diff --git a/image_generation/popiskill-image-real2anime/SKILL.md b/image_generation/popiskill-image-real2anime/SKILL.md new file mode 100644 index 0000000..d79b265 --- /dev/null +++ b/image_generation/popiskill-image-real2anime/SKILL.md @@ -0,0 +1,179 @@ +--- +name: popiskill-image-real2anime +version: 1.0.0 +description: > + 真人照片转动漫风格，支持多种动漫风格转换 +homepage: https://llmapitest.popi.art +user-invocable: true +metadata: + { + "openclaw": + { + "emoji": "🎭", + "requires": { "bins": ["python3"], "env": ["POPI_OPENAPI_KEY"] }, + "primaryEnv": "POPI_OPENAPI_KEY", + }, + } +--- +# 真人照片转动漫风格 + +将真人照片转换为多种动漫/艺术风格，支持16种风格选择。 + +## 环境配置指引 + +### 密钥配置 + +本 Skill 需要 Popi API 密钥才能正常工作。 + +#### 方式一：环境变量 + +**Linux / macOS：** + +```bash +export POPI_OPENAPI_KEY="your-api-key-here" +``` + +如需持久化： + +```bash +echo 'export POPI_OPENAPI_KEY="your-api-key-here"' >> ~/.zshrc +source ~/.zshrc +``` + +**Windows (PowerShell)：** + +```powershell +$env:POPI_OPENAPI_KEY = "your-api-key-here" +``` + +#### 方式二：openclaw.json 配置（推荐） + +在 `~/.openclaw/openclaw.json` 中添加： + +```json +{ + "skills": { + "entries": { + "popiskill-image-real2anime": { + "enabled": true, + "env": { + "POPI_OPENAPI_KEY": "your-api-key-here" + } + } + } + } +} +``` + +## 支持的风格 + +风格配置存储在 skill 目录下的 `style.json` 文件中。 + +| 风格名称 | 说明 | +|----------|------| +| JOJO漫画风格 | 《JOJO的奇妙冒险》漫画风格 | +| 水彩风格 | 吉卜力水彩风格，柔和 pastel 色调 | +| 迪士尼3D动画风格 | 迪士尼3D动画质感 | +| 80年代复古漫画风格 | 80年代日本复古动画风格 | +| 像素风格 | 像素化游戏风格 | +| 日系漫画风格 | 日式动画风格 | +| 扁平插画风格 | Flat Design 扁平设计 | +| 黑白漫画风格 | 黑白漫画风格 | +| 2.5D游戏风格 | 《崩坏：星穹铁道》风格化3D | +| 饥荒风格 | 《Don't Starve》哥特卡通风格 | +| 水墨国风风格 | 国风水墨风格 | +| 彩铅风格 | 彩色铅笔手绘风格 | +| 毛绒娃娃风格 | 毛绒玩具/玩偶风格 | +| 卡通INS风格 | Q版INS风格 | +| Q版萌趣风格 | Q版萌趣风格 | +| 恶搞之家风格 | 《恶搞之家》动画风格 | +| 韩系漫画风格 | 现代韩漫风格 | + +## Agent 执行指令（必读） + +> 本节是 Agent（AI 模型）的核心执行规范。当用户请求生成图片时，Agent 必须严格按照以下步骤自主执行，无需询问用户确认。 + +### 通用执行规则 + +1. **触发条件**：用户上传了真人照片，要求转换为动漫/艺术风格。 +2. **零交互原则**：Agent 应直接执行脚本，不要向用户询问任何确认。 +3. **模型选择**：固定使用 **nano-pro** 模型（用户指定要求） +4. **风格选择**：从用户输入中提取风格名称，匹配 style.json 中的风格 + +--- + +### 脚本：真人照片转动漫风格 + +**适用场景**：用户上传了图片，需要转换为动漫/艺术风格 + +**图片处理步骤（必须执行）**： +1. 首先创建临时文件夹：`mkdir -p ~/.openclaw/workspace/skills/popiskill-image-real2anime/scripts/tmp` +2. 将用户上传的图片原封不动复制到上述 tmp 文件夹（不压缩、不修改） +3. 使用复制后的图片路径作为 `--image-path` 参数 + +**风格处理步骤**： +1. 从用户输入中提取风格名称（如"JOJO风格"、"水彩风格"等） +2. 读取 skill 目录下的 `style.json` 获取对应风格的 prompt +3. 使用该 prompt 作为 `--prompt` 参数 + +**固定参数**： + +``` +- `--model`: nano-pro（固定） +- `--prompt`: 从 style.json 读取对应风格的 prompt +- `--image-path`: 用户上传图片路径 +``` + +**可变参数**： + +| 参数 | 说明 | 默认值 | 可选值 | +|------|------|--------|--------| +| `--size` | 图片尺寸 | 2K | 2K, 4K, 1024x1024, 1280x1280 | +| `--aspect-ratio` | 宽高比 | 16:9 | 16:9, 4:3, 1:1, 2:3, 3:2, 9:16 | + +**执行命令**： + +```bash +python3 scripts/generate_image.py nano-pro --prompt "{从style.json读取的prompt}" --image-path "tmp/uploaded_image.png" --size 2K --aspect-ratio 16:9 +``` + +**输出示例**： + +```json +{ + "path": "C:/Users/xxx/.openclaw/workspace/scripts/image_20250325_161200.png", + "model": "nano-pro", + "size": "2K", + "aspect_ratio": "16:9" +} +``` + +### 完整调用示例 + +**示例 1：JOJO漫画风格** +```bash +python3 scripts/generate_image.py nano-pro --prompt "将参考图中的角色转化为《JOJO的奇妙冒险》漫画风格。