https://github.com/OpenImageIO/oiio/pull/658
It's a pretty big change to the DPX module, which is a fairly important file format for us, so it does make me a little nervous. It would be very helpful if you could build the branch referenced in the pull request (git pull https://github.com/lgritz/oiio lg-dpx) and give that a spin on a bunch of your files, that would give me a lot more confidence about committing it to master. -- lg On Aug 15, 2013, at 11:40 AM, Mark Boorer wrote: > Wow, you are a legend! Thank you so much for putting the time in to fix it! > > Cheers, > Mark > > On 15/08/2013 7:15 PM, "Larry Gritz" <[email protected]> wrote: > Update: It turned out that it was a bit of an overhaul of DPXOutput to get it > in shape to output subimage, but I dove in and did it last night. I also > found some bugs in the underlying implementation of libdpx. > > But I got success! -- I can write multi-image DPX files, read them back > correctly, as well as read the sample image you provided. Whew! > > I left it late last night working, but with some cleanup left to do. If I > can get some of my real work out of the way first, I'll try to get a chance > to tidy things up enough to send out a pull request today. Tonight at the > latest. > > -- lg > > > On Aug 14, 2013, at 12:51 PM, Mark Boorer wrote: > >> Hi Larry, It should contain an image with L and an image with R. The view of >> the plant in the center is also from a different camera in the R eye. >> >> I'm not sure if Nuke allows for subimages in DPX files, and I've never been >> able to view these images in Nuke. (They are able to be read by our own >> grading / playback software). >> >> It could be that our library is doing things wrong, but not having an >> alternative library to write a stereo DPX, I was unable to test it ;) >> >> To answer your previous question, your initial summation was correct. I am >> unable to read or write stereo DPX's with the OpenImageIO DPX plugin >> (assuming my stereo DPX is valid). >> >> >> Thanks! >> Mark >> >> >> On Wed, Aug 14, 2013 at 8:34 PM, Larry Gritz <[email protected]> wrote: >> I loaded the image in Nuke, added a "SideBySide" node between the read node >> and the viewer, and what I see in the preview window is side by side images >> where they both have a big "L", neither has "R". >> >> Can you confirm? >> >> >> >> On Aug 14, 2013, at 12:18 PM, Larry Gritz wrote: >> >>> Aha, I misunderstood. I thought you meant that you could neither read nor >>> write. >>> >>> The underlying "libDPX" looks like it should support multiple image >>> elements (that's what DPX calls subimages) for both reading and writing. >>> >>> The dpxinput.cpp code looks like it supports reading multiple elements, but >>> although the test file you sent is reported as having two elements, they >>> both look the same (with an "L", I presume the other is intended to have >>> "R"?). I don't yet know if that is because we have a bug, or because your >>> file has two identical elements. Can you verify on your end? Do you have an >>> app that can view the two subimages and be sure that they are actually >>> different? >>> >>> The dpxoutput.cpp is, indeed, written in such a way that it ignores the >>> possibility of multiple elements when writing the file, but I think it's a >>> pretty simple matter to fix it up. Let me take a stab at it. >>> >>> >>> >>> On Aug 14, 2013, at 11:09 AM, Mark Boorer wrote: >>> >>>> Thanks again Larry, >>>> >>>> Just confirming though, this is for reading sub image DPX's. The writing >>>> is still very much unimplemented? >>>> >>>> Cheers, >>>> Mark >>>> >>>> On 14/08/2013 6:30 PM, "Larry Gritz" <[email protected]> wrote: >>>> Thanks for the test case. I briefly glanced at the code yesterday, and >>>> it's structured as if it expects subimages to work in the DPX code. So >>>> it's probably just a bug, not a case of it being written without >>>> considering subimages at all and needs a total rewrite. I'll try to take >>>> a quick look and see if it's something obvious, I'll let you know what I >>>> find. >>>> >>>> -- lg >>>> >>>> >>>> On Aug 14, 2013, at 1:34 AM, Mark Boorer wrote: >>>> >>>>> Hi Larry, >>>>> >>>>> Thanks for your quick reply and willingness to take a look. >>>>> I've attached an example file (its a test plate so no copyright issues). >>>>> It was written using an internal, quite old library, and is fine for >>>>> reading back by said library (it also seems okay in RV). I'm looking to >>>>> transition new code over to OIIO, so as I mentioned, if you don't have >>>>> time to tinker with it, I'm happy to take a stab :) >>>>> >>>>> File can be downloaded from here: >>>>> https://docs.google.com/file/d/0B7tMukhco9MwVjZWa09ZWW04Vm8/edit?usp=sharing >>>>> >>>>> Thanks! >>>>> Mark >>>>> >>>>> >>>>> On Tue, Aug 13, 2013 at 8:20 PM, Larry Gritz <[email protected]> wrote: >>>>> Nobody is working on it as far as I know. We use the DPX support >>>>> extensively here, but I think our convention is to save L & R as separate >>>>> images, so it hadn't been brought to my attention that our DPX subimage >>>>> support was lacking. A fix for that would be very welcome. >>>>> >>>>> But before you dive in too deep... if you want to send me (privately is >>>>> ok, I won't share) one of these stereo DPX files, I would be happy to >>>>> take a *quick* look at the DPX code -- which I didn't write, but am at >>>>> least passingly familiar with -- just to see if it's an extremely minor >>>>> alteration (or even an obvious bug) to fix subimages, and if so I can >>>>> just bang it out. But if it's more extensive, I'll let you know because >>>>> if I don't know immediately how to do it, you'll probably get a fix >>>>> faster if you are able to do it yourself rather than wait for me to free >>>>> up a bigger block of time. >>>>> >>>>> -- lg >>>>> >>>>> >>>>> On Aug 13, 2013, at 11:50 AM, Mark Boorer wrote: >>>>> >>>>> > Hi, >>>>> > >>>>> > I'm looking to read/write some stereo DPX files I have, and I noticed >>>>> > that the DPXOutput class explicitly does not support subimages. >>>>> > I've also tried reading some files that contain subimages (written >>>>> > from a custom library), but I'm unable to get the data out of them >>>>> > after calling seek_subimage() (though the seek call works as >>>>> > expected). >>>>> > >>>>> > Is anyone presently working on supporting subimages in the DPX writer? >>>>> > I'm not too familiar with the guts of the format, but I do have access >>>>> > to the spec and may be motivated to add support. >>>>> > >>>>> > Cheers, >>>>> > Mark >>>>> > >>>>> >>>>> -- >>>>> 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 >>>> >>>> _______________________________________________ >>>> Oiio-dev mailing list >>>> [email protected] >>>> http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org >>> >>> -- >>> Larry Gritz >>> [email protected] >>> >>> >> >> -- >> 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 > > _______________________________________________ > 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
