[3dem] Data storage and compression

[Takanori Nakane]

Hi,

The dynamic range of electron counts in individual
frames is very small (4 or 8 bits). In contrast, the
gain reference is basically random floating point
numbers. Since each frame is compressed independently in TIFF,
the algorithm cannot realize that the gain value is common
over frames in the same physical pixel. This is why compression
is so bad.

Indeed, if the movie frames are transposed such that the time
axis becomes the continuous (fastest changing) axis
in the 3D array, they compress better. Nonetheless, the algorithm
still cannot use the fact that the gain reference is common
among movies. Moreover, this transposition makes is impossible to
decompress only one frame.

Best regards,

Takanori Nakane

On 2019/08/29 21:50, Matthew H. Cahn wrote:
> By the way, why don’t gain-corrected images compress well?  Doesn’t 
> gain-correction just change the dynamic range of the pixels?
> 
> Matthew
> 
> *||  Matthew Cahn  |  Linux Administrator  |  Dept. of Molecular Biology 
> / Research Computing  |  Princeton University  |  (609) 258-5404  | 
> mcahn at princeton.edu <mailto:mcahn at princeton.edu>  ||
> *
>