Jump to content

Good picture: Difference between revisions

From InkanChain
Gordon (talk | contribs)
mNo edit summary
Gordon (talk | contribs)
mNo edit summary
 
Line 23: Line 23:
Some object like banknotes may have transparent areas
Some object like banknotes may have transparent areas


As we ask for 2 pictures and use both to analyze similarity it is a good practice to move slightly this object over it's background between picture #1 and picture #2.  
As we ask for 2 pictures and use both to analyze similarity it is a good practice to move slightly this object over it's background between picture #1 and picture #2.


This will artificially create a difference that will void usage of this transparency area, and is rather a simple and helpfull methode,
Still keeping the same framing, this will artificially create a difference of the background of "see trough" transparent area, that automatically will be detected and void the usage of this transparent area. This rather a simple and helpful method,

Latest revision as of 16:11, 10 October 2025

Good pictures are essential for best results

During registration

  • Keep in mind that a similar picture will be needed during Authentication.
  • The quality of picture used during Authentication are important but less critical.
  • The first picture will become the "Reference" make it your best.


Photo should be taken as close as possible to the object but not too close to avoid blurriness

Lighting condition should be as constant as possible, and we strongly recommend using the flash (we use a flash detection algorithm to automatically reject usage of overexposed area, so feel safe)

Use a point of view facing the flattest surface of the object.

For very small object (pils, jewels, ...) , use a Camera with Macro capabilities like iPhone Pro (3 lenses) ,

but be careful that the flash should be constantly lighting the same area, this require cares to have a good registration.


Tips for transparency

Some object like banknotes may have transparent areas

As we ask for 2 pictures and use both to analyze similarity it is a good practice to move slightly this object over it's background between picture #1 and picture #2.

Still keeping the same framing, this will artificially create a difference of the background of "see trough" transparent area, that automatically will be detected and void the usage of this transparent area. This rather a simple and helpful method,