https://github.com/OpenImageIO/oiio/pull/558


On Apr 2, 2013, at 5:59 PM, Matt Chambers wrote:

> Currently we're pulling from master...which I think its advertising itself as 
> 1.1.  I'm sure you could knock out the --chnames flag faster than me, 
> specially if your doing a big refactor.  I'll work in getting those verbose 
> messages in place.  Thanks
> 
> 
> On Wed, Apr 3, 2013 at 3:43 AM, Larry Gritz <[email protected]> wrote:
> Are you guys concerned primarily with master, or are you using 1.1?
> 
> maketx is in transition right now; we're switching to a new code path 
> (internal to libOpenImageIO, rather than all the good bits just in the maketx 
> binary, and quite overhauled for performance improvements), so it's a bit 
> different in master and 1.1.
> 
> In this instance, it may be easier for me to make the --chnames change, 
> unless you really had your heart set on trying it yourself.   (If you want to 
> take a crack at something, the verbose messages thing you mentioned might be 
> a good toe in the water.)
> 
> 
> On Apr 2, 2013, at 12:52 AM, Matt Chambers wrote:
> 
>> Ha, I didn't see that oiiotool option, I'm going to use that for now.
>> 
>> I think think for most cases that is a good assumption...for a particular 
>> internal case we need it in R.  I'd love for there to be no extra step 
>> because then it makes this really easy for all departments when every single 
>> texture publish can be done with a single maketx command.
>> 
>> I'd love to give it a try. Would I add a --chnames option that would just 
>> override the channel names for any operation?
>> 
>> -Matt
>> 
>> 
>> On Tue, Apr 2, 2013 at 8:36 PM, Larry Gritz <[email protected]> wrote:
>> Fundamentally, I think the problem is with ImageSpec::default_channel_names 
>> ()  (in src/libOpenImageIO/formatspec.cpp).  There's an assumption that if 
>> channels aren't named (hello, TIFF) and there's only one channel in an 
>> image, it will be "A" in the absence of any other information.  I dunno, 
>> maybe that's a bad assumption, but nobody every complained before.
>> 
>> oiiotool --chnames can be used after the fact to rename channels.  If you 
>> don't want that extra step, we could probably put a similar option in maketx 
>> directly.
>> 
>> Yes, we would welcome the patch you describe (or any other).
>> 
>>      -- lg
>> 
>> 
>> On Apr 2, 2013, at 12:25 AM, Matt Chambers wrote:
>> 
>>> Yeah, pretty much.  The input image is 1 channel, I need it to go into R, 
>>> not A...an internal thing.  I've poked around the code a bit but I can't 
>>> tell where its happening.  I'd be happy to add an option if you can point 
>>> me to the right place.
>>> 
>>> Also, somewhat related, my colleagues have been using (or trying to use) a 
>>> lot of the optimization options like monochrome-detect and opaque detect, 
>>> but there is actually a few other checks maketx does before enabling those 
>>> options.  Can I push you a patch where in verbose mode it would tell you if 
>>> those things were ignored because they conflicted with another option or 
>>> the number of channels in the source image was too high?
>>> 
>>> -Matt
> 
> --
> Larry Gritz
> [email protected]
> 
> 
> 
> _______________________________________________
> Oiio-dev mailing list
> [email protected]
> http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
> 
> 
> _______________________________________________
> Oiio-dev mailing list
> [email protected]
> http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org

--
Larry Gritz
[email protected]


_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org

Reply via email to