保持构图和核心特征" --image-path "tmp/photo.jpg" --size 2K --aspect-ratio 16:9 +``` + +**示例 2：水彩风格** +```bash +python3 scripts/generate_image.py nano-pro --prompt "Transform the characters in the reference image into《Ghibli》style: soft watercolor edges, gentle pastel washes, clean white background, high contrast, muted palette with selective color pops, warm yet whimsical Studio-Ghibli vibe. Masterpiece quality, intricate hand-painted detail" --image-path "tmp/photo.jpg" --size 2K --aspect-ratio 16:9 +``` + +**示例 3：迪士尼3D动画风格** +```bash +python3 scripts/generate_image.py nano-pro --prompt "将参考图中的角色转变为具有《迪士尼3d动画》质感的动画风格，背景变为白色背景" --image-path "tmp/photo.jpg" --size 2K --aspect-ratio 16:9 +``` + +### Agent 须避免的行为 + +- 只打印脚本路径而不执行 +- 向用户询问"是否要执行图片生成"——应直接执行 +- 忘记读取输出结果中的 `path` 并返回给用户 +- 图像生成失败时，自行编造图片路径 +- 使用非 nano-pro 模型（用户明确要求使用 nano-pro） +- 忽略用户指定的风格，使用默认 prompt + +## API 参考文档 + +暂无 diff --git a/image_generation/popiskill-image-real2anime/generation/gemini_edited_20260328_120648.png b/image_generation/popiskill-image-real2anime/generation/gemini_edited_20260328_120648.png new file mode 100644 index 0000000..a38b30a Binary files /dev/null and b/image_generation/popiskill-image-real2anime/generation/gemini_edited_20260328_120648.png differ diff --git a/image_generation/popiskill-image-real2anime/scripts/_common_new.py b/image_generation/popiskill-image-real2anime/scripts/_common_new.py new file mode 100644 index 0000000..6ae3684 --- /dev/null +++ b/image_generation/popiskill-image-real2anime/scripts/_common_new.py @@ -0,0 +1,158 @@ +"""Popiart AI OpenAPI 公共模块：HTTP 请求、鉴权、图片保存等工具函数""" + +import base64 +import datetime +import json +import os +import re +import struct +import sys +import time +from typing import Optional, Tuple +import urllib.request +import urllib.error + +# ========== 配置 ========== +POPI_OPENAPI_URL = os.environ.get("POPI_OPENAPI_URL", "https://llmapitest.popi.art") +API_KEY = os.environ.get("POPI_OPENAPI_KEY", "") + +if not API_KEY: + print("错误：请设置 POPI_OPENAPI_KEY 环境变量", file=sys.stderr) + print(" export POPI_OPENAPI_KEY=sk-xxxx", file=sys.stderr) + sys.exit(1) + +# 轮询配置 +POLL_INTERVAL = 30 # 秒 +MAX_WAIT_TIME = 600 # 秒 +# ========================== + + +def headers(content_type="application/json"): + h = {"Authorization": f"Bearer {API_KEY}"} + if content_type: + h["Content-Type"] = content_type + return h + + +def post_json(path: str, body: dict, timeout: int = 120) -> dict: + url = f"{POPI_OPENAPI_URL.rstrip('/')}{path}" + data = json.dumps(body).encode("utf-8") + req = urllib.request.Request(url, data=data, method="POST", headers=headers()) + try: + with urllib.request.urlopen(req, timeout=timeout) as resp: + return json.loads(resp.read().decode("utf-8")) + except urllib.error.HTTPError as e: + body_text = e.read().decode("utf-8") if e.fp else "" + print(f"❌ HTTP {e.code}: {body_text}", file=sys.stderr) + sys.exit(1) + except urllib.error.URLError as e: + print(f"❌ 网络错误: {e.reason}", file=sys.stderr) + sys.exit(1) + + +def get_json(path: str, timeout: int = 30) -> dict: + url = f"{POPI_OPENAPI_URL.rstrip('/')}{path}" + req = urllib.request.Request(url, method="GET", headers=headers(content_type=None)) + try: + with