2011.07.29 00:49 "Re: [Tiff] Using photon lists rather than rasters", by Ryan Wong

Actually, the sticking point is not whether something is raster. Here is the real sticking point:

http://www.awaresystems.be/imaging/tiff/tifftags/baseline.html

Any tags that are listed on this TIFF Baseline page, are not subject to modification or re-interpretation. To give an example, an TIFF page must be either Tiled or Striped, and if it is Tiled it must have certain required fields indicating the tile size and data locations, likewise if it is striped it must have some other required fields. The number of strip/tile must be a positive integer.

A file that has a TIFF header but violates the Baseline assumption, must not use the TIFF magic number (0x49 0x49 0x2a 0x00, or 0x4d 0x4d 0x00 0x2a) or else it would be properly called a "malicious TIFF image file". It could even be automatically deleted by virus scanners.

These are about the only "dogmatic" rules about TIFF. Beyond that, TIFF is quite a liberal container format.

Thanks,
Ryan Wong

rwong_002@hotmail.com <mailto:rwong_002@hotmail.com>

Disclaimer: The opinions expressed here are the views of the writer(s) and do not necessarily reflect the views and opinions of the writers’ employer(s). Neither the writer(s), nor the writers’ employers, make any claim as to the accuracy of the opinions expressed here.

This has been an interesting thread. The final decision is to include the image in the TIFF file and simply add the photon list as private meta data. We will use integration methods to achieve compression (integrating 1000 16-bit images gives a 2000:1 compression ratio) and the TIFF file will include photon lists from all images used in the integration. That will allow all TIFF readers to display the integrated image and it will allow my software (and any other software that knows how to read the photon lists) to use the photon lists to create images with shorter exposures if desired. Including the image also allows users to browse the hard drive when trying to find a specific set of images.

In the future, I may choose to compress images further by setting the bits/sample to the largest bit-depth in the image and then packing the image. So images that are not integrated will be 1-bit/sample images, which gives me another 8:1 compression ratio.

To: libtil