惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

V2EX - 技术
V2EX - 技术
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
T
Threat Research - Cisco Blogs
T
The Exploit Database - CXSecurity.com
S
Schneier on Security
S
Securelist
P
Privacy & Cybersecurity Law Blog
Scott Helme
Scott Helme
T
Threatpost
C
Cybersecurity and Infrastructure Security Agency CISA
L
LINUX DO - 热门话题
Cyberwarzone
Cyberwarzone
Cisco Talos Blog
Cisco Talos Blog
量子位
博客园 - Franky
cs.CL updates on arXiv.org
cs.CL updates on arXiv.org
Latest news
Latest news
T
Troy Hunt's Blog
N
News | PayPal Newsroom
Google Online Security Blog
Google Online Security Blog
Apple Machine Learning Research
Apple Machine Learning Research
N
Netflix TechBlog - Medium
小众软件
小众软件
P
Palo Alto Networks Blog
Spread Privacy
Spread Privacy
C
Cyber Attacks, Cyber Crime and Cyber Security
C
Check Point Blog
aimingoo的专栏
aimingoo的专栏
WordPress大学
WordPress大学
L
Lohrmann on Cybersecurity
L
LINUX DO - 最新话题
D
Darknet – Hacking Tools, Hacker News & Cyber Security
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
The Last Watchdog
The Last Watchdog
S
Security @ Cisco Blogs
P
Privacy International News Feed
Last Week in AI
Last Week in AI
Microsoft Security Blog
Microsoft Security Blog
T
Tailwind CSS Blog
博客园_首页
云风的 BLOG
云风的 BLOG
V
Vulnerabilities – Threatpost
D
DataBreaches.Net
Recent Announcements
Recent Announcements
酷 壳 – CoolShell
酷 壳 – CoolShell
CTFtime.org: upcoming CTF events
CTFtime.org: upcoming CTF events
罗磊的独立博客
Engineering at Meta
Engineering at Meta
Forbes - Security
Forbes - Security
T
Tenable Blog

