Supprimer l'Arrière-Plan — Outil en ligne gratuit

Free background remover — transparent PNG in your browser

Outilolis Retourground Remover strips backgrounds from images entirely in your browser. Non uploads, no signups, no watermarks. It uses a Canvas-based flood-fill algorithm that detects the background color and replaces it with transparency, giving you a clean PNG cutout in seconds.

How background removal works

Unlike AI-powered services that rely on cloud servers and machine learning models, this tool uses a deterministic algorithm. It samples the top-left corner pixel of your image as the reference background color, then performs a flood-fill from each edge of the image. Every pixel whose color falls within your specified tolerance is made transparent. The result is a clean cutout that works especially well with studio photography, product shots, passport photos, and any image taken against a solid-color background.

The tolerance slider controls how aggressively similar colors are matched. A low tolerance (10-20) only removes pixels that are nearly identical to the reference color, which is ideal for pure white or pure green backgrounds. A higher tolerance (40-70) handles gradients, slight shadows, and off-white backgrounds but may start removing parts of the subject if colors overlap.

When to use this tool vs. AI-based removers

This browser-based tool excels at removing solid or near-solid backgrounds such as white studio backdrops, green screens, or single-color walls. For these scenarios, it matches or exceeds the speed of cloud-based tools because everything happens locally without upload/download latency.

For complex backgrounds with trees, crowds, indoor scenes, or intricate hair detail, AI-based tools like remove.bg, Adobe Photoshop, or Canva's background remover use trained neural networks that understand object boundaries at a semantic level and will produce superior results. Consider this tool a fast, private alternative for the 70% of use cases involving clean backgrounds.

Best practices for clean results

• Use high-contrast images — the greater the difference between subject and background, the cleaner the cutout.
• Start with lower tolerance — increase gradually until the background disappears without eating into the subject.
• Enable edge feathering — this softens anti-aliased edges and prevents harsh, pixelated borders around the subject.
• Crop first if needed — if the top-left corner of your image is part of the subject rather than the background, crop or rotate so the background occupies a corner.
• Use PNG output — JPG does not support transparency. The download is always PNG to preserve the alpha channel.

Common use cases

• E-commerce product photos — remove white backgrounds for marketplace listings on Amazon, eBay, or Shopify.
• Passport and ID photos — replace colored backgrounds with white or transparent for official documents.
• Presentations and slides — place subjects on custom backgrounds without Photoshop skills.
• Social media graphics — create clean cutouts for Instagram stories, YouTube thumbnails, and TikTok content.
• Design mockups — extract subjects for compositing in Figma, Canva, or other design tools.

Technical details

The algorithm uses a BFS (breadth-first search) flood-fill starting from all four edges of the image. Color distance is computed in RGB space using the Euclidean distance formula: sqrt((r1-r2)^2 + (g1-g2)^2 + (b1-b2)^2). A pixel is made transparent when its distance from the reference color is less than the tolerance value scaled to the 0-441 range (441 being the maximum possible RGB distance). Edge feathering applies a Gaussian-like alpha gradient to pixels near the transparency boundary, creating a smoother transition.

All processing happens on the HTML5 Canvas using getImageData and putImageData. For a 4000x3000 image, this processes 12 million pixels, which typically completes in 1-3 seconds on modern hardware. Non WebAssembly, no ONNX models, no external dependencies.