Skip to content

ReShot — copy the shot, not the actors. Video → depth map, OpenPose skeleton or canny lines for Seedance 2.0/2.5 reference video, MiniMax H3 Fun ControlNet and Wan VACE. Apache-2.0. Open-sourced by Maosika 猫斯卡 (www.maosika.com), the AI short-drama production system.

0.4.0: local web page — reshot with no arguments opens a Studio Air UI (drop clip → target → quality → depth map, side-by-side preview, download, prompt); stdlib only

wangfeng1a14bbf31761
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 3e55592..20b5b22 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,12 @@
# Changelog
+## 0.4.0 — 2026-09-12
+- **Web page.** `reshot` with no arguments (or `reshot web`) starts a local UI in the browser:
+ drop a clip, pick the target model and quality, run, preview reference and depth side by
+ side, download, copy the prompt line for Seedance / MiniMax H3. Studio Air look (the
+ Maosika design system), Chinese and English. Standard library only — no new dependencies;
+ listens on 127.0.0.1; results in `~/ReShot`. Verified in Chrome with the real model.
+
## 0.3.8 — 2026-09-12
- Fix: on Apple Silicon with torch ≥ 2.14 the model failed to load with
`invalid low watermark ratio 1.4` — the MPS low watermark is now set alongside the high one.
diff --git a/README.md b/README.md
index b0e545f..8ba0611 100644
--- a/README.md
+++ b/README.md
@@ -57,6 +57,15 @@ on a short clip with --metrics and show me the numbers. Don't say it's done unti
### 2. Make the depth map
+**Easiest — the web page.** Type `reshot` with nothing after it: a page opens in your browser (everything stays on your computer). Drop the clip in, pick which model it is for, click **Make depth map**. You get the reference and the depth map side by side, the numbers, a download button, and the prompt line to paste into Seedance or MiniMax H3.
+
+```bash
+reshot # opens http://127.0.0.1:8765 — results land in ~/ReShot
+reshot web --port 9000 --out ./depth --no-browser # options, if you want them
+```
+
+**Or the command line**, for scripts and folders:
+
```bash
reshot reference.mp4 -o depth.mp4 --target seedance
```
@@ -129,6 +138,8 @@ Things we learned making the demo:
| `fast` (default) | 644×364 · 364×644 · 490×364 | ~3 GB | 34 ms/frame on a 3080 Ti | large shapes identical to full; a loose strand of hair merges into the cheek |
| `full` | 924×518 · 518×924 · 686×518 | ~11 GB | 83 ms/frame on a 3080 Ti | sharper fine silhouettes |
+The web page has the same two choices under **Quality**; the command line takes `--quality`.
+
We recommend `fast`: for copying blocking and camera it is all a video model needs — the demo takes were made from a depth map shrunk to 320×176 before it even reached MiniMax H3, far below either setting. Measured full against fast on the same 294-frame clip: mean difference 5.3 of 255 grey levels, 95 % of pixels within 15, edge energy −4.5 %. Use `full` for close-ups where thin silhouettes matter and you have the VRAM. The run prints the exact resolution it uses (`model fast: the model sees 644x364`) and records it in `--metrics`. Experts can set any short side with `--input-size`.
Verified on rented cards on 2026-09-12; the numbers are in the `--metrics` output of those runs.
diff --git a/README.zh-CN.md b/README.zh-CN.md
index c43d91a..6b94874 100644
--- a/README.zh-CN.md
+++ b/README.zh-CN.md
@@ -57,6 +57,15 @@ depth.mp4 没生成出来之前不要说完成。
### 2. 抽深度图
+**最省事——网页。** 只敲 `reshot`,后面什么都不加,浏览器会打开一个页面(所有处理都在你自己电脑上)。把片子拖进去,选给哪家模型用,点**生成深度图**。参考片和深度图并排预览,数字、下载按钮、要粘到 Seedance 或 MiniMax H3 的提示词都在同一屏。
+
+```bash
+reshot # 打开 http://127.0.0.1:8765,结果存到 ~/ReShot
+reshot web --port 9000 --out ./depth --no-browser # 想改端口 / 目录 / 不自动开浏览器
+```
+
+**或者命令行**,适合脚本和批量:
+
```bash
reshot 参考片.mp4 -o 深度图.mp4 --target seedance
```
@@ -129,6 +138,8 @@ is transferred onto <Subject 1> and <Subject 2>; its grey depth look is not tran
| `fast`(默认) | 644×364 · 364×644 · 490×364 | 约 3 GB | 3080 Ti 上 34 ms/帧 | 大形状与 full 一模一样;脸颊旁一缕散发会糊进脸里 |
| `full` | 924×518 · 518×924 · 686×518 | 约 11 GB | 3080 Ti 上 83 ms/帧 | 细轮廓更锐 |
+网页上「画质」就是这两档;命令行用 `--quality`。
+
我们推荐 `fast`:复制走位和运镜,视频模型要的它全有——演示里的三段片,深度图交给 MiniMax H3 之前还缩到了 320×176,比这两档都小得多。同一段 294 帧的片上 full 对 fast 实测:平均差 5.3 个灰阶(满量程 255),95% 的像素差在 15 以内,边缘锐度低 4.5%。特写里在意细轮廓、显存又够,再用 `full`。运行时会打印实际用的分辨率(`model fast: the model sees 644x364`),也写进 `--metrics`。要指定任意短边用 `--input-size`。
2026-09-12 在租来的显卡上实测;数字都在那几次运行的 `--metrics` 输出里。
diff --git a/demo/requirements.txt b/demo/requirements.txt
index 08e576d..ed684df 100644
--- a/demo/requirements.txt
+++ b/demo/requirements.txt
@@ -1,5 +1,5 @@
--extra-index-url https://download.pytorch.org/whl/cpu
torch
torchvision
-reshot[ffmpeg]==0.3.8
+reshot[ffmpeg]==0.4.0
gradio>=4.0
diff --git a/docs/USAGE.md b/docs/USAGE.md
index 6e7ab7a..4641c6c 100644
--- a/docs/USAGE.md
+++ b/docs/USAGE.md
@@ -29,6 +29,23 @@ Model weights (111 MB) download from Hugging Face on first run into `~/.cache/hu
Offline machines: copy that cache, or pass `--checkpoint /path/video_depth_anything_vits.pth`.
In China: `export HF_ENDPOINT=https://hf-mirror.com`.
+## Web page
+
+`reshot` with no arguments starts a local web page and opens it in your browser. It is the
+same pipeline behind a form: drop a clip, choose the target model and quality, run, preview
+reference and depth side by side, download. Nothing leaves your machine; the server listens on
+127.0.0.1 only. Results are saved to `~/ReShot` with `<clip>_depth_<target>.mp4` names (never
+overwriting an earlier result) and a `.json` of metrics next to each.
+
+```bash
+reshot # http://127.0.0.1:8765 (next free port if taken)
+reshot web --port 9000 --out ./depth # choose port and folder
+reshot web --no-browser # print the URL only, e.g. over SSH
+```
+
+The model loads once when the first job runs and stays loaded; later runs start immediately.
+Jobs queue one after another (one GPU).
+
## Command line
```
diff --git a/docs/USAGE.zh-CN.md b/docs/USAGE.zh-CN.md
index da29220..884ee90 100644
--- a/docs/USAGE.zh-CN.md
+++ b/docs/USAGE.zh-CN.md
@@ -28,6 +28,20 @@ PyTorch 不锁 CUDA 版本;如果 `pip` 装错了,先按 https://pytorch.org
离线机器:把这个缓存目录拷过去,或用 `--checkpoint /路径/video_depth_anything_vits.pth`。
国内:`export HF_ENDPOINT=https://hf-mirror.com`。
+## 网页
+
+只敲 `reshot` 不带参数,会在本机起一个网页并用浏览器打开。它和命令行是同一条流水线,只是套了个表单:
+拖入片子、选给哪家模型用和画质、运行、参考片与深度图并排预览、下载。什么都不会离开你的电脑,服务只监听
+127.0.0.1。结果存在 `~/ReShot`,文件名 `<片名>_depth_<目标>.mp4`(不会覆盖之前的结果),旁边有一份同名 `.json` 指标。
+
+```bash
+reshot # http://127.0.0.1:8765(被占就用下一个空闲端口)
+reshot web --port 9000 --out ./depth # 指定端口和目录
+reshot web --no-browser # 只打印地址不开浏览器,比如通过 SSH 用
+```
+
+模型在第一次运行时加载一次,之后常驻;后面的任务立刻开始。任务排队一个个跑(一块显卡)。
+
## 命令行
```
diff --git a/pyproject.toml b/pyproject.toml
index 1fdad20..ce7dec9 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -56,6 +56,7 @@ include = ["reshot*"]
[tool.setuptools.package-data]
"reshot.third_party.video_depth_anything" = ["LICENSE"]
"reshot" = ["py.typed"]
+"reshot.web" = ["index.html"]
[tool.ruff]
line-length = 120
diff --git a/reshot/_version.py b/reshot/_version.py
index 4ad67eb..6a9beea 100644
--- a/reshot/_version.py
+++ b/reshot/_version.py
@@ -1 +1 @@
-__version__ = "0.3.8"
+__version__ = "0.4.0"
diff --git a/reshot/cli.py b/reshot/cli.py
index a030cbf..dd602e8 100644
--- a/reshot/cli.py
+++ b/reshot/cli.py
@@ -4,6 +4,8 @@
reshot in.mp4 -o out.mp4 --target h3 # 24 fps, ×32 dims, ≤15 s for MiniMax H3
reshot in.mp4 -o out.mp4 --npz d.npz --metrics run.json
reshot clips/*.mp4 -o depth/ --target seedance # batch: one model load, out/<name>_depth.mp4
+ reshot # no arguments: local web UI, opens in the browser
+ reshot web --port 8765 --no-browser --out ~/ReShot
Batch mode (several inputs, or `-o` naming a directory): `--npz` / `--metrics`, if given,
are directories too and get `<name>.npz` / `<name>.json` per clip. One bad clip is
@@ -141,7 +143,30 @@ def config_from_args(args: argparse.Namespace) -> RunConfig:
return configs_from_args(args)[0]
+def _web_parser() -> argparse.ArgumentParser:
+ p = argparse.ArgumentParser(
+ prog="reshot web", description="Local web UI for ReShot (everything stays on this machine)."