Hacker News: Show HN

PurrrrrFocus: Pomodoro Timer App - App Store Workflow Engine — Multi-Step Orchestration for Bun RapidPhoto: Pro Photo Editor App - App Store GitHub - DheerG/swarms: Achieve extraordinary results with claude code across a variety of tasks SPICE simulation → oscilloscope → verification with Claude Code — Lucas Gerads Show HN: VCoding – A 5 MB native Windows IDE with no dynamic dependencies Show HN: LLMs don't hallucinate because they're bad at math, it's the format GitHub - Agent-FM/agentfm-core: AgentFM is a peer-to-peer network that turns everyday computers into a decentralized AI supercomputer. AgentFM lets you run massive AI workloads directly across a global mesh of idle CPUs and GPUs. Show HN: Tracking Top US Science Olympiad Alumni over Last 25 Years GitHub - Potarix/agent-hub: One place to talk to all your agents Show HN: Runtime security for AI agents(injection,tool abuse, data exfiltration) GitHub - dubeyKartikay/lazyspotify: Terminal Spotify client for macOS and Linux GitHub - the-banana-tool/king-louie: Easy to use GUI Personal AI Assistant. Win/Linux/Mac. Show HN I made my vacation rental bookable by AI agents–no Airbnb, 0% commission GitHub - basteez/jsf-autoreload: maven plugin to enable hot reload on jsf projects uvm32/hosts/host-gdbstub at main · ringtailsoftware/uvm32 GitHub - labsai/EDDI: Config-driven engine that turns JSON into production-grade AI agents. Multi-agent orchestration, 12+ LLM providers, MCP/A2A protocols, RAG, persistent memory, and enterprise compliance (EU AI Act, GDPR, HIPAA). Built on Quarkus. GitHub - glitchnsec/fortyone-oss: AI Executive Assistant Platform Quickstart | Alien GitHub - muxshed/shed: One stream in, or many. Every destination, simultaneously. No cloud middleman, no per-channel fees, no limits. GitHub - ocrbase-hq/ocrbase: 📄 PDF/IMG ->.MD/JSON Document OCR API for PaddleOCR and GLMOCR. Self-hostable. GitHub - impactjo/home-memory: MCP server that lets your AI assistant remember everything about your home. GitHub - Sets88/dbcls: DbCls is a powerful terminal database client that supports various databases GitHub - neptun2000/heor-agent-mcp GitHub - SeanFDZ/macmind: Single-layer transformer in HyperTalk for the classic Macintosh RollQuation: Math Puzzles - Apps on Google Play GitHub - dropbox/witchcraft Show HN: Agent-cache – Multi-tier LLM/tool/session caching for Valkey and Redis GitHub - opentalon/opentalon: OpenTalon is an open-source platform built from the ground up in Go as a robust alternative to OpenClaw LinkedIn™ 职位抓取工具 - Chrome 应用商店 GitHub - EdoardoBambini/Agent-Armor-Iaga: AI agents are getting tool access — shell, file system, databases, APIs, secrets. But **nobody is governing what they actually do with it**. Frameworks like LangChain, CrewAI, AutoGen, and Claude Code give agents the power to execute. Agent Armor gives you the power to control, audit, and approve every single action before it happens. HN Vibes — Week 15, Apr 7–13 2026 GitHub - chojs23/ec: Easy terminal-native 3-way git mergetool vim-like workflow GitHub - SethPyle376/hiraeth: Local AWS emulator focused on fast integration testing, with SQS support, SQLite-backed state, and a debug-friendly web UI. GitHub - JakOb-dotcom/cloud-sandbox-security-analysis: Technical analysis and Proof of Concept (PoC) regarding environment variable exfiltration in containerized cloud sandboxes via side-channel data leaks. Springboards - Flint Alpha Show HN: A simpler coding agent harness GitHub - audiodude/sudomake-friends GitHub - 256thFission/mini-mythos: OSS clone of Anthropic’s Mythos harness to locate C/C++ memory vulnerabilities Show HN: OpenParallax: OS-level privilege separation for AI agent execution Hacker News Sorted - Chrome 应用商店 Show HN: How to Install Docker on Ubuntu 24.04 LTS: Complete 2026 Guide GitHub - himanshudongre/smriti GitHub - sverrirsig/claude-control: macOS desktop dashboard for monitoring and managing multiple Claude Code sessions GitHub - ory/dockertest: Write better integration tests! Dockertest helps you boot up ephermal docker images for your Go tests with minimal work. Chiral - Chrome 应用商店 Show HN: Two Claudes collaborating through shared memory on a $100 mini-PC GitHub - pmichaillat/latex-cv: Minimalist LaTeX template for academic CVs GitHub - oguzbilgic/posse: A web UI for Anthropic Managed Agents. GitHub - sshiraz/depsly: Dependency risk analysis tool for npm packages ABI Add safari/agent-harness — Safari browser automation via safari-mcp by achiya-automation · Pull Request #212 · HKUDS/CLI-Anything GitHub - Halfblood-Prince/trustcheck: Verify PyPI package attestations and improve Python supply-chain security GitHub - oguzbilgic/kern-ai: Agents that do the work and show it. GitHub - bruits/satteri: High-performance Markdown and MDX processing for the JavaScript ecosystem GitHub - tylergibbs1/feedstock: High-performance web crawler and scraper for TypeScript, powered by Bun and Playwright GitHub - Grimm67123/grimmbot: The self-improving sandboxed and open-source AI agent. With persistent memory and scheduling. GitHub - whitevanillaskies/whitebloom: Local whiteboard that blooms. GitHub - hwdsl2/docker-whisper: Docker image for a self-hosted Whisper speech-to-text server with speaker diarization and OpenAI-compatible transcription and translation APIs. Powered by faster-whisper. Supports all Whisper models, NVIDIA GPU (CUDA) acceleration, JSON/SRT/VTT output, SSE streaming, offline mode, and multi-arch (amd64, arm64). GitHub - yisding/reviewwiggum GitHub - MarwanAlsoltany/serrors: Structured errors for Go: sentinel hierarchies, typed data, custom formatting, and slog integration. GitHub - soatok/age-php GitHub - Luthiraa/markitme GitHub - stagas/rtdiff: realtime git diff gui and AI-assisted commits GitHub - tombedor/excalicharts GitHub - wh1le/excalidraw-edit: Open and edit .excalidraw files from the terminal. Offline, auto-saves to disk. MalExt Sentry - Malicious Extension Scanner - Chrome 应用商店 GitHub - syi0808/asciianimesvg: Generate animated ASCII art SVGs from text. CLI, Rust library, WASM, and web editor. GitHub - zaina-ml/ml_forge: A visual-based graph node editor for training computer vision models. GitHub - anakin87/llm-rl-environments-lil-course: 🌱 A little course on Reinforcement Learning Environments for evaluating and training Language Models GitHub - takaakit/superpowers-uml: Superpowers-UML modifies Superpowers to ensure a software development workflow in which AI agents design through UML modeling. AdriByte Studio - Sviluppo Web e Soluzioni Digitali GitHub - chouligi/angel-copilot: Your personalized Angel Investment Advisor Show HN: MoodSense AI (ML and FastAPI and Gradio, Deployed on Hugging Face) Moodsense Ai - a Hugging Face Space by aman179102 GitHub - agenteractai/lodmem: Level Of Detail Context Management for Agents GitHub - ostefani/subnetlens: A fast, concurrent network scanner with a TUI and plain-text CLI, built in Go. It discovers live hosts on your network, scans their open ports, resolves hostnames, and fingerprints operating systems—delivered. Cyber Pulse: Agentic Intel - Apps on Google Play Whisper API: Self-Hostable Speech to Text Transcription The Agent-Web Protocol Stack: A Research Thesis GitHub - msmarkgu/RelayFreeLLM: A restful API designed to route user prompts to various AI model providers. Show HN: Provepy – A Python decorator that proves your code using Lean and LLMs Show HN: Pardonned.com – A searchable database of US Pardons GitHub - patrickdappollonio/dux: Dux is a terminal UI that lets you run multiple AI coding agents side by side, each in its own git worktree, with full companion terminals, macros, commit generation, and a command palette that knows more tricks than you do. kMC Crystal Simulator Show HN: HyperFlow – A self-improving agent framework built on LangGraph GitHub - stef41/vibescore: 🎵 Grade your vibe-coded project. One command, instant letter grade across security, quality, dependencies, and testing. GitHub - stef41/lmscan: 🔍 Detect AI-generated text and fingerprint which LLM wrote it. Open-source GPTZero alternative. Zero dependencies, works offline. imgur.com GitHub - visionscaper/collabmem: Enabling long-term collaboration with Agentic AI - building up episodic and world model memory over time with in-context awareness 在 Steam 上购买 FriedrichAI: Offline AI 立省 10% GitHub - atripati/ark: AI Runtime Kernel — a context operating system for AI agents. Eliminates tool bloat, loads only what’s needed, and gives LLMs their reasoning space back. GitHub - nowork-studio/toprank: Open-source Claude Code skills for SEO, SEM, Google Ads GitHub - tacomanator/sash: Lightweight macOS menu bar app for reliably cycling through windows of the current application. Appents | Social Media Management for Product-First Teams GitHub - pnhoang/youtube-spam-blocker: Automatically detects and hides spam messages in YouTube Live chat. Set rate limits, keyword filters, and block repeat offenders. GitHub - decisionnode/DecisionNode: CLI + Local MCP - A shared structured memory store across Claude Code, Cursor, Windsurf, Antigravity, and every MCP client. Semantically queryable. GitHub - AvaCodeSolutions/django-email-learning: An open source Django app for creating email-based learning platforms with IMAP integration and React frontend components. The $100K Gap in Kubernetes Security Tooling Function Calling Harness: From 6.75% to 100%
GitHub - useknockout/api
tlorents · 2026-04-26 · via Hacker News: Show HN

