[Radiance-general] Re: difference between ascii and binary output
Greg Ward
[email protected]
Fri, 20 Feb 2004 21:49:35 -0800
Hi Alexa,
The latest version of pvalue (from the HEAD distribution) includes a
-dF option to output byte-swapped floating point problems. If your
source and destination machines have different byte orderings (little
endian vs. big endian), then this might solve your problem. You also
should apply the -H option (in addition to the -h option) to avoid
output of the image resolution string, which will likely mess up the
reading of the data.
-Greg
> From: "Alexa I. Ruppertsberg"
> Date: February 20, 2004 5:25:52 AM PST
>
> Hi there,
>
> before I despair I ask the RADIANCE-world (maybe there is a difference
> even if there shouldn't be one):
>
> When writing out data with 'pvalue' you can write ascii or binary
> files.
> I chose 'pvalue -o -h -df ' for a binary file
> and 'pvalue -o -h ' for an ascii file.
>
> I then read both files into Matlab (under MS Windows): The ascii files
> are fine, they show the exact same values as when viewing the
> corresponding pic-image with ximage, or when just looking at the
> ascii-file itself. (The world is consistent! Great!)
>
> Then I read the binary file in (I have taken into account that the
> files were generated on a big-endian machine and are now read on a
> little-endian machine).
> And: The numbers that you yield look reasonable, similar to what you
> would expect, BUT they are not the same, even though the binary and
> the ascii file were generated from the very same pic-file.
>
> I looked at the minimum and the maximum values of either file, I
> computed the sum over all pixel values: they are just not the same!
> (just in case the matrix I deal with is in a different orientation)
>
> WHY???? Please help!
>
> Thanks!!!!!!!!!!!!!!!
>
> Alexa
>
> P.S. yes, I could use ascii-files, but they are huge and it takes
> longer to read them in. So, that's why I want to store them as
> binaries.