+ )
+ p.add_argument(
+ "--port", type=int, default=8765, help="port to listen on (default 8765; the next free one is used if taken)"
+ )
+ p.add_argument("--out", type=Path, default=None, help="where depth videos are saved (default ~/ReShot)")
+ p.add_argument("--no-browser", action="store_true", help="don't open the browser automatically")
+ p.add_argument("-v", "--verbose", action="store_true")
+ return p
+
+
def main(argv: list[str] | None = None) -> int:
+ argv = list(sys.argv[1:] if argv is None else argv)
+ # `reshot` alone, or `reshot web …`: the web UI. Anything else is the classic CLI.
+ if not argv or argv[0] == "web":
+ wargs = _web_parser().parse_args(argv[1:] if argv else [])
+ logging.basicConfig(
+ level=logging.INFO if wargs.verbose else logging.WARNING, format="%(levelname)s %(name)s: %(message)s"
+ )
+ from .web import serve
+
+ return serve(out_dir=wargs.out, port=wargs.port, open_browser=not wargs.no_browser)
args = build_parser().parse_args(argv)
logging.basicConfig(
level=logging.INFO if args.verbose else logging.WARNING, format="%(levelname)s %(name)s: %(message)s"
diff --git a/reshot/web/__init__.py b/reshot/web/__init__.py
new file mode 100644
index 0000000..f90f0a2
--- /dev/null
+++ b/reshot/web/__init__.py
@@ -0,0 +1,386 @@
+"""The local web UI: `reshot` with no arguments (or `reshot web`) starts it and opens a browser.
+
+One HTML page (``index.html`` next to this file) talks to a small JSON API served by the
+standard library — no framework, no extra dependency, so ``pip install reshot`` is all a
+user needs. Everything stays on the user's machine: uploads go to a temp folder, results to
+``~/ReShot`` (or ``--out``).
+
+API
+ POST /api/upload multipart "file" → source info (id, size, fps, frames, seconds)
+ GET /api/source/<id> the uploaded clip, for the preview player
+ POST /api/jobs JSON options → {"id"}; runs in the single worker thread
+ GET /api/jobs/<id> status, progress steps, result / error
+ GET /api/output/<id> the depth video (``?download=1`` for an attachment)
+ GET /api/info version, device, VRAM, output folder
+
+Why one worker thread with a cached model: the depth model is loaded once per server
+process (3–6 s) and every job reuses it; two jobs at once would fight for the GPU, so they
+queue. Why polling instead of SSE: 500 ms polls are plenty for a 20-second job and keep
+the server a plain ``http.server``.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import mimetypes
+import queue
+import socket
+import sys
+import threading
+import time
+import traceback
+import uuid
+import webbrowser
+from email.parser import BytesParser
+from email.policy import HTTP
+from http import HTTPStatus
+from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
+from pathlib import Path
+from urllib.parse import parse_qs, urlparse
+
+from .._version import __version__
+from ..config import RunConfig
+from ..errors import ReshotError
+from ..io import probe_video
+from ..pipeline import _cuda_total_bytes, _make_backend, _run_one
+from ..targets import TARGETS
+
+log = logging.getLogger("reshot.web")
+
+MAX_UPLOAD_BYTES = 500 * 2**20
+DEFAULT_PORT = 8765
+
+
+class _Job:
+ """One depth run: options in, progress steps and a result (or error) out."""
+
+ def __init__(self, cfg: RunConfig, source_name: str) -> None:
+ self.id = uuid.uuid4().hex[:12]
+ self.cfg = cfg
+ self.source_name = source_name
+ self.status = "queued" # queued | running | done | error
+ self.steps: list[dict] = []
+ self.result: dict | None = None
+ self.error: str | None = None
+ self.created = time.time()
+ self.started: float | None = None
+ self.finished: float | None = None
+
+ # Reporter protocol — the pipeline calls this for every stage.
+ def step(self, tag: str, message: str) -> None:
+ self.steps.append({"tag": tag, "message": message, "t": round(time.time() - (self.started or time.time()), 1)})
+
+ def to_dict(self) -> dict:
+ return {
+ "id": self.id,
+ "status": self.status,
+ "source_name": self.source_name,
+ "target": self.cfg.target,
+ "quality": self.cfg.quality,
+ "steps": self.steps,
+ "result": self.result,
+ "error": self.error,
+ "elapsed": round((self.finished or time.time()) - (self.started or self.created), 1),
+ }
+
+
+class _State:
+ """Everything the handlers share. One instance per server."""
+
+ def __init__(self, out_dir: Path, tmp_dir: Path) -> None:
+ self.out_dir = out_dir
+ self.tmp_dir = tmp_dir
+ self.sources: dict[str, dict] = {}
+ self.jobs: dict[str, _Job] = {}
+ self.queue: queue.Queue[_Job] = queue.Queue()
+ self.backend = None
+ self.backend_name = ""
+ self.lock = threading.Lock()
+ self.device = "?"
+
+ def worker(self) -> None:
+ while True:
+ job = self.queue.get()
+ job.status = "running"
+ job.started = time.time()
+ try:
+ if self.backend is None:
+ job.step("model", "loading the depth model (once per session)")
+ self.backend, self.backend_name = _make_backend(job.cfg)
+ self.device = self.backend.device
+ res = _run_one(job.cfg, job, self.backend, self.backend_name)
+ job.result = {
+ "output": str(res.output),
+ "output_url": f"/api/output/{job.id}",
+ "frames": res.frames,
+ "fps": res.fps,
+ "width": res.width,
+ "height": res.height,
+ "device": res.device,
+ "precision": res.precision,
+ "ms_per_frame": round(res.ms_per_frame, 1),
+ "depth_seconds": round(res.depth_seconds, 1),
+ "total_seconds": round(res.total_seconds, 1),
+ "model_input_resolution": res.metrics.get("model_input_resolution"),
+ "gpu_peak_reserved_bytes": res.metrics.get("gpu_peak_reserved_bytes"),
+ "output_bytes": res.metrics.get("output_bytes"),
+ }
+ job.status = "done"
+ except ReshotError as exc:
+ job.error = str(exc)
+ job.status = "error"
+ except Exception as exc: # a bug, not a user problem — show it rather than hang
+ job.error = f"{type(exc).__name__}: {exc}"
+ job.status = "error"
+ log.error("job %s crashed:\n%s", job.id, traceback.format_exc())
+ finally:
+ job.finished = time.time()
+ self.queue.task_done()
+
+
+def _safe_stem(name: str) -> str:
+ stem = Path(name).stem or "clip"
+ return "".join(c if c.isalnum() or c in "-_ ." else "_" for c in stem)[:80].strip() or "clip"
+
+
+class _Handler(BaseHTTPRequestHandler):
+ state: _State # set by serve()
+ index_html: bytes
+
+ # ---- plumbing -------------------------------------------------------------
+ def log_message(self, fmt, *args): # quiet unless -v
+ log.debug("%s " + fmt, self.address_string(), *args)
+
+ def _json(self, obj, status=HTTPStatus.OK) -> None:
+ body = json.dumps(obj, ensure_ascii=False).encode()
+ self.send_response(status)
+ self.send_header("Content-Type", "application/json; charset=utf-8")
+ self.send_header("Content-Length", str(len(body)))
+ self.send_header("Cache-Control", "no-store")
+ self.end_headers()
+ self.wfile.write(body)
+
+ def _file(self, path: Path, download_name: str | None = None) -> None:
+ if not path.is_file():
+ return self._json({"error": "not found"}, HTTPStatus.NOT_FOUND)
+ size = path.stat().st_size
+ ctype = mimetypes.guess_type(str(path))[0] or "application/octet-stream"
+ # Range support so the <video> element can seek.
