diff --git a/DEPLOY.md b/DEPLOY.md
index 2462dfb..f78405f 100644
--- a/DEPLOY.md
+++ b/DEPLOY.md
@@ -217,6 +217,8 @@ redirect back to `huggingface.co`, which recent `huggingface_hub` rejects with
 | `BGFILTER_CONFIG` | `configs/default.yaml` | Pipeline config file |
 | `BGFILTER_DEVICE` | `cpu` | Overrides **both** model and segmentation device (`cuda` for GPU) |
 | `BGFILTER_MAX_IMAGE_PIXELS` | `4194304` | Reject larger inputs with `413` (~4 MP) |
+| `BGFILTER_FETCH_MAX_BYTES` | `26214400` | `image_url` mode: cap the downloaded body (~25 MB) → `413`. No SSRF filtering — keep auth on / callers trusted. |
+| `BGFILTER_FETCH_TIMEOUT` | `15` | `image_url` mode: read timeout (s) for the fetch (connect fixed at 5 s). Outbound fetch concurrency is fixed at `2 × cpu_workers` (hard cap 64). |
 | `BGFILTER_PRELOAD` | `1` | Load default models at startup (first request isn't cold) |
 | `BGFILTER_CROSS_CHECK` | config (`on`) | Force the cross-check veto on/off (`0` disables; needs `BiRefNet_HR-matting`) |
 | `BGFILTER_CPU_WORKERS` | config `server.cpu_workers` (`1`) | Thread-pool size for request pipelining. `1` = strictly serial (one request at a time). `>1` lets that many requests run concurrently, overlapping their CPU stages while a GPU mutex serializes the model forwards — keeps the GPU fed instead of idle. Biggest win when the pipeline is CPU-bound (e.g. cross-check off). Set it in the config (`server:` section) or override per box with this env var. |
diff --git a/DEPLOY_ZH.md b/DEPLOY_ZH.md
index 5d7cb49..eaef015 100644
--- a/DEPLOY_ZH.md
+++ b/DEPLOY_ZH.md
@@ -200,6 +200,8 @@ export HF_HUB_DISABLE_XET=1                 # 这些 repo 是 Xet 存储；强
 | `BGFILTER_CONFIG` | `configs/default.yaml` | 管线配置文件 |
 | `BGFILTER_DEVICE` | `cpu` | 同时覆盖 model 和 segmentation 的 device（GPU 用 `cuda`）|
 | `BGFILTER_MAX_IMAGE_PIXELS` | `4194304` | 超过则返回 `413`（约 4 MP）|
+| `BGFILTER_FETCH_MAX_BYTES` | `26214400` | `image_url` 模式：下载体上限（约 25 MB）→ `413`。无 SSRF 过滤——务必开鉴权 / 调用方可信。 |
+| `BGFILTER_FETCH_TIMEOUT` | `15` | `image_url` 模式：fetch 读超时（秒；连接超时固定 5 秒）。出站 fetch 并发固定为 `2 × cpu_workers`（硬顶 64）。 |
 | `BGFILTER_PRELOAD` | `1` | 启动时预加载默认模型（首个请求不用冷加载）|
 | `BGFILTER_CROSS_CHECK` | 配置（`on`）| 强制开/关 cross-check 否决（`0` 关闭；需 `BiRefNet_HR-matting`）|
 | `BGFILTER_CPU_WORKERS` | 配置 `server.cpu_workers`（`1`）| 请求流水线的线程池大小。`1` = 严格串行（一次一个请求）。`>1` 允许这么多请求并发,重叠各自的 CPU 段,同时用一把 GPU 互斥锁串行化模型前向——让 GPU 别闲着。CPU 为瓶颈时(如关 cross-check)收益最大。可写进配置的 `server:` 段,或用此 env 变量按机器覆盖。 |
diff --git a/README.md b/README.md
index e6e3279..189d3a5 100644
--- a/README.md
+++ b/README.md
@@ -207,7 +207,8 @@ is expected to authenticate). `/healthz` is always open. See [DEPLOY.md](DEPLOY.
