What is the maximum resolution used for feature extraction?

  • 1
  • Question
  • Updated 11 years ago
Since the synther only uses 1.5 megapixels (or 2 megapixels, depending who you ask) versions of our photos to analyse for image features and there is no setting to change this, even when synthing relatively small numbers of images, the obvious workaround is to include tiles or crops of the original images alongside the originals all in one synth.

My question is as follows: What is the largest size width or height that I can make these tiles without the synther resizing them down again (as far as the feature extraction is concerned)?

Does the synther:
a) make a copy of each photo to analyse that is separate from the DZIs in the %temp%/photosynther/images/ folder (e.g.
static - with a specific maximum width|height,
dynamic - anything that multiplies to 1,500,000 pixels or the next lowest number based on the aspect ratio of the image, or
the square root of 1,500,000), or
b) do you simply utilise a specific zoom level in the DZI folder structure, meaning that in each case the maximum width|height of the image used for feature extraction is slightly different (dependent on the original resolution of the image)?

My interest is only that I wish to absolutely minimise the number of partial-image tiles in my synth while ensuring that every last pixel is scanned for image features.

If there's some secret command line way to tell the synther to use the full resolution for feature extraction, though, I'll take it in a heartbeat. =] I'm no more keen than you to see synths where each image quad is covered by a grid of smaller quads.
Photo of Nathanael Lawrence

Nathanael Lawrence

  • 795 Posts
  • 55 Reply Likes
  • ready to ask for an answer after more than a year of wondering.

Posted 11 years ago

  • 1

Be the first to post a reply!