+ start, end = 0, size - 1
+ rng = self.headers.get("Range")
+ status = HTTPStatus.OK
+ if rng and rng.startswith("bytes="):
+ a, _, b = rng[6:].partition("-")
+ start = int(a) if a else max(0, size - int(b))
+ end = int(b) if b and a else size - 1
+ end = min(end, size - 1)
+ status = HTTPStatus.PARTIAL_CONTENT
+ self.send_response(status)
+ self.send_header("Content-Type", ctype)
+ self.send_header("Accept-Ranges", "bytes")
+ self.send_header("Content-Length", str(end - start + 1))
+ if status == HTTPStatus.PARTIAL_CONTENT:
+ self.send_header("Content-Range", f"bytes {start}-{end}/{size}")
+ if download_name:
+ self.send_header("Content-Disposition", f'attachment; filename="{download_name}"')
+ self.end_headers()
+ with open(path, "rb") as fh:
+ fh.seek(start)
+ remaining = end - start + 1
+ while remaining > 0:
+ chunk = fh.read(min(1 << 20, remaining))
+ if not chunk:
+ break
+ self.wfile.write(chunk)
+ remaining -= len(chunk)
+
+ # ---- routes -----------------------------------------------------------------
+ def do_GET(self) -> None:
+ url = urlparse(self.path)
+ parts = [p for p in url.path.split("/") if p]
+ try:
+ if not parts or parts == ["index.html"]:
+ self.send_response(HTTPStatus.OK)
+ self.send_header("Content-Type", "text/html; charset=utf-8")
+ self.send_header("Content-Length", str(len(self.index_html)))
+ self.send_header("Cache-Control", "no-store")
+ self.end_headers()
+ self.wfile.write(self.index_html)
+ elif parts == ["api", "info"]:
+ self._json(self._info())
+ elif parts[:2] == ["api", "source"] and len(parts) == 3:
+ src = self.state.sources.get(parts[2])
+ self._file(Path(src["path"])) if src else self._json({"error": "unknown source"}, HTTPStatus.NOT_FOUND)
+ elif parts[:2] == ["api", "jobs"] and len(parts) == 3:
+ job = self.state.jobs.get(parts[2])
+ self._json(job.to_dict()) if job else self._json({"error": "unknown job"}, HTTPStatus.NOT_FOUND)
+ elif parts == ["api", "jobs"]:
+ self._json([j.to_dict() for j in sorted(self.state.jobs.values(), key=lambda j: j.created)])
+ elif parts[:2] == ["api", "output"] and len(parts) == 3:
+ job = self.state.jobs.get(parts[2])
+ if not job or not job.result:
+ return self._json({"error": "no output yet"}, HTTPStatus.NOT_FOUND)
+ q = parse_qs(url.query)
+ self._file(Path(job.result["output"]), Path(job.result["output"]).name if q.get("download") else None)
+ else:
+ self._json({"error": "not found"}, HTTPStatus.NOT_FOUND)
+ except (BrokenPipeError, ConnectionResetError):
+ pass # the browser closed the video stream; normal
+
+ def do_POST(self) -> None:
+ parts = [p for p in urlparse(self.path).path.split("/") if p]
+ if parts == ["api", "upload"]:
+ return self._upload()
+ if parts == ["api", "jobs"]:
+ return self._create_job()
+ self._json({"error": "not found"}, HTTPStatus.NOT_FOUND)
+
+ def _info(self) -> dict:
+ st = self.state
+ vram = _cuda_total_bytes("cuda") if st.device == "cuda" else 0
+ return {
+ "version": __version__,
+ "device": st.device,
+ "vram_bytes": vram,
+ "out_dir": str(st.out_dir),
+ "targets": {
+ k: {"fps": v.fps, "multiple": v.multiple, "max_seconds": v.max_seconds} for k, v in TARGETS.items()
+ },
+ "model_loaded": st.backend is not None,
+ }
+
+ def _upload(self) -> None:
+ length = int(self.headers.get("Content-Length") or 0)
+ if length <= 0 or length > MAX_UPLOAD_BYTES:
+ return self._json(
+ {"error": f"upload must be 1 byte – {MAX_UPLOAD_BYTES // 2**20} MB"}, HTTPStatus.BAD_REQUEST
+ )
+ ctype = self.headers.get("Content-Type", "")
+ if "multipart/form-data" not in ctype:
+ return self._json({"error": "expected multipart/form-data"}, HTTPStatus.BAD_REQUEST)
+ raw = self.rfile.read(length)
+ # The email parser handles multipart/form-data correctly and is in the stdlib.
+ msg = BytesParser(policy=HTTP).parsebytes(b"Content-Type: " + ctype.encode() + b"\r\n\r\n" + raw)
+ part = next((p for p in msg.iter_parts() if p.get_param("name", header="content-disposition") == "file"), None)
+ if part is None:
+ return self._json({"error": "no 'file' field"}, HTTPStatus.BAD_REQUEST)
+ filename = part.get_filename() or "clip.mp4"
+ data = part.get_payload(decode=True)
+ sid = uuid.uuid4().hex[:12]
+ path = self.state.tmp_dir / f"{sid}_{_safe_stem(filename)}{Path(filename).suffix.lower() or '.mp4'}"
+ path.write_bytes(data)
+ try:
+ w, h, fps, n = probe_video(path)
+ except Exception as exc:
+ path.unlink(missing_ok=True)
+ return self._json({"error": f"cannot read that file as a video ({exc})"}, HTTPStatus.BAD_REQUEST)
+ if w <= 0 or h <= 0:
+ path.unlink(missing_ok=True)
+ return self._json({"error": "that file has no video stream"}, HTTPStatus.BAD_REQUEST)
+ info = {
+ "id": sid,
+ "name": filename,
+ "path": str(path),
+ "bytes": len(data),
+ "width": w,
+ "height": h,
+ "fps": round(fps, 3),
+ "frames": n,
+ "seconds": round(n / fps, 2) if fps else 0,
+ "url": f"/api/source/{sid}",
+ }
+ self.state.sources[sid] = info
+ self._json({k: v for k, v in info.items() if k != "path"})
+
+ def _create_job(self) -> None:
+ length = int(self.headers.get("Content-Length") or 0)
+ try:
+ body = json.loads(self.rfile.read(length) or b"{}")
+ except json.JSONDecodeError:
+ return self._json({"error": "bad JSON"}, HTTPStatus.BAD_REQUEST)
+ src = self.state.sources.get(str(body.get("source_id", "")))
+ if not src:
+ return self._json({"error": "upload a video first"}, HTTPStatus.BAD_REQUEST)
+ target = str(body.get("target", "seedance"))
+ stem = _safe_stem(src["name"])
+ out = self.state.out_dir / f"{stem}_depth_{target}.mp4"
+ n = 2
+ while out.exists(): # never overwrite an earlier result
+ out = self.state.out_dir / f"{stem}_depth_{target}_{n}.mp4"
+ n += 1
+ try:
+ cfg = RunConfig(
+ input=Path(src["path"]),
+ output=out,
+ target=target,
+ quality=str(body.get("quality", "fast")),
+ max_res=int(body.get("max_side") or 1280) if body.get("max_side") else 1280,
+ max_frames=int(body["max_frames"]) if body.get("max_frames") else None,
+ invert=bool(body.get("invert", False)),
+ clip_percent=float(body.get("clip_percent", 0.0) or 0.0),
+ gamma=float(body.get("gamma", 1.0) or 1.0),
+ metrics=out.with_suffix(".json"),
+ force=bool(body.get("force", False)),
+ )
+ except (ReshotError, ValueError) as exc:
+ return self._json({"error": str(exc)}, HTTPStatus.BAD_REQUEST)
+ job = _Job(cfg, src["name"])
+ self.state.jobs[job.id] = job
+ self.state.queue.put(job)
+ self._json({"id": job.id})
+
+
+def _free_port(preferred: int) -> int:
+ for port in range(preferred, preferred + 20):
+ with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+ try:
+ s.bind(("127.0.0.1", port))
+ return port
+ except OSError:
+ continue
+ return 0
+
+
+def make_server(out_dir: Path, port: int = 0, tmp_dir: Path | None = None) -> ThreadingHTTPServer:
+ """Build (but don't run) the server — used by `serve()` and by the tests."""
+ import tempfile
+
+ out_dir = Path(out_dir).expanduser()
+ out_dir.mkdir(parents=True, exist_ok=True)
+ tmp = Path(tmp_dir) if tmp_dir else Path(tempfile.mkdtemp(prefix="reshot-web-"))
+ tmp.mkdir(parents=True, exist_ok=True)
+ state = _State(out_dir, tmp)
+ threading.Thread(target=state.worker, name="reshot-worker", daemon=True).start()
+
+ handler = type("ReShotHandler", (_Handler,), {})
+ handler.state = state
+ handler.index_html = (Path(__file__).parent / "index.html").read_bytes()
+ server = ThreadingHTTPServer(("127.0.0.1", port), handler)
+ server.daemon_threads = True
+ server.reshot_state = state # type: ignore[attr-defined]
+ return server
+
+
+def serve(out_dir: Path | None = None, port: int = DEFAULT_PORT, open_browser: bool = True) -> int:
+ """Run the UI until Ctrl-C. Returns an exit code."""