urllib.request.urlopen(req, timeout=timeout) as resp: + return json.loads(resp.read().decode("utf-8")) + except urllib.error.HTTPError as e: + body_text = e.read().decode("utf-8") if e.fp else "" + print(f"❌ HTTP {e.code}: {body_text}", file=sys.stderr) + sys.exit(1) + except urllib.error.URLError as e: + print(f"❌ 网络错误: {e.reason}", file=sys.stderr) + sys.exit(1) + + +def post_multipart(path: str, fields: dict, file_fields: list = None, timeout: int = 30) -> dict: + """ + multipart/form-data POST。 + fields: {name: value} + file_fields: [(field_name, filename, bytes, mime_type), ...] 或 None，支持多个同名字段 + """ + import requests as _requests + url = f"{POPI_OPENAPI_URL.rstrip('/')}{path}" + # requests 支持 list of tuples 以传递同名多字段 + files = [(k, (None, v)) for k, v in fields.items()] + for fname, filename, data, mime in (file_fields or []): + files.append((fname, (filename, data, mime))) + resp = _requests.post(url, headers={"Authorization": f"Bearer {API_KEY}"}, files=files, timeout=timeout) + if resp.status_code != 200: + print(f"❌ HTTP {resp.status_code}: {resp.text}", file=sys.stderr) + sys.exit(1) + return resp.json() + + +def download_stream(path: str, output_path: str, timeout: int = 120) -> bool: + """流式下载文件到 output_path，返回是否成功""" + import requests as _requests + url = f"{POPI_OPENAPI_URL.rstrip('/')}{path}" + resp = _requests.get(url, headers={"Authorization": f"Bearer {API_KEY}"}, stream=True, timeout=timeout) + if resp.status_code != 200: + print(f"❌ 下载失败 HTTP {resp.status_code}: {resp.text}", file=sys.stderr) + return False + total = int(resp.headers.get("content-length", 0)) + downloaded = 0 + with open(output_path, "wb") as f: + for chunk in resp.iter_content(chunk_size=8192): + if chunk: + f.write(chunk) + downloaded += len(chunk) + if total: + print(f"\r {downloaded/total*100:.1f}% ({downloaded}/{total} bytes)", end="") + print() + return True + + +def save_base64_image(b64_data: str, prefix: str = "image") -> str: + """将 base64 图片数据保存为 PNG 文件，返回文件路径""" + ts = datetime.datetime.now().strftime("%Y%m%d_%H%M%S") + filename = f"{prefix}_{ts}.png" + script_dir = os.path.dirname(os.path.abspath(__file__)) + generation_dir = os.path.dirname(script_dir) + filepath = os.path.join(generation_dir,"generation", filename) + with open(filepath, "wb") as f: + f.write(base64.b64decode(b64_data)) + # 输出超链接格式（使用 file:// 协议） + print(f"\n🔗 点击查看图片: file://{filepath}") + return filepath + + + +def extract_image_urls(markdown_text: str) -> list: + """从 markdown 文本中提取图片 URL""" + return re.findall(r'!\[.*?\]\((https?://[^)]+)\)', markdown_text) + + +def poll_video(video_id: str) -> Optional[str]: + """ + 轮询视频生成状态，返回最终 task_id（用于下载），失败返回 None。 + 网关返回结构：{ code, data: { status, progress, result_url, ... } } + """ + url = f"/v1/video/generations/{video_id}" + start = time.time() + while True: + elapsed = int(time.time() - start) + if elapsed > MAX_WAIT_TIME: + print(f"⏱️ 超时：已等待 {MAX_WAIT_TIME}s") + return None + result = get_json(url) + data = result.get("data") or result + status = data.get("status", "") + progress = data.get("progress", 0) + print(f" [{elapsed:>4}s] status={status} progress={progress}%") + if status == "SUCCESS": + url_result = data.get("result_url", "") + if url_result: + print(f" result_url: {url_result}") + return data.get("task_id") or video_id + elif status == "FAILURE": + print(f"❌ 生成失败: {data.get('fail_reason', '')}") + return None + print(f" 等待 {POLL_INTERVAL}s...") + time.sleep(POLL_INTERVAL) + + +def ts() -> str: + return datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S") diff --git a/image_generation/popiskill-image-real2anime/scripts/generate_image.py b/image_generation/popiskill-image-real2anime/scripts/generate_image.py new file mode 100644 index 0000000..288c4ad --- /dev/null +++ b/image_generation/popiskill-image-real2anime/scripts/generate_image.py @@ -0,0 +1,245 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- +import sys +import io +sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8') +sys.stderr = io.TextIOWrapper(sys.stderr.buffer, encoding='utf-8') + +""" +图像生成 skill 脚本 +用法: + python3 generate_image.py --prompt TEXT [选项] + +模型: + seedream SeeDream 文生图（OpenAI Images API） + nano-pro nano-banana-pro 文生图 / 图生图（Gemini 原生 API） + nano-2 nano-banana-2 文生图 / 图生图（Gemini 原生 API） + sora Sora Image 文生图 / 图生图（Chat Completions API） + 不传则默认 seedream + +图生图触发条件: + nano-pro/nano-2 传 --image-path（本地图片路径） + sora 传 --image-list（一张或多张在线图片 URL） + +必填参数: + --prompt TEXT 图像生成提示词 + +可选参数: + --image-path PATH ... 本地图片路径列表（nano-pro / nano-2 图生图，支持多张） + --image-list URL ... 原图 URL 列表（sora 图生图，支持多张） + --n N 生成数量，>=1，>1 时循环调用（sora，默认 1） + --size SIZE 图片尺寸，如 2K / 4K （默认 2K） + --quality Q 图片质量，standard/hd（seedream，默认 hd） + --ratio W:H 图片比例，如 2:3（sora 文生图，默认 2:3） + --aspect-ratio W:H 宽高比，如 16:9（nano，默认 16:9） + +示例: + python3 generate_image.py seedream --prompt "一只猫" ✅ 完成 + python3 generate_image.py seedream --prompt "一只猫" --n 2 ✅ 完成（n未生效，只返回一张图） + python3 generate_image.py nano-pro --prompt "一只猫" ✅ 完成 + python3 generate_image.py nano-pro --prompt "改成油画风格" --image-path ./cat.jpg ✅ 完成 + python3 generate_image.py nano-pro --prompt "融合两张图" --image-path ./a.jpg ./b.jpg ✅ 完成 + python3 generate_image.py nano-2 --prompt "一只猫" --aspect-ratio 1:1 ✅ 完成 + python3 generate_image.py nano-2 --prompt "改成油画风格" --image-path ./cat.jpg ✅ 完成 + python3 generate_image.py nano-2 --prompt "融合两张图" --image-path ./cat.jpg ./dog.jpg ✅ 完成 + python3 generate_image.py sora --prompt "一只柴犬" ✅ 完成 + python3 generate_image.py sora --prompt "一只柴犬" --n 2 ✅ 完成 + python3 generate_image.py sora --prompt "把毛色改成彩虹色" --image-list https://example.com/dog.jpg ✅ 完成 + python3 generate_image.py sora --prompt "融合两张图" --image-list https://a.com/1.jpg https://b.com/2.jpg ✅ 完成 +""" + +import argparse +import base64 +import os +import sys + +sys.path.insert(0, os.path.dirname(__file__)) +from _common_new import post_json, save_base64_image, extract_image_urls, ts, POPI_OPENAPI_URL, API_KEY + +DEFAULT_MODEL = "seedream" + + +# ---------- seedream ---------- + +def gen_seedream_text(prompt: str, size: str = "2K", quality: str = "hd", n: int = 1): + print(f"模型: seedream 4.5 模式: 文生图 尺寸: {size} 质量: {quality} 数量: {n}") + result = post_json("/v1/images/generations", { + "model": "seedream-4-5-251128", + "prompt": prompt, + "size": size, + "quality": quality, + "n": n, + "response_format": "url", + }, timeout=120) + if "data" in result: + for item in result["data"]: + if "url" in item: + print(f"🔗 {item['url']}") + return item['url'] + print(f"❌ {result}") + return False + + + +# ---------- gemini ---------- + +def gen_gemini_text(prompt: str, model_name: str, aspect_ratio: str = "16:9", size: str = "2K"): + print(f"模型: {model_name} 模式: 文生图 宽高比: {aspect_ratio} 尺寸: {size}") + result = post_json(f"/v1beta/models/{model_name}:generateContent", { + "contents": [{"parts": [{"text": prompt}]}], + "generationConfig": { + "responseModalities": ["IMAGE"], + "imageConfig": {"aspectRatio": aspect_ratio, "imageSize": size}, + }, + }, timeout=300) + if "candidates" in result: + try: + b64 = result["candidates"][0]["content"]["parts"][0]["inlineData"]["data"] + path = save_base64_image(b64, prefix="gemini") + print(f"🖼️ 已保存: {path}") + # return True + return path + except (KeyError, IndexError) as e: + print(f"❌ 解析响应失败: {e}") + print(f"❌ {result}") + return False + + +def gen_gemini_image(image_paths: list, prompt: str, model_name: str, + aspect_ratio: str = "16:9", size: str = "2K"): + print(f"模型: {model_name} 模式: 图生图 图片: {image_paths}") + parts = [{"text": prompt}] + for image_path in image_paths: + try: + with open(image_path, "rb") as f: + img_b64 = base64.b64encode(f.read()).decode() + except FileNotFoundError: + print(f"❌ 文件不存在: {image_path}") + return False + mime = "image/jpeg" if image_path.lower().endswith((".jpg", ".jpeg")) else "image/png" + parts.append({"inlineData": {"mimeType": mime, "data": img_b64}}) + result = post_json(f"/v1beta/models/{model_name}:generateContent", { + "contents": [{"parts": parts}], + "generationConfig": { + "responseModalities": ["IMAGE"], + "imageConfig": {"aspectRatio": aspect_ratio, "imageSize": size}, + }, + }, timeout=300) + if "candidates" in result: + try: + b64 = result["candidates"][0]["content"]["parts"][0]["inlineData"]["data"] + path = save_base64_image(b64, prefix="gemini_edited") + print(f"🖼️ 已保存: {path}") + return {"path": path} + except (KeyError, IndexError) as e: + print(f"❌ 解析响应失败: {e}") + print(f"❌ {result}") + return False + + +# ---------- sora ---------- + +def gen_sora(prompt: str, ratio: str = "2:3", image_list: list = None, n: int = 1): + """ + sora_image 文生图 / 图生图统一入口。 + - image_list 为空 → 文生图，prompt 末尾附加比例标记 + - image_list 非空 → 图生图，每张图作为 image_url 传入 + - n >= 1，大于 1 时循环调用接口生成对应数量图片 + """ + if n < 1: + print("❌ n 必须大于等于 1") + return False + + image_list = image_list or [] + mode = "图生图" if image_list else "文生图" + print(f"模型: sora_image 模式: {mode} 比例: {ratio} 数量: {n}") + if image_list: + print(f"原图: {image_list}") + + def _build_messages(): + if image_list: + content = [{"type": "text", "text": prompt}] + for url in image_list: + content.append({"type": "image_url", "image_url": {"url": url}}) + return [{"role": "user", "content": content}] + else: + p = f"{prompt}【{ratio}】" if ratio in ("2:3", "3:2", "1:1") else prompt + return [{"role": "user", "content": p}] + + success = 0 + for i in range(n): + if n > 1: + print(f"\n[{i+1}/{n}]") + result = post_json("/v1/chat/completions", { + "model": "sora_image", + "messages": _build_messages(), + }, timeout=120) + if "choices" in result: + content = result["choices"][0]["message"]["content"] + urls = extract_image_urls(content) + if urls: + print(f"🔗 {urls[0]}") + else: + print(f"内容: {content[:200]}") + success += 1 + else: + print(f"❌ {result}") + + return success == n + + +# ---------- 主入口 ---------- + +def main(): + parser = argparse.ArgumentParser( + description="图像生成 skill", + formatter_class=argparse.RawDescriptionHelpFormatter, + epilog=__doc__, + ) + parser.add_argument("model", nargs="?", default=DEFAULT_MODEL, + choices=["seedream", "nano-pro", "nano-2", "sora"], + help=f"模型别名，默认 {DEFAULT_MODEL}") + parser.add_argument("--prompt", required=True, help="图像生成提示词（必填）") + parser.add_argument("--image-path", nargs="+", default=[], metavar="PATH", + help="本地图片路径列表（nano-pro / nano-2 图生图，支持多张）") + parser.add_argument("--image-list", nargs="+", default=[], metavar="URL", + help="原图 URL 列表（sora 图生图，支持多张）") + parser.add_argument("--n", type=int, default=1, help="生成数量，>=1（sora，默认 1）") + parser.add_argument("--size", default="2K", help="图片尺寸，如 2K / 4K / 1024x1024（默认 2K）") + parser.add_argument("--quality", default="hd", help="图片质量 standard/hd（seedream，默认 hd）") + parser.add_argument("--ratio", default="2:3", help="图片比例（sora 文生图，默认 2:3）") + parser.add_argument("--aspect-ratio", default="16:9", help="宽高比（gemini，默认 16:9）") + args = parser.parse_args() + + model = args.model.lower() + prompt = args.prompt + + print(f"⏰ {ts()} POPI_OPENAPI_URL={POPI_OPENAPI_URL}") + print(f"提示词: {prompt}\n") + + ok = False + if model == "seedream": + ok = gen_seedream_text(prompt, size=args.size, quality=args.quality, n=args.n) + elif model == "nano-pro": + if args.image_path: + ok = gen_gemini_image(args.image_path, prompt, "gemini-3-pro-image-preview", + aspect_ratio=args.aspect_ratio, size=args.size) + else: + ok = gen_gemini_text(prompt, "gemini-3-pro-image-preview", + aspect_ratio=args.aspect_ratio, size=args.size) + elif model == "nano-2": + if args.image_path: + ok = gen_gemini_image(args.image_path, prompt, "gemini-3.1-flash-image-preview", + aspect_ratio=args.aspect_ratio, size=args.size) + else: + ok = gen_gemini_text(prompt, "gemini-3.1-flash-image-preview", + aspect_ratio=args.aspect_ratio, size=args.size) + elif model == "sora": + ok = gen_sora(prompt, ratio=args.ratio, image_list=args.image_list, n=args.n) + + print(f"\n{'✅ 完成' if ok else '❌ 失败'} {ts()}") + sys.exit(0 if ok else 1) + + +if __name__ == "__main__": + main() diff --git a/image_generation/popiskill-image-real2anime/scripts/tmp/1.jpg b/image_generation/popiskill-image-real2anime/scripts/tmp/1.jpg new file mode 100644 index 0000000..0bc1745 Binary files /dev/null and b/image_generation/popiskill-image-real2anime/scripts/tmp/1.jpg differ diff --git a/image_generation/popiskill-image-real2anime/style.json b/image_generation/popiskill-image-real2anime/style.json new file mode 100644 index 0000000..c3957e4 --- /dev/null +++ b/image_generation/popiskill-image-real2anime/style.json @@ -0,0 +1,70 @@ +{ + "jojo漫画风格": { + "prompt": "将参考图中的角色转化为《JOJO的奇妙冒险》漫画风格。