 
 | Field | Required | Values | Meaning |
 | --- | --- | --- | --- |
-| `file` | yes | image file | Source image (field name fixed as `file`). |
+| `file` | one of `file`/`image_url` | image file | Source image as an upload (field name fixed as `file`). |
+| `image_url` | one of `file`/`image_url` | `http(s)://…` | Source image by URL — the service fetches it (bounded by size/time; follows ≤3 redirects). Provide exactly one of `file` or `image_url`. No SSRF filtering: the endpoint trusts authenticated callers, so keep auth on and callers trusted. |
 | `background_mode` | no | `flat` (default) / `complex` | `flat` = solid-colour background (colour-keyed pipeline). `complex` = non-flat/scene background: no colour key, the segmenter alone drives the matte. `screen_color` is ignored in `complex`; expect flat backgrounds to stay stronger on hair detail. |
 | `screen_color` | no | `#RRGGBB`, or omit/empty/`auto` | Background-colour prior; omit to auto-detect from the border. Ignored when `background_mode=complex`. |
 | `seg_model` | no | `birefnet` (default) / `anime-seg` | Segmentation backend. |
@@ -228,9 +229,9 @@ is expected to authenticate). `/healthz` is always open. See [DEPLOY.md](DEPLOY.
 | Code | When |
 | --- | --- |
 | `200` | Success — body is the RGBA PNG. |
-| `400` | Bad input: unreadable image, invalid `screen_color`/`seg_model`/`cross_check`/`background_mode`, or auto-detect failed (pass `screen_color`). |
+| `400` | Bad input: unreadable image, not exactly one of `file`/`image_url`, bad `image_url` / fetch failed, invalid `screen_color`/`seg_model`/`cross_check`/`background_mode`, or auto-detect failed (pass `screen_color`). |
 | `401` | Auth enabled and the token is missing or wrong. |
-| `413` | Image exceeds `BGFILTER_MAX_IMAGE_PIXELS` (~4 MP by default). |
+| `413` | Image exceeds `BGFILTER_MAX_IMAGE_PIXELS` (~4 MP by default), or an `image_url` body exceeds `BGFILTER_FETCH_MAX_BYTES` (~25 MB). |
 | `500` | Inference failed. |
 | `503` | Service still loading models (not ready). |
 
@@ -256,6 +257,13 @@ curl -sS \
   -F "file=@scene.jpg" \
   -F "background_mode=complex" \
   http://127.0.0.1:18083/remove-background -o output.png
+
+# by URL instead of an upload (service fetches it; use image_url OR file, not both)
+curl -sS \
+  -F "image_url=https://example.com/input.png" \
+  -F "screen_color=#CFEFFF" \
+  -H "X-Genarrative-Image-Token: <token>" \
+  http://127.0.0.1:18083/remove-background -o output.png
 ```
 
 Config comes from `BGFILTER_CONFIG` (defaults to `configs/default.yaml` when present);
diff --git a/README_ZH.md b/README_ZH.md
index 1452df0..2fa9db0 100644
--- a/README_ZH.md
+++ b/README_ZH.md
@@ -186,7 +186,8 @@ python -m uvicorn app:app --host 127.0.0.1 --port 18083 --workers 1
 
 | 字段 | 必填 | 取值 | 含义 |
 | --- | --- | --- | --- |
-| `file` | 是 | 图片文件 | 源图(字段名固定为 `file`)。 |
+| `file` | `file`/`image_url` 二选一 | 图片文件 | 以上传方式提供源图(字段名固定为 `file`)。 |
+| `image_url` | `file`/`image_url` 二选一 | `http(s)://…` | 以 URL 提供源图,服务端自行 fetch(受大小/超时限制,跟随 ≤3 跳重定向)。`file` 与 `image_url` 必须恰好提供一个。无 SSRF 过滤:端点信任已鉴权的调用方,请保持鉴权开启且调用方可信。 |
 | `background_mode` | 否 | `flat`(默认)/ `complex` | `flat` = 纯色背景(走色键管线)。`complex` = 非纯色/实景背景:无色键,完全由分割器驱动抠图;此模式下 `screen_color` 被忽略,发丝细节通常弱于纯色模式。 |
 | `screen_color` | 否 | `#RRGGBB`,或留空/`auto` | 背景色先验;留空则从边框自动探测。`background_mode=complex` 时忽略。 |
 | `seg_model` | 否 | `birefnet`(默认)/ `anime-seg` | 分割后端。 |
diff --git a/app.py b/app.py
index 5e27064..deb7133 100644
--- a/app.py
+++ b/app.py
@@ -16,7 +16,9 @@ import time
 from concurrent.futures import ThreadPoolExecutor
 from contextlib import asynccontextmanager
 from pathlib import Path
+from urllib.parse import urlparse
 
+import httpx
 import numpy as np
 from fastapi import Depends, FastAPI, File, Form, Header, HTTPException, Response, UploadFile