+ out = Path(out_dir) if out_dir else Path.home() / "ReShot"
+ chosen = _free_port(port)
+ if not chosen:
+ print(f"reshot: no free port between {port} and {port + 19}", file=sys.stderr)
+ return 1
+ server = make_server(out, chosen)
+ url = f"http://127.0.0.1:{chosen}/"
+ print(f"ReShot {__version__} — open {url} (results go to {out}; Ctrl-C to stop)", file=sys.stderr)
+ if open_browser:
+ threading.Timer(0.4, lambda: webbrowser.open(url)).start()
+ try:
+ server.serve_forever()
+ except KeyboardInterrupt:
+ print("\nreshot: stopped", file=sys.stderr)
+ finally:
+ server.server_close()
+ return 0
+
+
+__all__ = ["DEFAULT_PORT", "make_server", "serve"]
diff --git a/reshot/web/index.html b/reshot/web/index.html
new file mode 100644
index 0000000..3a4610f
--- /dev/null
+++ b/reshot/web/index.html
@@ -0,0 +1,441 @@
+<!doctype html>
+<html lang="zh">
+<head>
+<meta charset="utf-8">
+<meta name="viewport" content="width=device-width,initial-scale=1">
+<title>ReShot · 猫斯卡</title>
+<style>
+/* Studio Air tokens — the Maosika visual system (dark studio, lime accent, system fonts). */
+:root{
+ --bg:#0d0d0f;--surface:#1a1a1e;--surface-2:#28282c;--card:#101216;
+ --text:#eaeaef;--text-2:#adadb8;--text-3:#72727c;
+ --accent:#d8ff42;--accent-bright:#e2f060;--accent-border:rgba(216,255,66,.45);--accent-ink:#2a3208;--active-ink:#0a0b0d;
+ --error:#ff6b6b;--line:rgba(255,255,255,.07);--line-2:rgba(255,255,255,.12);
+ --cinema-bg:#0b0f0f;--cinema-deep:#080b0b;--teal:#5cc4bc;--teal-border:rgba(92,196,188,.32);--teal-soft:rgba(92,196,188,.12);
+ --font:-apple-system,BlinkMacSystemFont,"SF Pro Text","Segoe UI",Roboto,"Helvetica Neue",Arial,"PingFang SC","Hiragino Sans GB","Microsoft YaHei","Noto Sans CJK SC",sans-serif;
+ --mono:ui-monospace,SFMono-Regular,"SF Mono",Menlo,Consolas,monospace;
+ --ease:cubic-bezier(.22,1,.36,1);--radius:22px;--radius-sm:16px;
+}
+*{box-sizing:border-box}
+html,body{margin:0;background:var(--bg);color:var(--text);font:15px/1.55 var(--font);letter-spacing:-.01em;-webkit-font-smoothing:antialiased}
+a{color:var(--accent);text-decoration:none}
+button{font:inherit;letter-spacing:inherit;cursor:pointer}
+[hidden]{display:none!important}
+
+/* header */
+.hdr{height:60px;display:flex;align-items:center;justify-content:space-between;padding:0 28px;border-bottom:.5px solid rgba(255,255,255,.06);background:rgba(13,13,15,.85);backdrop-filter:blur(14px);position:sticky;top:0;z-index:5}
+.logo{display:inline-flex;align-items:center;gap:10px;font-weight:750;font-size:1.25rem;color:var(--text);white-space:nowrap}
+.logo:hover{color:var(--accent-bright)}
+.logo-mark{display:inline-flex;align-items:center;justify-content:center;width:32px;height:32px;border-radius:10px;font-size:1.05rem;font-weight:800;color:var(--accent-ink);background:linear-gradient(160deg,var(--accent-bright),var(--accent));border:.5px solid var(--accent-border);box-shadow:0 2px 10px rgba(216,255,66,.18)}
+.logo-sep{color:var(--text-3);font-weight:500;margin:0 2px}
+.logo-app{color:var(--text)}
+.hdr-right{display:flex;align-items:center;gap:14px;font-size:.875rem;color:var(--text-2)}
+.hdr-right a{color:var(--text-2)}.hdr-right a:hover{color:var(--accent-bright)}
+.lang{border:1px solid var(--line-2);background:transparent;color:var(--text-2);border-radius:999px;padding:4px 11px;font-size:.78rem;font-weight:600}
+.lang:hover{color:var(--text);border-color:rgba(255,255,255,.22)}
+
+/* page */
+.wrap{max-width:1120px;margin:0 auto;padding:26px 28px 60px}
+.intro{margin:6px 0 20px}
+.intro h1{margin:0 0 6px;font-size:1.5rem;font-weight:700;letter-spacing:-.02em}
+.intro p{margin:0;color:var(--text-2);font-size:.95rem;max-width:760px}
+
+/* the one card */
+.card{background:var(--card);border:1px solid var(--line);border-radius:var(--radius);display:grid;grid-template-columns:minmax(0,1.15fr) minmax(360px,.85fr);overflow:hidden}
+.stage{background:var(--cinema-deep);border-right:1px solid var(--line);padding:22px;display:flex;flex-direction:column;gap:14px;min-height:520px}
+.opts{padding:22px 24px 24px;display:flex;flex-direction:column;gap:20px}
+
+/* stage: drop zone / players */
+.drop{flex:1;display:flex;flex-direction:column;align-items:center;justify-content:center;gap:10px;border-radius:18px;background:linear-gradient(180deg,#1c2128,#12151a);border:1px solid var(--line);cursor:pointer;transition:border-color .18s,background .18s;text-align:center;padding:30px}
+.drop:hover,.drop.is-over{border-color:var(--teal-border);background:linear-gradient(180deg,#1e262b,#12181a)}
+.drop .big{font-size:1.05rem;font-weight:650;color:var(--text)}
+.drop .sub{font-size:.85rem;color:var(--text-2)}
+.drop .hint{font-size:.78rem;color:var(--text-3);margin-top:8px}
+.drop .icon{width:44px;height:44px;border-radius:12px;background:var(--teal-soft);border:1px solid var(--teal-border);display:flex;align-items:center;justify-content:center;color:var(--teal);font-size:1.3rem}
+.players{display:grid;grid-template-columns:1fr;gap:12px;align-content:start}
+/* portrait clips sit side by side, landscape clips stack — either way the boxes hug the video */
+.players.two.portrait{grid-template-columns:1fr 1fr}
+.player{position:relative;border-radius:18px;overflow:hidden;background:#000;border:1px solid var(--line)}
+.player video{display:block;width:100%;height:auto;max-height:56vh;object-fit:contain;background:#000}
+.players.two.portrait .player video{max-height:62vh}
+.player .tag{position:absolute;top:10px;left:10px;font-size:.72rem;font-weight:650;letter-spacing:.02em;padding:4px 9px;border-radius:999px;background:rgba(0,0,0,.55);color:var(--text);border:1px solid rgba(255,255,255,.12)}
+.player .tag.depth{background:rgba(92,196,188,.18);border-color:var(--teal-border);color:#dff7f4}
+.meta{display:flex;flex-wrap:wrap;gap:6px 14px;font-size:.8rem;color:var(--text-2);font-family:var(--mono)}
+.meta b{color:var(--text);font-weight:600;font-family:var(--font)}
+.meta .name{font-family:var(--font);color:var(--text);max-width:100%;overflow:hidden;text-overflow:ellipsis;white-space:nowrap}
+.stage-foot{display:flex;align-items:center;justify-content:space-between;gap:10px;flex-wrap:wrap}
+.ghost{border:1px solid var(--line-2);background:transparent;color:var(--text);border-radius:12px;padding:7px 12px;font-size:.82rem;font-weight:600}
+.ghost:hover{border-color:rgba(255,255,255,.22);background:rgba(255,255,255,.04)}
+.ghost.teal{border-color:var(--teal-border);color:#dff7f4;background:var(--teal-soft)}
+
+/* options */
+.opt h3{margin:0 0 8px;font-size:.8rem;font-weight:650;letter-spacing:.02em;color:var(--text-2);text-transform:uppercase}
+.seg{display:flex;background:var(--surface);border:1px solid var(--line);border-radius:18px;padding:4px;gap:2px}
+.seg button{flex:1;border:0;background:transparent;color:var(--text-2);border-radius:13px;padding:9px 6px;font-size:.86rem;font-weight:650;white-space:nowrap;transition:color .15s,background .15s}
+.seg button:hover{color:var(--text)}
+.seg button.on{background:var(--accent);color:var(--active-ink)}
+.seg button:focus-visible{outline:2px solid var(--accent);outline-offset:2px}
+.note{margin:8px 2px 0;font-size:.8rem;color:var(--text-3);line-height:1.5}
+.note b{color:var(--text-2);font-weight:600}
+.warn{margin-top:8px;font-size:.8rem;color:#ffd08a;background:rgba(255,184,108,.08);border:1px solid rgba(255,184,108,.25);border-radius:12px;padding:8px 11px}
