Oops, thanks for the tip on the docs being out of date. I will guess that at 
some point we realized that this call needed to work for certain cases.


> On Jul 4, 2019, at 1:18 PM, Colin Doncaster <[email protected]> wrote:
> 
> Thank you for clarifying, I think this suits our needs jut fine fine and 
> keeping get_texture_info the more robust option seems fair, as long as it’s 
> clear.  
> 
> After posting yesterday I did find this in the UDIM section of the docs and 
> felt a little bad for asking, but I guess it’s changed?
> 
> "Please note that most other calls, including most queries for get  
> <page181image17295104.png>texture  <page181image17301248.png>info(), will 
> fail with one of these special filenames, since it’s not a real file and the 
> system doesn’t know which concrete file you it corresponds to in the absence 
> of specific texture coordinates."
> 
> I appreciate the detailed response. 
> 
> Colin
> 
> 
>> On Jul 4, 2019, at 4:05 PM, Larry Gritz <[email protected] 
>> <mailto:[email protected]>> wrote:
>> 
>> On Jul 3, 2019, at 12:04 PM, Colin Doncaster <[email protected] 
>> <mailto:[email protected]>> wrote:
>>> 
>>> Hello - 
>>> 
>>> I’m curious to know what the expected behaviour is when using 
>>> TextureHandles with filenames that include UDIM’s?
>>> 
>>> ie. if I create a texture handle with myTexture.<UDIM>.tif and 
>>> myTexture.1000.tif exists will this handle be successfully created and 
>>> good(myHandle) return true?  I realize that without knowing what 
>>> coordinates might potentially be sampled there still could be an out of 
>>> range UDIM value, but I’m looking for a first line of defence. 
>> 
>> Yes, if it's a special UDIM name, the initial 
>> get_image_handle/get_texture_handle will always succeed and appear to be 
>> good. It's not clear what it means to fail, until you give it specific 
>> coordinates that resolve to a specific real live file that can be present or 
>> missing or somehow broken.
>> 
>> Up for debate: should merely asking for the handle do a *search* 
>> (potentially expensive) for files that could match any u,v range with that 
>> filename template, and succeed if any one is found and fail if none at all 
>> are found? (See below, that's what get_image_info does.)
>> 
>> 
>>> Would this also be a valid way of testing to see if a texture exists and if 
>>> it’s usable by OIIO?  Or would using get_image_info on the image cache be 
>>> better?  If so would filenames with <UDIM> pose a problem?  
>> 
>> get_image_info/get_texture_info has the behavior described above -- it has 
>> to get the data you ask for from somewhere, but you haven't supplied a u,v. 
>> So it will look real files matching the template and return the information 
>> for the lowest-numbered one it can find. It assumes that you understand that 
>> if asking for the template name (like "foo_<UDIM>.exr") instead of a 
>> concrete specific name, any one file will do, and probably you will be in 
>> trouble anyway if important things like number of channels is not the same 
>> on all the udim tiles. But if NO files are found matching the template, the 
>> get_image_info/get_texture_info call will fail.
>> 
>> So you could use get_image_info to check for any metadata that will always 
>> be present, like number of channels, just to find out if any udim files 
>> exist at all based on that naming template.
>> 
>> 
>> If this is missing something important you need, please do let me know. None 
>> of this is set in stone, I'd be happy to beef up udim support.
>> 
>> 
>> --
>> Larry Gritz
>> [email protected] <mailto:[email protected]>
>> 
>> 
>> 
>> 
>> _______________________________________________
>> Oiio-dev mailing list
>> [email protected] <mailto:[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