保持构图和核心特征", + "base_img": "" + }, + "水彩风格": { + "prompt": "Transform the characters in the reference image into《Ghibli》style: soft watercolor edges, gentle pastel washes, clean white background, high contrast, muted palette with selective color pops, warm yet whimsical Studio-Ghibli vibe. Masterpiece quality, intricate hand-painted detail", + "base_img": "" + }, + "迪士尼3D动画风格": { + "prompt": "将参考图中的角色转变为具有《迪士尼3d动画》质感的动画风格，背景变为白色背景", + "base_img": "" + }, + "80年代复古漫画风格": { + "prompt": "将参考图中的角色转变为具有《80s Anime》质感的日本动画风格。背景变为白色背景", + "base_img": "" + }, + "像素风格": { + "prompt": "将参考图中的角色转变为像素风格，背景变为白色背景", + "base_img": "" + }, + "日系漫画风格": { + "prompt": "将参考图中的角色转变为动画风格，背景变为白色背景", + "base_img": "" + }, + "扁平插画风格": { + "prompt": "将参考图中的角色转变为Flat Design（扁平设计）风格，背景变为白色背景", + "base_img": "" + }, + "黑白漫画风格": { + "prompt": "将参考图中的角色转变为黑白漫画风格，背景变为白色背景", + "base_img": "" + }, + "2.5D游戏风格": { + "prompt": "将参考图中的角色转变为具有《崩坏：星穹铁道》质感的 Stylized 3D 风格。", + "base_img": "" + }, + "饥荒风格": { + "prompt": "Transform the characters in the reference image into gothic cartoon style, in the signature “Don't Starve” game aesthetic: shaky dark-ink outlines, textured paper background, The character's pupils constricted.high contrast, muted palette with selective color pops, whimsical yet macabre vibe. Masterpiece quality, intricate details.", + "base_img": "" + }, + "水墨国风风格": { + "prompt": "Transform the characters in the reference image into Guofeng ink style: flowing xuan-paper brush lines, wet-on-wet ink gradations, blank-leave negative space, high contrast, muted palette with selective color pops, serene yet mysterious Eastern vibe. Masterpiece quality, intricate ink details.", + "base_img": "" + }, + "彩铅风格": { + "prompt": "According to the reference image, transform this character into a colored pencil style: delicate paper fiber texture, soft overlapping colored pencil strokes, slightly rough edges of the blank spaces, high contrast, low-saturation main colors with partial contrasting colors, creating a warm yet slightly naive fairy-tale atmosphere.Masterpiece quality， intricate layering details.", + "base_img": "" + }, + "毛绒娃娃风格": { + "prompt": "Convert the character in the reference image into《Plush Toy》style: soft velour seams, plush-fabric pile texture, matte stitch outlines, clean white background, high contrast, muted palette with selective color pops, cuddly yet whimsical stuffed-animal vibe. Masterpiece quality, intricate embroidery details. Adopt the above look for the character in the provided reference image.背景变为白色背景", + "base_img": "" + }, + "卡通INS风格": { + "prompt": "A将参考图中的角色转变为具有Q版风格。保持参考图1的构图和核心特征，背景变为白色，风格参考图2（Q版风格）", + "base_img": "https://popitest-public-1313913486.cos.ap-guangzhou.myqcloud.com/media/2026/0209/2468.png" + }, + "Q版萌趣风格": { + "prompt": "A将参考图中的角色转变为具有Q版风格。保持参考图1的构图和核心特征，背景变为白色，风格参考图2（Q版风格）", + "base_img": "https://popitest-public-1313913486.cos.ap-guangzhou.myqcloud.com/media/2026/0209/2469.png" + }, + "恶搞之家风格": { + "prompt": "将参考图中的角色转变为具有《恶搞之家》动画风格。保持参考图1的构图和核心特征，背景变为白色，风格参考图2（《恶搞之家》动画风格）", + "base_img": "https://popitest-public-1313913486.cos.ap-guangzhou.myqcloud.com/media/2026/0209/2470.jpg" + }, + "韩系漫画风格": { + "prompt": "将参考图中的角色转变为具有现代韩漫风格。保持参考图1的构图和核心特征，背景变为白色，风格参考图2（现代韩漫风格）", + "base_img": "https://popitest-public-1313913486.cos.ap-guangzhou.myqcloud.com/media/2026/0209/2478.jpg" + } +} \ No newline at end of file