+
+details.adv{border:1px solid var(--line);border-radius:var(--radius-sm);background:var(--surface)}
+details.adv summary{cursor:pointer;list-style:none;padding:11px 14px;font-size:.85rem;font-weight:600;color:var(--text-2);display:flex;justify-content:space-between;align-items:center}
+details.adv summary::-webkit-details-marker{display:none}
+details.adv summary::after{content:"+";font-size:1rem;color:var(--text-3)}
+details.adv[open] summary::after{content:"−"}
+.adv-body{padding:2px 14px 14px;display:grid;grid-template-columns:1fr 1fr;gap:12px 16px}
+.field label{display:block;font-size:.78rem;color:var(--text-2);margin-bottom:5px;font-weight:600}
+.field input[type=number]{width:100%;background:var(--surface-2);border:1px solid var(--line);color:var(--text);border-radius:10px;padding:8px 10px;font:inherit;font-size:.9rem}
+.field input[type=number]:focus{outline:none;border-color:var(--accent-border)}
+.field .tip{font-size:.72rem;color:var(--text-3);margin-top:4px;line-height:1.4}
+.check{display:flex;align-items:center;gap:8px;font-size:.85rem;color:var(--text);cursor:pointer;grid-column:1/-1}
+.check input{accent-color:var(--accent);width:16px;height:16px}
+
+.cta{margin-top:auto;border:0;border-radius:14px;padding:14px 18px;font-size:1rem;font-weight:700;color:var(--accent-ink);background:var(--accent);transition:transform .18s var(--ease),box-shadow .18s,background .15s}
+.cta:hover:not(:disabled){background:var(--accent-bright);transform:scale(1.02);box-shadow:0 4px 14px rgba(216,255,66,.32)}
+.cta:active:not(:disabled){transform:scale(.98)}
+.cta:disabled{opacity:.45;cursor:not-allowed}
+
+/* progress + result */
+.prog{border:1px solid var(--line);border-radius:var(--radius-sm);background:var(--surface);padding:12px 14px;font-size:.82rem}
+.prog .row{display:flex;gap:10px;padding:3px 0;color:var(--text-2)}
+.prog .row .t{font-family:var(--mono);color:var(--text-3);width:52px;flex:none}
+.prog .row .k{font-family:var(--mono);color:var(--teal);width:60px;flex:none}
+.prog .row.warn-row .k{color:#ffd08a}
+.prog .spin{display:inline-block;width:12px;height:12px;border:2px solid rgba(216,255,66,.25);border-top-color:var(--accent);border-radius:50%;animation:spin .9s linear infinite;vertical-align:-2px;margin-right:6px}
+@keyframes spin{to{transform:rotate(360deg)}}
+.result{border:1px solid var(--teal-border);background:var(--teal-soft);border-radius:var(--radius-sm);padding:14px}
+.result h4{margin:0 0 8px;font-size:.95rem;font-weight:700;color:#dff7f4}
+.chips{display:flex;flex-wrap:wrap;gap:6px;margin-bottom:12px}
+.chip{font-size:.74rem;font-family:var(--mono);color:var(--text-2);border:1px solid var(--line-2);border-radius:999px;padding:3px 9px}
+.chip b{color:var(--text);font-weight:600}
+.btns{display:flex;gap:8px;flex-wrap:wrap}
+.btn{border-radius:12px;padding:9px 14px;font-size:.85rem;font-weight:650;border:1px solid var(--line-2);background:transparent;color:var(--text)}
+.btn:hover{border-color:rgba(255,255,255,.25);background:rgba(255,255,255,.04)}
+.btn.primary{background:var(--accent);color:var(--accent-ink);border-color:transparent}
+.btn.primary:hover{background:var(--accent-bright)}
+.path{margin-top:10px;font-size:.74rem;color:var(--text-3);font-family:var(--mono);word-break:break-all}
+.next{margin-top:12px;border-top:1px solid rgba(255,255,255,.08);padding-top:10px;font-size:.82rem;color:var(--text-2);line-height:1.55}
+.next code{display:block;margin-top:6px;padding:9px 11px;border-radius:10px;background:rgba(0,0,0,.35);color:var(--text);font-family:var(--font);font-size:.8rem;white-space:pre-wrap}
+.err{border:1px solid rgba(255,107,107,.4);background:rgba(255,107,107,.08);color:#ffb3b3;border-radius:var(--radius-sm);padding:12px 14px;font-size:.85rem;white-space:pre-wrap}
+
+footer{margin-top:26px;display:flex;flex-wrap:wrap;gap:6px 18px;font-size:.8rem;color:var(--text-3)}
+footer a{color:var(--text-2)}footer a:hover{color:var(--accent-bright)}
+
+@media (max-width:900px){
+ .card{grid-template-columns:1fr}
+ .stage{border-right:0;border-bottom:1px solid var(--line);min-height:0}
+ .wrap{padding:18px 16px 40px}.hdr{padding:0 16px}
+ .adv-body{grid-template-columns:1fr}
+ .players.two{grid-template-columns:1fr}
+}
+@media (prefers-reduced-motion:reduce){*{animation:none!important;transition:none!important}}
+</style>
+</head>
+<body>
+<header class="hdr">
+ <a class="logo" href="https://www.maosika.com" target="_blank" rel="noopener" title="www.maosika.com">
+ <span class="logo-mark" aria-hidden="true">M</span>
+ <span data-i18n="brand">猫斯卡</span><span class="logo-sep">·</span><span class="logo-app">ReShot</span>
+ </a>
+ <div class="hdr-right">
+ <a href="https://www.maosika.com" target="_blank" rel="noopener" data-i18n="hdr.maosika">去猫斯卡做 AI 短剧 ↗</a>
+ <a href="https://github.com/maosika-ai/reshot" target="_blank" rel="noopener">GitHub</a>
+ <button class="lang" id="lang" type="button">EN</button>
+ </div>
+</header>
+
+<main class="wrap">
+ <div class="intro">
+ <h1 data-i18n="intro.h1">复制走位,不复制演员。</h1>
+ <p data-i18n="intro.p">把一段参考视频变成深度图视频,交给 Seedance 或 MiniMax H3 当参考,它就会照着这段的动作和运镜再拍一遍——人物换成你的。所有处理都在这台电脑上完成。</p>
+ </div>
+
+ <section class="card">
+ <!-- left: source / result stage -->
+ <div class="stage">
+ <div class="drop" id="drop" role="button" tabindex="0">
+ <div class="icon">▶</div>
+ <div class="big" data-i18n="drop.big">拖入参考视频,或点击选择</div>
+ <div class="sub" data-i18n="drop.sub">mp4 / mov / webm · 15 秒以内效果最好</div>
+ <div class="hint" data-i18n="drop.hint">视频不会上传到任何服务器</div>
+ <input type="file" id="file" accept="video/*,.mp4,.mov,.webm,.mkv" hidden>
+ </div>
+ <div class="players" id="players" hidden>
+ <div class="player" id="srcPlayer"><span class="tag" data-i18n="tag.src">参考片</span><video id="srcVideo" controls muted loop playsinline></video></div>
+ <div class="player" id="outPlayer" hidden><span class="tag depth" data-i18n="tag.depth">深度图</span><video id="outVideo" controls muted loop playsinline></video></div>
+ </div>
+ <div class="meta" id="meta" hidden></div>
+ <div class="stage-foot" id="stageFoot" hidden>
+ <button class="ghost" id="changeBtn" type="button" data-i18n="stage.change">换一段视频</button>
+ <button class="ghost teal" id="syncBtn" type="button" hidden data-i18n="stage.sync">同步播放</button>
+ </div>
+ </div>
+
+ <!-- right: options -->
+ <div class="opts">
+ <div class="opt">
+ <h3 data-i18n="opt.target">给谁用</h3>
+ <div class="seg" id="target">
+ <button type="button" data-v="seedance" class="on">Seedance</button>
+ <button type="button" data-v="h3">MiniMax H3</button>
+ <button type="button" data-v="wan">Wan</button>
+ <button type="button" data-v="none" data-i18n="opt.none">通用</button>
+ </div>
+ <div class="note" id="targetNote"></div>
+ <div class="warn" id="lenWarn" hidden></div>
+ </div>
+
+ <div class="opt">
+ <h3 data-i18n="opt.quality">画质</h3>
+ <div class="seg" id="quality">
+ <button type="button" data-v="fast" class="on" data-i18n="q.fast">快速(推荐)</button>
+ <button type="button" data-v="full" data-i18n="q.full">完整</button>
+ </div>
+ <div class="note" id="qualityNote"></div>
+ </div>
+
+ <details class="adv">
+ <summary data-i18n="adv.title">高级选项</summary>
+ <div class="adv-body">