State-of-the-art background removal API — open source, self-hostable, 40× cheaper than remove.bg.

MIT License npm version npm downloads GitHub stars Powered by Modal Model: BiRefNet Python TypeScript

Live API · Docs · Quick Start · API Reference · Self-hosting

useknockout before/after — background removal demo

Drop an image in. Get a transparent PNG out. ~200ms per call.

A production-grade background removal API powered by BiRefNet — the current SOTA on DIS5K, HRSOD, and COD benchmarks. Served on Modal's GPU infrastructure with scale-to-zero economics.

  • SOTA quality — matches or beats remove.bg, Photoroom, and Pixelcut on hair, fur, fine detail
  • Fast — ~200ms per image on a warm L4 GPU
  • Cheap — ~$0.00005 per image raw compute cost (4,000x cheaper than remove.bg PAYG)
  • MIT licensed — model weights and code, commercial use OK
  • Self-hostable — deploy to your own Modal workspace in one command

Works alpha-preserving (PNG with transparent bg) OR opaque (solid color / remote image as new bg).


Table of contents


Demo

Live endpoint: https://useknockout--api.modal.run

Interactive docs: https://useknockout--api.modal.run/docs

Input → Output:

Original After
Complex hair Clean wisps, no halo
Fur / pet photos Soft edges preserved
Product shots Sharp, clean cutout
Low-contrast subjects Accurate separation

Quick start

Public beta token — copy, paste, try it right now

During public beta, everyone shares this bearer token:

kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f

No signup. Just use it. We're free during beta. Paid tier launches soon — need your own key or higher limits? DM @useknockout.

Hit the API in 3 seconds

curl -X POST "https://useknockout--api.modal.run/remove" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "file=@your-image.jpg" \
  -o out.png

You get a PNG with a transparent alpha channel. Done.

With a URL instead of a file

curl -X POST "https://useknockout--api.modal.run/remove-url" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com/cat.jpg"}' \
  -o out.png

Replace the background with a color or remote image

# solid color background
curl -X POST "https://useknockout--api.modal.run/replace-bg" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "file=@cat.jpg" \
  -F "bg_color=#FF5733" \
  -F "format=jpg" \
  -o out.jpg

# use a remote image as the new background
curl -X POST "https://useknockout--api.modal.run/replace-bg" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "file=@cat.jpg" \
  -F "bg_url=https://example.com/mountains.jpg" \
  -o out.png

Batch — process up to 10 images in one call

# multipart batch
curl -X POST "https://useknockout--api.modal.run/remove-batch?format=png" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "files=@a.jpg" -F "files=@b.jpg" -F "files=@c.jpg"

# URL batch — JSON body
curl -X POST "https://useknockout--api.modal.run/remove-batch-url" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -H "Content-Type: application/json" \
  -d '{"urls":["https://a.jpg","https://b.jpg"], "format":"png"}'

Both return JSON: { "count": N, "format": "png", "results": [{ "success": true, "data_base64": "..." }, ...] }.

More presets (v0.3.0)

# Sticker — cutout + thick white outline (WhatsApp / iMessage sticker style)
curl -X POST "https://useknockout--api.modal.run/sticker" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "file=@photo.jpg" -F "stroke_width=24" -o sticker.png

# Smart crop — tight bounding box around subject
curl -X POST "https://useknockout--api.modal.run/smart-crop" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "file=@photo.jpg" -F "padding=32" -o cropped.png

# Studio shot — e-commerce preset (white bg + shadow + centered, 1:1 aspect)
curl -X POST "https://useknockout--api.modal.run/studio-shot" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "file=@photo.jpg" -F "aspect=1:1" -F "format=jpg" -o studio.jpg

# Shadow — subject composited onto new bg with a drop shadow
curl -X POST "https://useknockout--api.modal.run/shadow" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "file=@photo.jpg" -F "bg_color=#F3F4F6" -o shadow.png

# Compare — before/after side-by-side for marketing/social
curl -X POST "https://useknockout--api.modal.run/compare" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "file=@photo.jpg" -o compare.png

# Mask — just the black/white mask, for your own pipeline
curl -X POST "https://useknockout--api.modal.run/mask" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "file=@photo.jpg" -o mask.png

# Outline — subject on transparent bg with a thin outline
curl -X POST "https://useknockout--api.modal.run/outline" \
  -H "Authorization: Bearer kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f" \
  -F "file=@photo.jpg" -F "outline_color=#000000" -F "outline_width=4" -o outline.png

Health check

curl https://useknockout--api.modal.run/health
# {"status":"ok","model":"ZhengPeng7/BiRefNet"}

API reference

Base URL: https://useknockout--api.modal.run

POST /remove

Remove the background from an uploaded image.

