16bit works, but only on ATI hardware in QC (at least last time I tested it). The intel GPUs are probably limited to 8bit anyway. Both nvidia and ATI support 32f formats.
I work heavily with 16/32bit formats in QC without much problem. For safety, I suggest just using 32f for any processing, and dropping back to 16 to save if you need that (TIFF can save 32f formats, so I just pull out the 32f pixel data and write to that directly). You can test this easily, drop a >8bit image into the editor and connect it to a billboard. With a 12bit RAW file, the pixel format is listed as Internal_RGB16 for me. I've been processing them with CI/GLSL without any issues (now and then you'll want to put your billboards/glsl into a render in image, and set the format to 32bit or it'll drop back to 8bit, but that's the only real issue I've had). Chris On 9 April 2010 03:51, vade <[email protected]> wrote: > My understanding is that you are limited to 8bpc RGBA/BGRA, or 32bpc RGBA > "float" image formats (also, the intensity formats as well but). I don't > believe you can shim in your own types because the > QCPluginOutputImageProvider protocol requires use of those specified types, > and also, as you've ascertained, not all plugins will respect and output the > same format image out as they processed in. > I would suggest looking into building an Image Unit using Core Image / CIKL, > which can handle any input type (with some small caveats and gotchas on > certain hardware - but its generally speaking sane and 'just works' from my > limited experience) and building your processing chain "outside" of QC - QCs > limited support for anything other than 8bpc across the board makes working > at higher bit depth a bit more complicated especially if you want to build > or use custom 3rd party plugins. That plus float texture differences across > hardware may prove to be a royal pain to manage. Core Image should keep all > of those hardware specific GL nuances behind the scenes and the same code > should "just work". > Even my own for the most part output only an 8bpc input image regardless of > the input format. I know, tsk tsk - but its not 100% clear how to even test > for that case (determining bit depth of an input QCImage - there is no > accessor, you can kind of do it in GL using some gets, but Im not 100% if > thats "right"). > Good luck! > > On Apr 8, 2010, at 7:51 PM, Josh Imbruglia wrote: > > Hi list! > I am trying to use quartz composer for image processing on 16bit per > component images (RGB16) and was wondering if that's possible? When I use > openGL profiler on the patch and check the texture parameters it says that > they're loaded as an RGB8 for the internal format. The same thing happens if > I attempt to use a floating point image. Quartz plug-ins allows me to create > a floating point buffer to read back into but if the values in quartz are > all down sampled to 8bit then when I read the float values out of the buffer > they will be rounded off and incorrect. I believe this to be the case but > could be missing something. > I'm also wondering if it's possible to create custom property types, so > instead of using a id<QCPlugInInputImageSource> I could use my own type > custom image type. > Any help would be most appreciated, > Josh > _______________________________________________ > Do not post admin requests to the list. They will be ignored. > Quartzcomposer-dev mailing list ([email protected]) > Help/Unsubscribe/Update your Subscription: > http://lists.apple.com/mailman/options/quartzcomposer-dev/doktorp%40mac.com > > This email sent to [email protected] > > _______________________________________________ > Do not post admin requests to the list. They will be ignored. > Quartzcomposer-dev mailing list ([email protected]) > Help/Unsubscribe/Update your Subscription: > http://lists.apple.com/mailman/options/quartzcomposer-dev/psonice%40gmail.com > > This email sent to [email protected] > _______________________________________________ Do not post admin requests to the list. They will be ignored. Quartzcomposer-dev mailing list ([email protected]) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com This email sent to [email protected]
