Re: [Openexr-devel] Header Size?

[Florian Kainz]

Hi Yanko,

the OpenEXR file header is an arbitrarily long sequence of
"attributes", followed by a single null byte.  An attribute
never begins with a null.  This allows you to parse the file
header incrementally: keep reading attributes until you find
one that appears to begin with a null byte.

Are you sure you have to write an OpenEXR file parser in
Python?  Python modules that read and write OpenEXR files
using ILM's C++ library are available online, for example
at http://excamera.com/sphinx/articles-openexr.html.

Florian



Yanko Sanchez wrote:
> I'm trying to read exr's using python's struct module. This is my first 
> time reading binary files and the PDF at the openexr has been very 
> helpful but I am a little puzzled on how to figure out what is the 
> header size with its 'optional' attributes... For example, nuke includes 
> a nuke hashstring but photoshop doesn't... before getting to the 
> hashstring I don't think I can tell if the file came form nuke or 
> photoshop... and if someone used their own plugin to create the EXR then 
> the header might also contain other data that might throw of my reader. 
> Is there a way to know where the header ends so I can skip all those 
> bytes and just go straight to the offset of scanlines and pixel data? 
> According to the docs there is a null byte at the end of the header, but 
> there are lots of null bytes at the end of a lot of attributes. Is there 
> a common way to dealing with this?
>
> Once I get through the header I am able to read a float uncompressed 
> scanline EXR file and display the data using matplotlib successfully. 
> Any help would be appreciated!
>
> Yanko
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Openexr-devel mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/openexr-devel