|
From: Peter G. <jpg...@gm...> - 2008-11-13 18:43:16
|
On Thu, Nov 13, 2008 at 1:28 PM, Roman Himmes <rom...@po...> wrote: > Hi, > > a few Months ago I was developing my own platform independed webcam > library, before I found this library. > At least the Mac version was very similiar to your version because > both take the sequenceGrabber and extract the video > picture by picture form the GWorld with the deprecated functions > GetGWorldPixMap and Un-/LockPixels. > > This is the way the QuickTime Sample Hack TV Carbon Example (http://developer.apple.com/samplecode/HackTVCarbon/index.html > ) or > BrideOfMungGrab Example (http://developer.apple.com/samplecode/BrideOfMungGrab/index.html > ) > has done it as well. > > Now as I understood the GWorld concept has been replaced by the new > CoreImage system, but I don't find any functions to simply get the > same functionality with CoreImage than the GetGWorldPixMap and Un-/ > LockPixels functions. > > Now my question: > > Is there somebody working on this? Or should I investigate further? I am not doing any libvidcap development at the moment. > I am not happy with the performance of the current version, since it > takes more cpu load than the Windows Version. Perhaps the Locking and > Unlocking of the GWorld is too expensive and there is a better way. What Apple would like everyone to do is upgrade to 10.5 and use their new Cocoa-based QtKit Capture framework to do video capture. That framework obviates all of the Carbon-based sequence grabber and video digitizer code used in libvidcap. Two problems: 1) Nobody has done the work to add QtKit capture as a source api for libvidcap 2) Only OS X 10.5 or later users will be able to take advantage of this sapi. It should be possible to have both exist in libvidcap and determine at runtime which to use though. Pete |