Headers

Header Required Description
Authorization Yes Bearer <API_TOKEN>
Content-Type Auto multipart/form-data (set by your client)

Bodymultipart/form-data

Field Type Required Description
file binary Yes Image to process (JPEG, PNG, WebP). Max 25 MB.

Query params

Param Type Default Description
format string png png (default) or webp. Both include alpha.

Responseimage/png or image/webp with a transparent background.

POST /remove-url

Fetch an image from a URL and remove its background.

Headers

Header Required Description
Authorization Yes Bearer <API_TOKEN>
Content-Type Yes application/json

Body — JSON

{
  "url": "https://example.com/image.jpg",
  "format": "png"
}

Response — same as /remove.

POST /replace-bg

Remove the background and composite the subject onto a new background — solid color or a remote image.

Headers

Header Required Description
Authorization Yes Bearer <API_TOKEN>
Content-Type Auto multipart/form-data (set by your client)

Bodymultipart/form-data

Field Type Required Description
file binary Yes Foreground image to process. Max 25 MB.
bg_color string No (default #FFFFFF) Hex color for the new background. Examples: #000000, #ff5733, #1a73e8.
bg_url string No Remote URL of a background image. Takes precedence over bg_color.
format string No (default png) Output format: png, webp, or jpg (smallest, opaque only).

Responseimage/png, image/webp, or image/jpeg with the subject composited onto the new background. Edges are cleaned via closed-form foreground matting (no color spill, no halo).

POST /remove-batch

Remove backgrounds from up to 10 images in one call.

Headers

Header Required Description
Authorization Yes Bearer <API_TOKEN>
Content-Type Auto multipart/form-data

Bodymultipart/form-data with repeated files fields.

Query params

Param Type Default Description
format string png png or webp. Applies to every result.

Response — JSON:

{
  "count": 3,
  "format": "png",
  "results": [
    { "filename": "a.jpg", "success": true, "format": "png", "size_bytes": 124503, "data_base64": "..." },
    { "filename": "b.jpg", "success": true, "format": "png", "size_bytes": 98321, "data_base64": "..." },
    { "filename": "c.jpg", "success": false, "error": "Invalid or unsupported image" }
  ]
}

Each data_base64 decodes to PNG/WebP bytes with a transparent background.

POST /remove-batch-url

Same as /remove-batch but takes a JSON array of remote URLs.

Body — JSON:

{
  "urls": ["https://example.com/a.jpg", "https://example.com/b.jpg"],
  "format": "png"
}

Response — same JSON shape as /remove-batch, with url in place of filename.

POST /mask

Return just the black/white alpha mask as a grayscale PNG/WebP. Useful for chaining into your own compositing pipeline (Photoshop actions, ffmpeg keying, custom workflows).

Field Type Default Description
file binary required Foreground image.
format string png png or webp.

Response — grayscale image (0 = background, 255 = subject).

POST /smart-crop

Auto-crop to the subject's tight bounding box + padding.

Field Type Default Description
file binary required Foreground image.
padding int 24 Pixels of padding around the bbox.
transparent bool true true → cropped cutout with transparent bg. false → cropped region from the original image (bg preserved).
format string png png, webp, or jpg (when transparent=false).

Response — cropped image.

POST /shadow

Composite the subject onto a new background with a configurable drop shadow.

Field Type Default Description
file binary required Foreground image.
bg_color string #FFFFFF Hex color for the new background.
bg_url string Optional remote URL. Takes precedence over bg_color.
shadow_color string #000000 Hex color for the shadow.
shadow_offset_x int 8 Shadow offset in pixels (X).
shadow_offset_y int 12 Shadow offset in pixels (Y).
shadow_blur int 14 Gaussian blur radius in pixels.
shadow_opacity float 0.45 0.0–1.0.
format string png png, webp, or jpg.

POST /sticker

Subject with a thick outline on a transparent background — iMessage / WhatsApp / Telegram sticker style.

Field Type Default Description
file binary required Foreground image.
stroke_color string #FFFFFF Outline color.
stroke_width int 20 Outline width in pixels (capped at 80).
format string png png or webp.

POST /outline

Subject on transparent background with a thin outline.

Field Type Default Description
file binary required Foreground image.
outline_color string #000000 Outline color.
outline_width int 4 Outline width in pixels (capped at 60).
format string png png or webp.

