AI-powered background removal with ultimate flexibility: free open-source model, premium API (€0.05/image), or self-hosted Docker app.
Replies
Best
Maker
📌
Hey Product Hunt! 👋
Background removal is hard. Getting clean edges, small details, and stable quality is a real challenge.
I built **withoutBG** to make this easier. The Focus v1.0.0 model is open-source, runs locally, and gives sharp results for free. If you want the highest quality possible, there’s also the withoutBG Pro API.
This is super cool. I see great results even for non-contrasting backgrounds. The spider web example is excellent. When you process the images, do you save them on your server? Would love to try the API
We don’t store any images at all. Nothing is saved on the server, and no thumbnails, hashes, or derivatives are kept or used for training. Everything is processed in RAM, then the buffers are cleared right after the response.
Replies
Hey Product Hunt! 👋
Background removal is hard. Getting clean edges, small details, and stable quality is a real challenge.
I built **withoutBG** to make this easier. The Focus v1.0.0 model is open-source, runs locally, and gives sharp results for free. If you want the highest quality possible, there’s also the withoutBG Pro API.
Three ways to use it:
- Open-source model: free, local, fast, high quality
- Pro API: best quality, €0.05/image
- Docker web app: self-hosted, one command to run
All options use the same Python SDK: pip install withoutbg
With local processing, your files never leave your computer. Everything is under the Apache 2.0 license.
What matters most to you? Privacy, cost, or quality? I’d love your feedback!
Quality examples (Focus model): https://withoutbg.com/resources/...
GitHub: https://github.com/withoutbg/wit...
Swytchcode
This is super cool. I see great results even for non-contrasting backgrounds. The spider web example is excellent.
When you process the images, do you save them on your server?
Would love to try the API
@chilarai Thanks so much.
We don’t store any images at all. Nothing is saved on the server, and no thumbnails, hashes, or derivatives are kept or used for training. Everything is processed in RAM, then the buffers are cleared right after the response.
Would love to have you try the API.
Swytchcode
@imran_kocabiyik Ya, I'll give it a try. Seems amazing.