别再一帧帧画了！AI漫剧全流程源码拆解：从剧本到成片，1200字带代码跑通

# script_gen.py
import json, openai

SCHEMA = {
  "type": "object",
  "properties": {
    "title": {"type": "string"},
    "characters": {"type": "array", "items": {
      "type": "object",
      "properties": {"name": {"type": "string"}, "desc": {"type": "string"}}
    }},
    "shots": {"type": "array", "items": {
      "type": "object",
      "properties": {
        "shot_id": {"type": "integer"},
        "scene": {"type": "string"},
        "camera": {"type": "string", "enum": ["wide","medium","close-up","over-shoulder"]},
        "dialogue": {"type": "string"},
        "speaker": {"type": "string"},
        "duration_sec": {"type": "number"}
      },
      "required": ["shot_id","scene","camera","dialogue","speaker","duration_sec"]
    }}
  },
  "required": ["title","characters","shots"]
}

def gen_script(topic):
    resp = openai.chat.completions.create(
        model="gpt-4o",
        response_format={"type": "json_schema", "json_schema": {"name":"manju","schema":SCHEMA}},
        messages=[{"role":"user","content":f"生成3分钟漫剧分镜，主题：{topic}，每镜5-8秒"}]
    )
    return json.loads(resp.choices[0].message.content)

# image_gen.py
import requests, hashlib

def gen_character_ref(name, desc, api_key):
    r = requests.post("https://api.openai.com/v1/images/generations",
        headers={"Authorization": f"Bearer {api_key}"},
        json={"model":"gpt-image-1","prompt":f"动漫角色设定图，{name}，{desc}，正面全身，白底",
              "size":"1024x1024","n":1})
    return r.json()["data"][0]["url"]

def gen_shot(shot, char_refs, api_key):
    # seed由角色名+镜头号哈希固定，保证可复现
    seed = int(hashlib.md5(f"{shot['speaker']}-{shot['shot_id']}".encode()).hexdigest()[:8], 16) % 2**31
    prompt = f"{shot['camera']}，{shot['scene']}，角色参考：{char_refs.get(shot['speaker'],'')}"
    r = requests.post("https://api.openai.com/v1/images/generations",
        headers={"Authorization": f"Bearer {api_key}"},
        json={"model":"gpt-image-1","prompt":prompt,"size":"1024x1024","n":1,"seed":seed})
    return r.json()["data"][0]["url"]

# tts.py
def tts(text, voice, out_path, api_key):
    r = requests.post("https://api.openai.com/v1/audio/speech",
        headers={"Authorization": f"Bearer {api_key}"},
        json={"model":"gpt-4o-mini-tts","input":text,"voice":voice,"response_format":"mp3"})
    open(out_path,"wb").write(r.content)
    # 返回实际时长，用于校正分镜duration
    return get_duration(out_path)

# 单镜头：静帧做缓慢推拉运镜，叠加音频
ffmpeg -loop 1 -i shot_01.png -i shot_01.mp3 \
  -filter_complex "[0:v]scale=1280:720,zoompan=z='min(zoom+0.0015,1.1)':d=125:s=1280x720[v]" \
  -map "[v]" -map 1:a -shortest -c:v libx264 -pix_fmt yuv420p shot_01.mp4

# 多镜头拼接
ffmpeg -f concat -safe 0 -i shots.txt -c copy final.mp4