 from PIL import Image, ImageOps, UnidentifiedImageError
@@ -46,6 +48,16 @@ def _int_env(name: str, default: int) -> int:
 
 
 MAX_IMAGE_PIXELS = _int_env("BGFILTER_MAX_IMAGE_PIXELS", 4_194_304)
+# /remove-background accepts either an uploaded `file` or an `image_url` the service
+# fetches itself. These bound the fetch so a huge or slow URL cannot hang a worker or
+# blow RAM. NOTE: no SSRF hardening here (no host/IP allowlist) -- the endpoint is
+# guarded by token auth and assumes trusted callers; do NOT expose it to untrusted
+# clients without adding private-address / redirect filtering to _fetch_image.
+FETCH_MAX_BYTES = _int_env("BGFILTER_FETCH_MAX_BYTES", 25 * 1024 * 1024)
+FETCH_TIMEOUT_S = float(os.environ.get("BGFILTER_FETCH_TIMEOUT", "15"))
+# Outbound fetch concurrency = 2 x cpu_workers, hard-capped here so a burst can never
+# exhaust file descriptors / ephemeral ports (each fetch = one socket).
+FETCH_CONCURRENCY_CAP = 64
 # We enforce our own pixel budget from the image header before decoding, so disable
 # Pillow's separate decompression-bomb ceiling (which would otherwise raise its own error).
 Image.MAX_IMAGE_PIXELS = None
@@ -81,6 +93,11 @@ _manager: PipelineManager | None = None
 # drains.
 _executor: ThreadPoolExecutor | None = None
 _sem: asyncio.Semaphore | None = None
+# Outbound image_url fetches: a FIFO admission bound (restores the throttle a direct
+# upload has implicitly) and a shared client for connection reuse. Both built in
+# lifespan once the pool size is known. See _fetch_image / §image_url mode.
+_fetch_sem: asyncio.Semaphore | None = None
+_http_client: httpx.AsyncClient | None = None
 
 
 def _build_settings():
@@ -106,13 +123,24 @@ def _build_settings():
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    global _manager, _executor, _sem
+    global _manager, _executor, _sem, _fetch_sem, _http_client
     settings = _build_settings()
     preload = os.environ.get("BGFILTER_PRELOAD", "1").strip().lower() not in ("0", "false", "no", "")
     # Pool size: config default (server.cpu_workers), env override wins.
     cpu_workers = max(1, _int_env("BGFILTER_CPU_WORKERS", settings.server.cpu_workers))
     _executor = ThreadPoolExecutor(max_workers=cpu_workers, thread_name_prefix="bgf")
     _sem = asyncio.Semaphore(cpu_workers)
+    # Outbound-fetch bound: 2x the pool (a small prefetch buffer), hard-capped. The
+    # shared client's max_connections matches so httpx never opens more sockets than
+    # the semaphore admits, and reuses connections across fetches.
+    fetch_conc = min(2 * cpu_workers, FETCH_CONCURRENCY_CAP)
+    _fetch_sem = asyncio.Semaphore(fetch_conc)
+    _http_client = httpx.AsyncClient(
+        follow_redirects=True,
+        max_redirects=3,
+        timeout=httpx.Timeout(connect=5.0, read=FETCH_TIMEOUT_S, write=10.0, pool=None),
+        limits=httpx.Limits(max_connections=fetch_conc),
+    )
     logger.info(
         "starting bgfilter service: device=%s default_seg=%s cross_check=%s preload=%s "
         "cpu_workers=%d gpu_concurrency=%d max_pixels=%d auth=%s",
@@ -130,7 +158,10 @@ async def lifespan(app: FastAPI):
             "BGFILTER_AUTH_TOKEN not set: /remove-background is OPEN (no app-level auth)"
         )
     _manager = PipelineManager(settings, preload=preload)
-    yield
+    try:
+        yield
+    finally:
+        await _http_client.aclose()
 
 
 app = FastAPI(title="BGfilter", version=VERSION, lifespan=lifespan)
@@ -205,6 +236,34 @@ def _normalize_background_mode(value: str | None) -> str | None:
     raise HTTPException(status_code=400, detail="invalid background_mode")
 
 
+async def _fetch_image(url: str) -> bytes:
+    """Download image bytes from a URL, bounded by size and time.
