len(result["candidates"]) > 0:
candidate = result["candidates"][0]
if "content" in candidate and "parts" in candidate["content"]:
parts = candidate["content"]["parts"]
# 遍历parts寻找图片数据
image_count = 0
for i, part in enumerate(parts):
# 检查是否有inlineData字段（注意大小写与响应一致）
if "inlineData" in part and "data" in part["inlineData"]:
image_count += 1
image_data = part["inlineData"]["data"]
image_type = part["inlineData"].get("mimeType", "image/png")
ext = image_type.split("/")[-1]
# 保存图片，使用与请求/响应文件相同的命名格式，为多图片添加序号
image_file = os.path.join(output_dir, f"{request_id}_generated_{image_count}.{ext}")
save_base64_image(image_data, image_file)
print(f"生成成功！图片 {image_count} 已保存到: {image_file}")
if image_count == 0:
print("响应中未找到图片数据，请检查响应JSON文件")
else:
print(f"共保存了 {image_count} 张生成的图片")
else:
print("响应结构中未找到content或parts字段，请检查响应JSON文件")
else:
print("响应中未找到candidates字段，请检查响应JSON文件")
# 打开保存目录
print(f"所有文件已存储到: {output_dir}")
os.startfile(output_dir)
except Exception as e:
print(f"处理响应时出错: {e}")
print("请查看保存的 JSON 文件以了解完整响应结构")
except requests.exceptions.RequestException as e:
print(f"请求错误: {str(e)}")
sys.exit(1)
except Exception as e:
print(f"发生错误: {str(e)}")
sys.exit(1)
if __name__ == "__main__":
main()
- API_KEY 填你自己的。
- 127.0.0.1:1081 是 V2RayN 的默认入口，改成自己梯子的入口。
- 如果用论坛佬友的代理端口就直接注释掉代理那行。
- ~ -
- 如果不选择图片（摁 Esc）则使用文生图
- 图片编辑用的是 Imagen2（目测鉴定）
效果：
如果效果不好想用上次的 Promot 就去翻同名 json 
不用那么麻烦了，改成直接输出到控制台了。
(\ _ /)
( ･-･)
/っ 
就是这样。目前 Gemini 生图只是实验阶段的玩具。
去找 Gemini 大王们要，我就一个 Key。