Just so you know, I haven't merged that fix yet! (And when I do, it'll only be 
in the master branch.) You'll need to use the other suggested solutions I 
mentioned in the original email.

        -- lg


On Mar 30, 2015, at 11:22 AM, Sebastian Kral <[email protected]> wrote:

> Hi Larry,
> 
> that's awesome. I still didn't have the chance to test this. Thank you. I 
> will tey to ger to it tomorrow.
> 
> Cheers,
> Sebastian
> 
> On Monday, March 30, 2015, Larry Gritz <[email protected]> wrote:
> As a follow-up, I have implemented the -trim (explicit trim in place) that I 
> mentioned on this thread would have been convenient.
> 
>  https://github.com/OpenImageIO/oiio/pull/1105
> 
> 
> On Mar 26, 2015, at 10:13 AM, Larry Gritz <[email protected]> wrote:
> 
>> How about this (presuming it's a 4 channel RGBA):
>> 
>>      oiiotool in.exr -dup -mulc 10000 -clamp:min=0:max=1 -ch R=A,G=A,B=A,A 
>> -mul -autorim -o out.exr
>> 
>> This reads in.exr, duplicates it on the stack, multiplies pixel values 10000 
>> then clamps to [0,1] (that has the effect of turning all nonzero alpha into 
>> 1.0), replicates the thresholded alpha value into all 4 channels, multiplies 
>> that mask by the original image (still below it on the stack!) -- now we've 
>> squashed all color to 0 in pixels that were original alpha 0 -- and finally 
>> write it out with autotrim!
>> 
>> I haven't tested this, so take with a grain of salt that I've spelled 
>> everything correctly and whatnot. But I think the overall methodology is 
>> sound.
>> 
>> If you are using master (so you have expression substitution), another 
>> clever trick might be:
>> 
>>      oiiotool in.exr --ch A -autotrim -o alphatrim.exr
>>      oiiotool alphatrim.exr in.exr -crop {IMG[1].geom} -o out.exr
>> 
>> This extracts just the alpha channel and saves to a temporary file with 
>> auto-trimming. Then it uses the size of the saved auto-trimmed alpha to 
>> explicitly crop the original. I would prefer to do it in one step, but it 
>> looks like --autotrim only works upon the saved output of "-o", and I 
>> neglected to implement a --trim (trim the outer black "in place" so you can 
>> then use the trimmed image).
>> 
>> 
>> 
>> On Mar 25, 2015, at 10:42 PM, Jason Iversen <[email protected]> wrote:
>> 
>>> Perhaps you can premultiply (--premult) before autotrimming?
>>> 
>>> On Wed, Mar 25, 2015 at 8:48 PM, Sebastian Kral <[email protected]> 
>>> wrote:
>>> Hi,
>>> 
>>> I was wondering if it's possible to create a DataWindow in an EXR with the 
>>> --autotrim flag which is based purely on the Alpha instead of black pixels.
>>> We have some rather large mattepaintings which I am trying to trim but 
>>> since the background isn't black the autotrim doesn't seem to work.
>>> 
>>> Perhaps I am missing something and someone can help me out.
>>> 
>>> The command I am running:
>>> oiiotool in.exr --scanline --compress zips --autotrim -o out.exr
>>> 
>>> Thank you.
>>> 
>>> Cheers,
>>> Sebastian
>>> 
>>> _______________________________________________
>>> Oiio-dev mailing list
>>> [email protected]
>>> http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
>>> 
>>> 
>>> 
>>> 
>>> -- 
>>> jason iversen
>>>   http://www.odforce.net
>>> _______________________________________________
>>> 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
> 
> --
> Larry Gritz
> [email protected]
> 
> 
> 
> 
> 
> -- 
> Sent from mobile device
> _______________________________________________
> 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