POST /studio-shot

E-commerce preset: remove background → tight crop → center on solid-color canvas → optional drop shadow → standardized aspect ratio.

Field Type Default Description
file binary required Foreground image.
bg_color string #FFFFFF Canvas color.
aspect string 1:1 W:H format. Examples: 1:1, 4:5, 16:9, 3:2.
padding int 48 Padding around the subject in pixels.
shadow bool true Include a soft drop shadow.
format string jpg png, webp, or jpg.

POST /compare

Before/after side-by-side preview — original on the left, transparent cutout (on a checkerboard) on the right. Great for marketing / social media screenshots.

Field Type Default Description
file binary required Foreground image.
format string png png or webp.

GET /health

Returns {"status":"ok","model":"ZhengPeng7/BiRefNet"}. No auth required.

GET /docs

Interactive OpenAPI (Swagger) UI.

Errors

Code Meaning
400 Invalid image, missing field, malformed URL, invalid hex color, or batch > 10 items
401 Missing Authorization header
403 Invalid bearer token
413 Image exceeds 25 MB limit
500 Server error (check dashboard logs)

Edge quality

All endpoints apply closed-form foreground matting (via pymatting) after mask prediction. This estimates pure foreground color at soft edges, eliminating color spill from the original background. Result: no halos, no fringing, even on backgrounds that differ sharply from the subject.


Client examples

Python

import requests

URL = "https://useknockout--api.modal.run/remove"
TOKEN = "kno_public_beta_4d7e9f1a3c5b2e8d6a9f7c1b3e5d8a2f"  # public beta token

with open("input.jpg", "rb") as f:
    resp = requests.post(
        URL,
        headers={"Authorization": f"Bearer {TOKEN}"},
        files={"file": f},
    )
resp.raise_for_status()

with open("output.png", "wb") as f:
    f.write(resp.content)

Node.js SDK (recommended)

npm i @useknockout/node
import { writeFile } from "node:fs/promises";
import { Knockout } from "@useknockout/node";

const client = new Knockout({ token: process.env.KNOCKOUT_TOKEN! });

// 1. Remove background → transparent PNG
const png = await client.remove({ file: "./input.jpg" });
await writeFile("out.png", png);

// 2. Replace background with a color
const jpg = await client.replaceBackground({
  file: "./input.jpg",
  bgColor: "#FF5733",
  format: "jpg",
});
await writeFile("out.jpg", jpg);

// 3. Replace background with a remote image
const composed = await client.replaceBackground({
  file: "./input.jpg",
  bgUrl: "https://example.com/mountains.jpg",
});

// 4. Batch — process 10 URLs in one call
const batch = await client.removeBatchUrl({
  urls: ["https://example.com/a.jpg", "https://example.com/b.jpg"],
});
for (const r of batch.results) {
  if (r.success) await writeFile(`out-${r.url}.png`, Buffer.from(r.data_base64!, "base64"));
}

Node.js (raw fetch, no SDK)

import { readFile, writeFile } from "node:fs/promises";

const URL = "https://useknockout--api.modal.run/remove";
const TOKEN = process.env.KNOCKOUT_TOKEN;

const buf = await readFile("input.jpg");
const form = new FormData();
form.set("file", new Blob([buf]), "input.jpg");

const res = await fetch(URL, {
  method: "POST",
  headers: { Authorization: `Bearer ${TOKEN}` },
  body: form,
});
if (!res.ok) throw new Error(await res.text());

await writeFile("output.png", Buffer.from(await res.arrayBuffer()));

TypeScript (browser / Next.js)

export async function removeBackground(file: File, token: string) {
  const form = new FormData();
  form.append("file", file);

  const res = await fetch("https://useknockout--api.modal.run/remove", {
    method: "POST",
    headers: { Authorization: `Bearer ${token}` },
    body: form,
  });

  if (!res.ok) throw new Error(`knockout error: ${res.status}`);
  return await res.blob(); // PNG with alpha
}