+ <div class="field"><label for="maxSide" data-i18n="adv.maxSide">输出长边上限</label><input type="number" id="maxSide" min="0" step="2" placeholder="0"><div class="tip" data-i18n="adv.maxSideTip">0 = 原片尺寸。给 MiniMax H3 当参考视频建议 320。</div></div>
+ <div class="field"><label for="maxFrames" data-i18n="adv.maxFrames">只处理前 N 帧</label><input type="number" id="maxFrames" min="1" step="1" placeholder="全部"><div class="tip" data-i18n="adv.maxFramesTip">试效果时用,比如 48。</div></div>
+ <div class="field"><label for="clip" data-i18n="adv.clip">裁掉两端 %</label><input type="number" id="clip" min="0" max="10" step="0.1" value="0"><div class="tip" data-i18n="adv.clipTip">有东西贴着镜头把其他全压黑时用 0.5–1。</div></div>
+ <div class="field"><label for="gamma">Gamma</label><input type="number" id="gamma" min="0.2" max="3" step="0.05" value="1"><div class="tip" data-i18n="adv.gammaTip">&gt;1 压暗中间调,把近处拉开。一般不动。</div></div>
+ <label class="check"><input type="checkbox" id="invert"><span data-i18n="adv.invert">反转:远白近黑(默认近白远黑,Seedance / H3 / ControlNet 都要这个)</span></label>
+ </div>
+ </details>
+
+ <div class="prog" id="prog" hidden></div>
+ <div class="err" id="err" hidden></div>
+ <div class="result" id="result" hidden></div>
+
+ <button class="cta" id="run" type="button" disabled data-i18n="cta">生成深度图</button>
+ </div>
+ </section>
+
+ <footer>
+ <span><span data-i18n="foot.by">出品</span> <a href="https://www.maosika.com" target="_blank" rel="noopener">猫斯卡 · www.maosika.com</a></span>
+ <span data-i18n="foot.what">AI 短剧 / AI 短视频自动生产系统</span>
+ <a href="https://github.com/maosika-ai/reshot" target="_blank" rel="noopener">GitHub</a>
+ <a href="https://github.com/maosika-ai/ComfyUI-ReShot" target="_blank" rel="noopener" data-i18n="foot.comfy">ComfyUI 节点</a>
+ <span id="ver"></span>
+ </footer>
+</main>
+
+<script>
+(() => {
+const I18N = {
+ zh: {
+ brand:"猫斯卡", "hdr.maosika":"去猫斯卡做 AI 短剧 ↗",
+ "intro.h1":"复制走位,不复制演员。",
+ "intro.p":"把一段参考视频变成深度图视频,交给 Seedance 或 MiniMax H3 当参考,它就会照着这段的动作和运镜再拍一遍——人物换成你的。所有处理都在这台电脑上完成。",
+ "drop.big":"拖入参考视频,或点击选择","drop.sub":"mp4 / mov / webm · 15 秒以内效果最好","drop.hint":"视频不会上传到任何服务器",
+ "tag.src":"参考片","tag.depth":"深度图","stage.change":"换一段视频","stage.sync":"同步播放",
+ "opt.target":"给谁用","opt.none":"通用","opt.quality":"画质","q.fast":"快速(推荐)","q.full":"完整",
+ "adv.title":"高级选项","adv.maxSide":"输出长边上限","adv.maxSideTip":"0 = 原片尺寸。给 MiniMax H3 当参考视频建议 320。",
+ "adv.maxFrames":"只处理前 N 帧","adv.maxFramesTip":"试效果时用,比如 48。","adv.clip":"裁掉两端 %","adv.clipTip":"有东西贴着镜头把其他全压黑时用 0.5–1。",
+ "adv.gammaTip":">1 压暗中间调,把近处拉开。一般不动。","adv.invert":"反转:远白近黑(默认近白远黑,Seedance / H3 / ControlNet 都要这个)",
+ cta:"生成深度图", "cta.running":"生成中…", "cta.again":"再生成一次",
+ "foot.by":"出品","foot.what":"AI 短剧 / AI 短视频自动生产系统","foot.comfy":"ComfyUI 节点",
+ t_seedance:"<b>Seedance 2.0 / 2.5 参考视频</b>:24 fps,宽高 16 的倍数,≤ 15 秒。上传后提示词写「参考@视频1的动作与运镜」。",
+ t_h3:"<b>MiniMax H3</b>:24 fps,宽高 32 的倍数,≤ 15 秒。当参考视频挂 &lt;Video 1&gt; 时,高级选项里把长边限到 320 效果最好。",
+ t_wan:"<b>Wan 2.1 VACE</b>:16 fps,宽高 16 的倍数,作为 control_video。(这条我们未端到端验证。)",
+ t_none:"<b>通用</b>:保持原 fps,只把宽高凑成偶数。给任何认深度视频的模型。",
+ q_fast:"模型看 <b>644×364</b>(16:9)/ 364×644(9:16)。约 3 GB 显存,快。复制走位和运镜足够。",
+ q_full:"模型看 <b>924×518</b> / 518×924。约 11 GB 显存,慢 2.5 倍,细轮廓更锐。特写、在意发丝时用。",
+ lenWarn:"这段 {s} 秒,{t} 只收 ≤ {m} 秒的参考视频——先剪短,或在高级选项里限制帧数。",
+ meta_fmt:"{w}×{h} · {fps} fps · {n} 帧 · {s} 秒 · {mb} MB",
+ uploading:"读取视频…", queued:"排队中", loading_model:"加载模型(每次启动只一次)",
+ step_plan:"计划", step_read:"解码", step_model:"模型", step_depth:"深度", step_size:"尺寸", step_wrote:"写出", step_warn:"提醒", step_vram:"显存", step_quality:"画质", step_metrics:"指标", step_npz:"npz", step_note:"注",
+ done:"完成", chips:{frames:"帧",fps:"fps",size:"尺寸",speed:"ms/帧",device:"设备",model:"模型看到",vram:"显存峰值",file:"文件"},
+ download:"下载深度图", openFolder:"保存在", next_title:"下一步",
+ next_seedance:"到 Seedance 把这个文件当「参考视频」上传,提示词开头写:",
+ next_seedance_p:"参考@视频1的动作与运镜,顺序与视频保持一致。\n(接着写你的人物、服装、场景和画风)",
+ next_h3:"在 MiniMax H3 里把它挂成 &lt;Video 1&gt;(定妆图挂 &lt;Picture 1&gt;),提示词里定义它并说明灰色外观不要抄:",
+ next_h3_p:"<Subject 3> is the fight choreography and camera movement shown in <Video 1>, a grey depth map in which near objects are white and far objects are black: (用一句话写清灰片里发生了什么)\n<Subject 3>: attribute_transfer - every action, position, timing and camera move of <Video 1> is transferred onto <Subject 1>; its grey depth look is not transferred.",
+ next_wan:"在 ComfyUI 里接到 WanVaceToVideo 的 control_video。", next_none:"交给任何认深度视频的模型。",
+ copy:"复制提示词", copied:"已复制 ✓", err_title:"没成功",
+ "info.device":"本机:{d}",
+ },
+ en: {
+ brand:"Maosika", "hdr.maosika":"Make AI short drama at Maosika ↗",
+ "intro.h1":"Copy the shot, not the actors.",
+ "intro.p":"Turn a reference video into a depth-map video, give it to Seedance or MiniMax H3 as the reference, and it repeats the moves and camera with your own characters. Everything runs on this computer.",
+ "drop.big":"Drop a reference video, or click to choose","drop.sub":"mp4 / mov / webm · under 15 s works best","drop.hint":"Nothing is uploaded to any server",
+ "tag.src":"Reference","tag.depth":"Depth map","stage.change":"Change video","stage.sync":"Play both",
+ "opt.target":"For which model","opt.none":"Generic","opt.quality":"Quality","q.fast":"Fast (recommended)","q.full":"Full",
+ "adv.title":"Advanced","adv.maxSide":"Max output long side","adv.maxSideTip":"0 = source size. 320 recommended for MiniMax H3 reference videos.",
+ "adv.maxFrames":"Only the first N frames","adv.maxFramesTip":"For a quick test, e.g. 48.","adv.clip":"Clip both ends %","adv.clipTip":"0.5–1 when something at the lens crushes everything else to black.",
+ "adv.gammaTip":">1 darkens mid-tones, spreading the near range. Rarely needed.","adv.invert":"Invert: far = white (default near = white, what Seedance / H3 / ControlNets expect)",
+ cta:"Make depth map", "cta.running":"Working…", "cta.again":"Run again",
+ "foot.by":"Made by","foot.what":"AI short drama & short video production system","foot.comfy":"ComfyUI nodes",
+ t_seedance:"<b>Seedance 2.0 / 2.5 reference video</b>: 24 fps, sides ×16, ≤ 15 s. Prompt with “参考@视频1的动作与运镜”.",
+ t_h3:"<b>MiniMax H3</b>: 24 fps, sides ×32, ≤ 15 s. As a reference video (&lt;Video 1&gt;) it works best small — set max long side 320 in Advanced.",
+ t_wan:"<b>Wan 2.1 VACE</b>: 16 fps, sides ×16, as control_video. (Not verified end to end by us.)",
+ t_none:"<b>Generic</b>: keep the source fps, only make dimensions even. For anything that reads a depth video.",
+ q_fast:"The model sees <b>644×364</b> (16:9) / 364×644 (9:16). ~3 GB VRAM, fast. Enough for blocking and camera.",
+ q_full:"The model sees <b>924×518</b> / 518×924. ~11 GB VRAM, 2.5× slower, sharper fine silhouettes. For close-ups.",
+ lenWarn:"This clip is {s} s; {t} accepts reference videos ≤ {m} s — trim it first, or limit frames in Advanced.",
+ meta_fmt:"{w}×{h} · {fps} fps · {n} frames · {s} s · {mb} MB",
+ uploading:"Reading video…", queued:"Queued", loading_model:"Loading the model (once per session)",
+ step_plan:"plan", step_read:"decode", step_model:"model", step_depth:"depth", step_size:"size", step_wrote:"wrote", step_warn:"note", step_vram:"vram", step_quality:"quality", step_metrics:"metrics", step_npz:"npz", step_note:"note",
+ done:"Done", chips:{frames:"frames",fps:"fps",size:"size",speed:"ms/frame",device:"device",model:"model saw",vram:"VRAM peak",file:"file"},
+ download:"Download depth map", openFolder:"Saved in", next_title:"Next",
