Adrian, FYI...
I tried a few tests this morning and found a special case that fails for the "ZIP 16 LINES" type. It throws a "TypeError: ord() expected a character, but string of length 0 found". I should say first of all that it worked great on *all* of the exr files that were created by Nuke: ZIP-1, ZIP-16, PIZ, RLE and B44. The one that *didn't* work was a ZIP-16 that was created by a VRay converter script: http://www.spot3d.com/vray/help/200R1/tools_vrimg2exr.htm However, those script-converted files *were* detected as ZIP-16 by your previous, more "brute force", script that allocates an arbitrary chunk of header data to read. It just so happens that we have a lot of older VRay renders that were converted to ZIP-16 with that vrimg to exr conveter script... (8^\ There must be something "non-standard" about the way the files are written, but not different enough to prevent Nuke's Read code from using them. In any case, thanks again for all your efforts, I really appreciate it - and I'm sure others do, too! Rich On May 22, 2012, at 7:28 PM, Adrian Baltowski wrote: > Hi > > In fact each channel name in exr can be 31 bytes long, so even 4096 could be > too less in some extreme cases. On the other side we don't want to read too > much. > The code below fix this issue: I did small modifications to read actual size > of channels list and skip over, no matter how many channels is in the file. I > added few comments to clarify. > > > ######################## > import struct > > compList = ['None', 'RLE', 'ZIP', 'ZIP 16 lines', 'PIZ', 'PXR24', 'B44', > 'B44A'] > > n = nuke.selectedNode() > file = nuke.filename(n, nuke.REPLACE) > fd = open(file, 'rb') > # read small portion of data > header = fd.read(256) > > # search for channels list tag. Well... in fact we could hardcode this > position but I prefer to write it more flexible > channels = header.find('channels') > > # skip the tags... > fd.seek(channels+16) > > # ...and read the actual size of channels list. We know, that it's 4-byte > integer > channelsSize = fd.read(4) > > # Ok, so we know how long (in bytes) is the channels list in our file. And we > simply jump over > fd.seek(struct.unpack('i', channelsSize)[0], 1) > > # Here we are > position = fd.tell() > > # read next portion of data and search for compression > header = fd.read(256) > index = header.find('compression') > fd.seek(position + index+28) > comp = ord(fd.read(1)) > print compList[comp] > > > ###################################### > > > > Best > Adrian > > > > >> Nathan, >> >> Thanks for the PDF link to the OpenEXR file layout - I was looking for >> something like that! >> >> Rich >> >> On May 22, 2012, at 1:23 PM, Nathan Rusch wrote: >> >>> The compression attribute will be in a different place within the header >>> depending on the number of channels in the file (due to the layout of the >>> EXR header). >>> >>> Adrian's snippet is simply finding the 'compression' attribute name within >>> the header, since that is one of the standard EXR attributes. From there, >>> based on known information about the EXR file layout, he can determine >>> where the actual compression attribute value is stored. >>> >>> The read size of 1024 is pretty much an arbitrary value; you may need to >>> read a larger chunk in order to get far enough into the file to locate the >>> compression attribute if your file has a lot of channels. If you know the >>> names of all the channels within the file, you could figure out exactly how >>> many bytes to read, or even start your read from a predefined offset to >>> keep your data buffer as small as possible; otherwise, you'll just need to >>> make a safe estimate. >>> >>> Check out this doc for a nice simple example of the structure of an EXR >>> file: http://www.openexr.com/openexrfilelayout.pdf >>> >>> Hope this helps. >>> >>> -Nathan >>> >>> >>> -----Original Message----- From: Dan Rosen >>> Sent: Tuesday, May 22, 2012 9:59 AM >>> To: Nuke user discussion >>> Subject: Re: [Nuke-users] How to check zip compression type for EXR >>> images...? >>> >>> This is great! Can you let us know what index number the datatype is >>> in? I'm finding it hard to tell. >>> >>> thx >>> Dan >>> >>> On Mon, May 21, 2012 at 7:30 PM, Richard Bobo <[email protected]> wrote: >>>> Adrian, >>>> >>>> Brilliant - I'll be making it into a nice little pulldown menu utility >>>> function! And, looking a bit deeper at your code, of course, so I can learn >>>> some more Python... 8^) >>>> >>>> Thanks! >>>> >>>> Rich >>>> >>>> Rich Bobo >>>> Senior VFX Compositor >>>> >>>> Mobile: (248) 840-2665 >>>> Web: http://richbobo.com/ >>>> >>>> "Man has been endowed with reason, with the power to create, so that he can >>>> add to what he's been given." >>>> - Anton Chekhov >>>> >>>> >>>> >>>> On May 21, 2012, at 6:40 PM, Adrian Baltowski wrote: >>>> >>>> Hi >>>> With just few lines of code and totally simplified >>>> >>>> ********************************** >>>> compList = ['None', 'RLE', 'ZIP', 'ZIP 16 lines', 'PIZ', 'PXR24', 'B44', >>>> 'B44A'] >>>> >>>> n = nuke.selectedNode() >>>> file = nuke.filename(n, nuke.REPLACE) >>>> fd = open(file, 'rb') >>>> header = fd.read(1024) >>>> index = header.find('compression') >>>> comp =ord(header[(index+28):(index+29)]) >>>> print compList[comp] >>>> >>>> *********************************** >>>> >>>> Each exr file MUST have compression info in the header and this info is >>>> placed just after channels info. It's simple to get actual size of channels >>>> list but I quickly set 1024 bytes of a headroom. >>>> >>>> Best >>>> Adrian >>> >>> _______________________________________________ >>> Nuke-users mailing list >>> [email protected], http://forums.thefoundry.co.uk/ >>> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >> >> _______________________________________________ >> Nuke-users mailing list >> [email protected], http://forums.thefoundry.co.uk/ >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >> > > _______________________________________________ > Nuke-users mailing list > [email protected], http://forums.thefoundry.co.uk/ > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users _______________________________________________ Nuke-users mailing list [email protected], http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
