Re: [Openexr-devel] Trouble creating deep scanline files

[Chip Collier]

That is helpful info! 

With the data coming from mantra I've been using Pz -> Z, C[0] -> R, C[1] -> G, C[2] -> B, C[4] -> A. But 'Of' I wasn't entirely sure what to do with so I just left the name as it is, unless it was written out with a tuple size > 1. In that case I went with Of.r, Of.g, Of.b.

Any other channel names are left named as they're found.

On Wed, Apr 17, 2013 at 3:05 PM, Peter Hillman <address@hidden> wrote:

Indeed, the EXR library does require the OutputFile objects to be destroyed cleanly to write a valid file. 'new'ing an OutputFile object and forgetting to delete it is a similar gotcha.

By the way, you should use "A" not "O" as the channel name for alpha/opacity: Nuke doesn't complain when trying to load the file, but the alpha channel is blank.
Internally, OpenEXR needs A to be alpha to be able to composite deep images into flat ones.

On 04/18/2013 05:59 AM, Chip Collier wrote:

Doh! Ok, the real offender in that code was the call to exit instead of just returning from main.

I believe this was preventing any threads from completely writing the file. I just started from some other tool and didn't even look at that. :) Thanks for the extra eyes!

On Wed, Apr 17, 2013 at 10:37 AM, Halfdan Ingvarsson <address@hidden> wrote:

On 13-04-17 01:17 PM, Chip Collier wrote:

Thanks Halfdan,

I've tried it both ways but get the same results.

Attached is the modifications I made to swap the x/y coordinates. Works fine for me after the changes.

Tested on 64-bit Linux.

Loads in Nuke, even :-)

 - ½

_______________________________________________
Openexr-devel mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/openexr-devel

_______________________________________________
Openexr-devel mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/openexr-devel