+ next_seedance:"Upload this file to Seedance as a reference video and start the prompt with:",
+ next_seedance_p:"参考@视频1的动作与运镜,顺序与视频保持一致。\n(then describe your people, wardrobe, set and look)",
+ next_h3:"In MiniMax H3 attach it as &lt;Video 1&gt; (character sheet as &lt;Picture 1&gt;); define it in the prompt and say the grey look is not to be copied:",
+ next_h3_p:"<Subject 3> is the fight choreography and camera movement shown in <Video 1>, a grey depth map in which near objects are white and far objects are black: (one sentence on what happens in the clip)\n<Subject 3>: attribute_transfer - every action, position, timing and camera move of <Video 1> is transferred onto <Subject 1>; its grey depth look is not transferred.",
+ next_wan:"Connect it to WanVaceToVideo → control_video in ComfyUI.", next_none:"Hand it to any model that reads a depth video.",
+ copy:"Copy prompt", copied:"Copied ✓", err_title:"That didn't work",
+ "info.device":"This machine: {d}",
+ }
+};
+let lang = (localStorage.getItem("reshot.lang") || (navigator.language || "zh").toLowerCase().startsWith("zh") ? "zh" : "en");
+if (localStorage.getItem("reshot.lang")) lang = localStorage.getItem("reshot.lang");
+const t = (k, vars) => { let s = (I18N[lang][k] ?? I18N.zh[k] ?? k); if (typeof s === "string" && vars) for (const [a,b] of Object.entries(vars)) s = s.replaceAll(`{${a}}`, b); return s; };
+const $ = id => document.getElementById(id);
+
+const state = { source:null, target:"seedance", quality:"fast", job:null, info:null, poll:null };
+
+function applyLang() {
+ document.documentElement.lang = lang;
+ document.querySelectorAll("[data-i18n]").forEach(el => { el.textContent = t(el.dataset.i18n); });
+ $("lang").textContent = lang === "zh" ? "EN" : "中文";
+ document.title = lang === "zh" ? "ReShot · 猫斯卡" : "ReShot · Maosika";
+ renderNotes(); renderMeta(); renderRunBtn();
+ if (state.job && state.job.status === "done") renderResult(state.job);
+}
+$("lang").onclick = () => { lang = lang === "zh" ? "en" : "zh"; localStorage.setItem("reshot.lang", lang); applyLang(); };
+
+function renderNotes() {
+ $("targetNote").innerHTML = t("t_" + state.target);
+ $("qualityNote").innerHTML = t("q_" + state.quality);
+ const info = state.info, tg = info && info.targets[state.target];
+ const w = $("lenWarn");
+ if (state.source && tg && tg.max_seconds && state.source.seconds > tg.max_seconds + 0.05 && !$("maxFrames").value) {
+ w.textContent = t("lenWarn", { s: state.source.seconds, t: ({seedance:"Seedance",h3:"MiniMax H3",wan:"Wan"})[state.target] || state.target, m: tg.max_seconds });
+ w.hidden = false;
+ } else w.hidden = true;
+}
+function seg(id, key) {
+ $(id).querySelectorAll("button").forEach(b => b.onclick = () => {
+ $(id).querySelectorAll("button").forEach(x => x.classList.toggle("on", x === b));
+ state[key] = b.dataset.v;
+ if (key === "target" && b.dataset.v === "h3" && !$("maxSide").value) $("maxSide").placeholder = "320";
+ renderNotes();
+ });
+}
+seg("target", "target"); seg("quality", "quality");
+$("maxFrames").addEventListener("input", renderNotes);
+
+// ---- source ----
+const drop = $("drop"), file = $("file");
+drop.onclick = () => file.click();
+drop.onkeydown = e => { if (e.key === "Enter" || e.key === " ") { e.preventDefault(); file.click(); } };
+["dragenter","dragover"].forEach(ev => drop.addEventListener(ev, e => { e.preventDefault(); drop.classList.add("is-over"); }));
+["dragleave","drop"].forEach(ev => drop.addEventListener(ev, e => { e.preventDefault(); drop.classList.remove("is-over"); }));
+drop.addEventListener("drop", e => { const f = e.dataTransfer.files && e.dataTransfer.files[0]; if (f) upload(f); });
+file.onchange = () => { if (file.files[0]) upload(file.files[0]); file.value = ""; };
+$("changeBtn").onclick = () => resetSource();
+document.addEventListener("dragover", e => e.preventDefault());
+document.addEventListener("drop", e => { e.preventDefault(); if (!drop.hidden) return; const f = e.dataTransfer.files && e.dataTransfer.files[0]; if (f) upload(f); });
+
+async function upload(f) {
+ hideErr(); clearResult();
+ drop.querySelector(".big").textContent = t("uploading");
+ const fd = new FormData(); fd.append("file", f, f.name);
+ try {
+ const r = await fetch("/api/upload", { method: "POST", body: fd });
+ const j = await r.json();
+ if (!r.ok) throw new Error(j.error || r.statusText);
+ state.source = j;
+ $("srcVideo").src = j.url; $("srcVideo").play().catch(()=>{});
+ drop.hidden = true; $("players").hidden = false; $("meta").hidden = false; $("stageFoot").hidden = false;
+ $("players").classList.toggle("portrait", j.height > j.width);
+ renderMeta(); renderNotes(); renderRunBtn();
+ } catch (e) { showErr(e.message); drop.querySelector(".big").textContent = t("drop.big"); }
+}
+function resetSource() {
+ state.source = null; clearResult(); hideErr();
+ $("srcVideo").removeAttribute("src"); $("srcVideo").load();
+ drop.hidden = false; $("players").hidden = true; $("meta").hidden = true; $("stageFoot").hidden = true;
+ drop.querySelector(".big").textContent = t("drop.big");
+ renderNotes(); renderRunBtn();
+}
+function renderMeta() {
+ const s = state.source; if (!s) return;
+ $("meta").innerHTML = `<span class="name"><b>${esc(s.name)}</b></span><span>${t("meta_fmt",{w:s.width,h:s.height,fps:s.fps,n:s.frames,s:s.seconds,mb:(s.bytes/1048576).toFixed(1)})}</span>`;
+}
+function renderRunBtn() {
+ const b = $("run"), running = state.job && (state.job.status === "queued" || state.job.status === "running");
+ b.disabled = !state.source || !!running;
+ b.textContent = running ? t("cta.running") : (state.job && state.job.status === "done" ? t("cta.again") : t("cta"));
+}
+
+// ---- run ----
+$("run").onclick = async () => {
+ if (!state.source) return;
+ hideErr(); clearResult();
+ const body = {
+ source_id: state.source.id, target: state.target, quality: state.quality,
+ max_side: +$("maxSide").value || 0, max_frames: +$("maxFrames").value || 0,
+ clip_percent: +$("clip").value || 0, gamma: +$("gamma").value || 1, invert: $("invert").checked,
+ };
+ try {
+ const r = await fetch("/api/jobs", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify(body) });
+ const j = await r.json(); if (!r.ok) throw new Error(j.error || r.statusText);
+ state.job = { id: j.id, status: "queued", steps: [] };
+ renderRunBtn(); renderProg(state.job);
+ clearInterval(state.poll); state.poll = setInterval(poll, 500); poll();
+ } catch (e) { showErr(e.message); }
+};
+async function poll() {
+ if (!state.job) return;
+ const r = await fetch("/api/jobs/" + state.job.id); if (!r.ok) return;
+ const j = await r.json(); state.job = j; renderProg(j);
+ if (j.status === "done") { clearInterval(state.poll); renderResult(j); renderRunBtn(); }
+ else if (j.status === "error") { clearInterval(state.poll); $("prog").hidden = true; showErr(j.error); renderRunBtn(); }
+}
+function renderProg(j) {
+ const p = $("prog"); p.hidden = false;
+ const rows = j.steps.map(s => `<div class="row${s.tag === "warn" ? " warn-row" : ""}"><span class="t">+${s.t}s</span><span class="k">${t("step_" + s.tag)}</span><span>${esc(s.message)}</span></div>`);
+ const head = j.status === "queued" ? `<div class="row"><span class="spin"></span>${t("queued")}</div>`
+ : j.status === "running" ? `<div class="row"><span class="spin"></span>${lang === "zh" ? "处理中" : "Working"} · ${j.elapsed}s</div>` : "";