Go

package main

import (
    "bytes"
    "io"
    "mime/multipart"
    "net/http"
    "os"
)

func removeBG(path, token string) ([]byte, error) {
    f, err := os.Open(path)
    if err != nil { return nil, err }
    defer f.Close()

    body := &bytes.Buffer{}
    w := multipart.NewWriter(body)
    part, _ := w.CreateFormFile("file", path)
    io.Copy(part, f)
    w.Close()

    req, _ := http.NewRequest("POST",
        "https://useknockout--api.modal.run/remove", body)
    req.Header.Set("Authorization", "Bearer "+token)
    req.Header.Set("Content-Type", w.FormDataContentType())

    resp, err := http.DefaultClient.Do(req)
    if err != nil { return nil, err }
    defer resp.Body.Close()
    return io.ReadAll(resp.Body)
}

cURL — WebP output (smaller files)

curl -X POST "https://useknockout--api.modal.run/remove?format=webp" \
  -H "Authorization: Bearer $TOKEN" \
  -F "file=@input.jpg" \
  -o output.webp

Benchmarks

Measured on Modal gpu="L4", Python 3.11, torch 2.4, batch size 1, 1024×1024 model input.

Image size Warm latency (p50) Cold start Output format
512×512 180 ms ~25 s PNG / WebP
1024×1024 220 ms ~25 s PNG / WebP
2048×2048 310 ms ~25 s PNG / WebP
4000×4000 520 ms ~25 s PNG / WebP

Quality vs. competitors

BiRefNet (the model we serve) consistently ranks first or second on public benchmarks:

  • DIS5K (Dichotomous Image Segmentation): #1 F-measure as of 2024
  • HRSOD (High-Resolution Salient Object Detection): #1 MAE
  • COD10K (Camouflaged Object Detection): #1 or #2 depending on metric

See the BiRefNet paper and leaderboards for details.


Self-hosting

Want to run your own instance? One command after Modal setup.

Prerequisites

pip install modal
modal token new

Clone & deploy

git clone https://github.com/useknockout/api.git
cd api

# create your bearer-token secret
modal secret create knockout-secrets API_TOKEN=$(openssl rand -hex 32)

# deploy
modal deploy main.py

Modal prints your live HTTPS URL. First deploy takes ~5 min (image build + weight bake). Subsequent deploys take seconds.

Tune for your workload

Edit main.py:

@app.cls(
    gpu="L4",              # or "A10", "A100", "H100"
    scaledown_window=60,   # seconds of idle before scale-to-zero
    max_containers=10,     # max concurrent containers
)
  • Latency-critical? Keep one warm: min_containers=1 (costs ~$0.80/hr 24/7).
  • Throughput-critical? Bump max_containers and use @modal.concurrent(max_inputs=4) to batch.
  • Higher quality? Change MODEL_INPUT_SIZE to (2048, 2048) — 4x slower, sharper edges.

Architecture

┌────────────┐      HTTPS       ┌───────────────────────────┐
│   Client   │ ───────────────▶ │  Modal ASGI (FastAPI)     │
│ (any lang) │                  │  ┌─────────────────────┐  │
└────────────┘                  │  │ Auth (bearer)       │  │
                                │  │ Validation          │  │
                                │  │ Image decode (PIL)  │  │
                                │  │ BiRefNet on L4 GPU  │  │
                                │  │ Encode (PNG/WebP)   │  │
                                │  └─────────────────────┘  │
                                │  Scale-to-zero, auto-HTTPS │
                                └───────────────────────────┘
  • One file (main.py), single Modal class, two endpoints + health + docs
  • Weights baked into image at build time — cold starts are just image pull + GPU model load (~25 s)
  • FastAPI handles multipart, JSON, CORS, OpenAPI schema generation

Pricing

The hosted API at useknockout--api.modal.run is in closed beta while we validate quality. Request an API key: contact.

When the paid tier goes live:

Tier Price Best for
Free 50 images / month, no card Personal, eval, open source
Pay-as-you-go $0.005 / image Side projects, early startups
Volume $0.003 / image at 100k+/mo Production workloads
Enterprise Custom, private endpoints Compliance, BYO-cloud

For reference — the same image on remove.bg is $0.20 at their PAYG rate.

Credits never expire. No subscriptions. You only pay for what you use.


Contact


License

MIT License — see LICENSE. Model weights (BiRefNet) are also MIT. Commercial use is allowed for both.


Built in a few hours because someone said it couldn't be done.