+
+    Robustness only -- no SSRF filtering (see FETCH_* note above): callers are
+    trusted via the auth token. Streams the body so an over-large response is
+    rejected without being fully buffered. Uses the shared client (connection reuse);
+    the caller holds _fetch_sem so outbound concurrency stays bounded.
+    """
+    parsed = urlparse(url)
+    if parsed.scheme not in ("http", "https") or not parsed.hostname:
+        raise HTTPException(status_code=400, detail="image_url must be an http(s) URL")
+    try:
+        async with _http_client.stream("GET", url) as resp:
+            if resp.status_code != 200:
+                raise HTTPException(
+                    status_code=400,
+                    detail=f"image_url fetch failed (status {resp.status_code})",
+                )
+            buf = bytearray()
+            async for chunk in resp.aiter_bytes():
+                buf += chunk
+                if len(buf) > FETCH_MAX_BYTES:
+                    raise HTTPException(status_code=413, detail="image_url body too large")
+            return bytes(buf)
+    except httpx.HTTPError:
+        raise HTTPException(status_code=400, detail="image_url fetch failed")
+
+
 def _decode_image(data: bytes) -> np.ndarray:
     if not data:
         raise HTTPException(status_code=400, detail="empty image")
@@ -235,7 +294,8 @@ def _encode_png(rgb: np.ndarray, alpha: np.ndarray) -> bytes:
 
 @app.post("/remove-background")
 async def remove_background(
-    file: UploadFile = File(...),
+    file: UploadFile | None = File(None),
+    image_url: str | None = Form(None),
     screen_color: str | None = Form(None),
     seg_model: str | None = Form(None),
     cross_check: str | None = Form(None),
@@ -245,6 +305,14 @@ async def remove_background(
     if _manager is None:
         raise HTTPException(status_code=503, detail="service not ready")
 
+    # Exactly one image source. (Auth already ran, so only authenticated callers can
+    # trigger an outbound fetch.)
+    url = image_url.strip() if image_url else None
+    if (file is None) == (not url):
+        raise HTTPException(
+            status_code=400, detail="provide exactly one of 'file' or 'image_url'"
+        )
+
     normalized_color = _normalize_screen_color(screen_color)
     normalized_seg = _normalize_seg_model(seg_model) or _manager.default_seg_model
     normalized_cc = _normalize_cross_check(cross_check)
@@ -260,11 +328,20 @@ async def remove_background(
     # Honors the lane lock (a locked cpu-fast box reports/runs "off" even on cross_check=on).
     effective_cc = _manager.effective_cross_check(normalized_cc)
 
+    # Acquire the raw bytes BEFORE the compute gate: a slow fetch must not hold a
+    # pool/compute slot. The upload body is already buffered by Starlette (fast);
+    # a URL is fetched under its own looser bound (_fetch_sem).
+    if url is not None:
+        async with _fetch_sem:
+            raw = await _fetch_image(url)
+    else:
+        raw = await file.read()
+
     # Bound in-flight requests to the pool size, then run on the pool. The GPU
     # sections are serialized inside the pipeline (a shared GPU mutex); the CPU
     # sections of concurrent requests overlap across the pool threads.
     async with _sem:
-        rgb = _decode_image(await file.read())
+        rgb = _decode_image(raw)
         start = time.perf_counter()
         loop = asyncio.get_running_loop()
         try:
diff --git a/requirements.txt b/requirements.txt
index 566302f..3b9a161 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -21,3 +21,4 @@ PyYAML
 fastapi
 uvicorn[standard]
 python-multipart
+httpx