+ p.innerHTML = head + rows.join("");
+}
+function renderResult(j) {
+ const r = j.result; if (!r) return;
+ $("prog").hidden = true;
+ $("outVideo").src = r.output_url + "?t=" + Date.now(); $("outPlayer").hidden = false; $("players").classList.add("two"); $("syncBtn").hidden = false;
+ const c = t("chips"), chips = [
+ [c.frames, r.frames], [c.fps, r.fps], [c.size, `${r.width}×${r.height}`], [c.speed, r.ms_per_frame], [c.device, `${r.device} ${r.precision}`],
+ r.model_input_resolution ? [c.model, r.model_input_resolution.join("×")] : null,
+ r.gpu_peak_reserved_bytes ? [c.vram, (r.gpu_peak_reserved_bytes / 2**30).toFixed(1) + " GB"] : null,
+ r.output_bytes ? [c.file, (r.output_bytes / 1048576).toFixed(2) + " MB"] : null,
+ ].filter(Boolean).map(([k, v]) => `<span class="chip">${k} <b>${v}</b></span>`).join("");
+ const nextKey = "next_" + j.target, prompt = I18N[lang][nextKey + "_p"];
+ $("result").innerHTML = `
+ <h4>${t("done")} · ${j.elapsed}s</h4>
+ <div class="chips">${chips}</div>
+ <div class="btns"><a class="btn primary" href="${r.output_url}?download=1" download>${t("download")}</a></div>
+ <div class="path">${t("openFolder")} ${esc(r.output)}</div>
+ <div class="next"><b>${t("next_title")}</b> · ${t(nextKey)}${prompt ? `<code id="promptText">${esc(prompt)}</code><div class="btns" style="margin-top:8px"><button class="btn" id="copyBtn" type="button">${t("copy")}</button></div>` : ""}</div>`;
+ $("result").hidden = false;
+ const cb = $("copyBtn"); if (cb) cb.onclick = async () => { await navigator.clipboard.writeText(prompt); cb.textContent = t("copied"); setTimeout(() => cb.textContent = t("copy"), 1600); };
+ $("srcVideo").currentTime = 0; $("outVideo").currentTime = 0;
+}
+$("syncBtn").onclick = () => { const a = $("srcVideo"), b = $("outVideo"); a.currentTime = 0; b.currentTime = 0; a.play(); b.play(); };
+function clearResult() { $("result").hidden = true; $("outPlayer").hidden = true; $("players").classList.remove("two"); $("syncBtn").hidden = true; $("prog").hidden = true; clearInterval(state.poll); if (state.job && state.job.status === "done") state.job = null; }
+function showErr(m) { $("err").textContent = `${t("err_title")}: ${m}`; $("err").hidden = false; }
+function hideErr() { $("err").hidden = true; }
+function esc(s) { return String(s).replace(/[&<>"]/g, c => ({"&":"&amp;","<":"&lt;",">":"&gt;",'"':"&quot;"})[c]); }
+
+// ---- boot ----
+fetch("/api/info").then(r => r.json()).then(i => { state.info = i; $("ver").textContent = `ReShot ${i.version}`; renderNotes(); }).catch(()=>{});
+applyLang();
+})();
+</script>
+</body>
+</html>
diff --git a/tests/test_web.py b/tests/test_web.py
new file mode 100644
index 0000000..4bb3d5e
--- /dev/null
+++ b/tests/test_web.py
@@ -0,0 +1,132 @@
+"""The local web UI's JSON API, end to end on a random port with the fake backend:
+upload → job → poll → download, plus the page itself and the error paths."""
+
+import json
+import os
+import threading
+import time
+import urllib.request
+from pathlib import Path
+
+import cv2
+import numpy as np
+import pytest
+
+from reshot.web import make_server
+
+os.environ["RESHOT_FAKE_BACKEND"] = "1"
+
+
+def _clip(path: Path, frames=12, w=200, h=120, fps=30):
+ vw = cv2.VideoWriter(str(path), cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))
+ for i in range(frames):
+ vw.write(np.full((h, w, 3), (i * 10) % 256, np.uint8))
+ vw.release()
+ return path
+
+
+def server(tmp_path):
+ srv = make_server(tmp_path / "out", port=0, tmp_dir=tmp_path / "tmp")
+ th = threading.Thread(target=srv.serve_forever, daemon=True)
+ th.start()
+ yield f"http://127.0.0.1:{srv.server_address[1]}", srv
+ srv.shutdown()
+ srv.server_close()
+
+
+def _multipart(field, filename, data, ctype="video/mp4"):
+ boundary = "----reshottest"
+ body = (
+ (
+ f'--{boundary}\r\nContent-Disposition: form-data; name="{field}"; filename="{filename}"\r\n'
+ f"Content-Type: {ctype}\r\n\r\n"
+ ).encode()
+ + data
+ + f"\r\n--{boundary}--\r\n".encode()
+ )
+ return body, f"multipart/form-data; boundary={boundary}"
+
+
+def _post(url, body, ctype):
+ req = urllib.request.Request(url, data=body, headers={"Content-Type": ctype}, method="POST")
+ try:
+ with urllib.request.urlopen(req) as r:
+ return r.status, json.loads(r.read())
+ except urllib.error.HTTPError as e:
+ return e.code, json.loads(e.read())
+
+
+def _get(url):
+ with urllib.request.urlopen(url) as r:
+ return r.status, r.read(), dict(r.headers)
+
+
+def test_page_and_info(server):
+ base, _ = server
+ st, body, _ = _get(base + "/")
+ assert st == 200 and b"ReShot" in body and b"maosika.com" in body
+ st, body, _ = _get(base + "/api/info")
+ info = json.loads(body)
+ assert info["targets"]["seedance"]["fps"] == 24 and info["out_dir"].endswith("out")
+
+
+def test_upload_job_download(server, tmp_path):
+ base, _ = server
+ clip = _clip(tmp_path / "ref clip.mp4", frames=30)
+ body, ctype = _multipart("file", "ref clip.mp4", clip.read_bytes())
+ st, src = _post(base + "/api/upload", body, ctype)
+ assert st == 200 and src["width"] == 200 and src["frames"] == 30 and src["seconds"] == 1.0
+ st, raw, _ = _get(base + src["url"])
+ assert st == 200 and len(raw) == clip.stat().st_size
+
+ st, j = _post(
+ base + "/api/jobs",
+ json.dumps({"source_id": src["id"], "target": "h3", "max_side": 160}).encode(),
+ "application/json",
+ )
+ assert st == 200
+ for _ in range(100):
+ _, body, _ = _get(base + f"/api/jobs/{j['id']}")
+ job = json.loads(body)
+ if job["status"] in ("done", "error"):
+ break
+ time.sleep(0.1)
+ assert job["status"] == "done", job
+ assert job["result"]["fps"] == 24 and job["result"]["width"] % 32 == 0 and job["result"]["width"] <= 160
+ assert any(s["tag"] == "wrote" for s in job["steps"])
+ out = Path(job["result"]["output"])
+ assert out.exists() and out.parent == tmp_path / "out" and out.name.startswith("ref clip_depth_h3")
+ st, raw, hdr = _get(base + job["result"]["output_url"] + "?download=1")
+ assert st == 200 and len(raw) == out.stat().st_size and "attachment" in hdr["Content-Disposition"]
+ # a second run must not overwrite the first
+ st, j2 = _post(
+ base + "/api/jobs", json.dumps({"source_id": src["id"], "target": "h3"}).encode(), "application/json"
+ )
+ for _ in range(100):
+ _, body, _ = _get(base + f"/api/jobs/{j2['id']}")
+ if json.loads(body)["status"] in ("done", "error"):
+ break
+ time.sleep(0.1)
+ assert Path(json.loads(body)["result"]["output"]).name == "ref clip_depth_h3_2.mp4"
+
+
+def test_range_request_for_video_seeking(server, tmp_path):
+ base, _ = server
+ clip = _clip(tmp_path / "c.mp4")
+ body, ctype = _multipart("file", "c.mp4", clip.read_bytes())
+ _, src = _post(base + "/api/upload", body, ctype)
+ req = urllib.request.Request(base + src["url"], headers={"Range": "bytes=0-99"})
+ with urllib.request.urlopen(req) as r:
+ assert r.status == 206 and len(r.read()) == 100 and r.headers["Content-Range"].startswith("bytes 0-99/")
+
+
+def test_bad_inputs(server):
+ base, _ = server
+ body, ctype = _multipart("file", "x.mp4", b"not a video")
+ st, j = _post(base + "/api/upload", body, ctype)
+ assert st == 400 and "error" in j
+ st, j = _post(base + "/api/jobs", json.dumps({"source_id": "nope"}).encode(), "application/json")
+ assert st == 400
+ st, j = _post(base + "/api/jobs", b"{", "application/json")
+ assert st == 400