Skip to main content

FAQ

Blur General​

Why is the quality of the logo in final blurred image really low?​

This happens when we have to transform the logo to fit into the shape of the plate at an angle.

Why is the blur API slower than the Snapshot Cloud solution?​

Snapshot doesn't perform any image modification and Blur has to first process the image with Snapshot before blurring.

How can I reduce the processing time?​

  1. Using the Snapshot SDK on-premise instead of Cloud API.
  2. Turning off face detection so only plates are processed.
  3. Ensure you are using platerecognizer/skew-correction instead of platerecognizer/skew-correction:no-avx.

How can I effectively process large images such as panoramic to find all plates?​

Use the image splitting parameters.

How can I choose the best splitting parameters?​

Enable DEBUG logging and test a sample image to preview the regions of splitting for easier adjustment of the configs.

How can I blur plates and/or faces in my video files?​

If you need to blur plates and faces in a video, you can make use of our video utilities.

How can I use on-premise Snapshot SDK?​

Two options:

  1. Use only the first compose file and set SDK_URL in .env file. Use this when Snapshot is already deployed elsewhere.
  2. Just include the second compose file. Use this to deploy Snapshot with compose.

How can I scale to process multiple images in parallel?​

By default, Blur onPremise runs one worker process which processes one image at a time. For multiprocessing of images, configure number of WORKERS and THREADS started by the (Gunicorn) server.

Floating button icon
CTRL + .