~registry/cinelerra-cv/master : changes

~registry/cinelerra-cv/master » Changes from revision 1044

From Revision 1044 to 1025

Rev	Summary	Authors	Date
1044	Merge branch 'bezier_patch' into construction_site Merge branch 'bezier_patch' into construction_site	Simeon Voelkel	12 years ago
1043	Use replacement for in-kernel videodev.h in v4l and buz driv... Use replacement for in-kernel videodev.h in v4l and buz drivers This change is required to compile cinelerra on 2.6.38+ kernels	Rafael Diniz	12 years ago
1042	Do not install ffmpeg's header files. Do not install ffmpeg's header files. The internal ffmpeg is not intended to be used for development of third- party software. We only want its libraries. Do not install the header files so that an existing ffmpeg installation is not disturbed. The libraries are already installed with a customized name so that the do not interfere.	Johannes Sixt	13 years ago
1041	Move Cinelerra to version 2.1.5CV Move Cinelerra to version 2.1.5CV Updated credits (now continuing on in a second column) and copyright year to 2010.	Simeon Voelkel	13 years ago
1040	Change return type of Thread:get_tid() Change return type of Thread:get_tid() pthread_t is defined as unsigned long, on some architectures; it cannot be silently converted to int.	Einar Rünkaru	13 years ago
1039	Eliminate useless indexes from YUV4MPEG2 file loader. Eliminate useless indexes from YUV4MPEG2 file loader. The YUV4MPEG2 loader was building a from-scratch transient index that required reading every byte in the raw file on every call to open. Open is called often, not just when the project is loaded. If the loader already had an index, it threw it away. Indexes existed only in memory, nothing was saved. This means that, for example, every undo required every y4m file in the project to rebuild indexes from scratch. On the movie I was working on, that was 20 minutes as it re-read just short of a terabyte of video. There is no practical reason for the index. In yuv4mpeg2 every frame (including the frame header) is a fixed size. Although the spec references theoretical support for a variable number of variably sized flags per frame header, the only actual flags defined are for interlacing order, which must appear in every frame and are all the same size. Thus, yuv4mpeg2 frame size does not vary. Remove the YUV4MPEG indexing code completely; rework open to determine file header and frame size in a robust fashion; rework seek to jump directly to a computed offset. This change, in my current project, reduces 'open' and 'undo' time from ten minutes to instantaneous. It also eliminates the bug where compositor playback hits a new track and the entire UI becomes unresponsive until the open call is finished rebuilding the transient index.	Monty	13 years ago
1038	Check error codes in OSS audio driver. Check error codes in OSS audio driver. In several places in the OSS driver, return codes are either not checked for error, or are checked, an error is printed, and the driver continues blindly on with invalid state. Performed a complete review of the driver backend and for all syscalls and ioctl()s, both check the return code, and also error out with consistent state if an error occurs. This corrects numerous observed crash-or-'destroy the speakers' bugs in the OSS backend, such as a crash when the configured device is unavailable, or full-amplitude noise output when device setup fails.	Monty	13 years ago
1037	Fix OSS latency calculation. Fix OSS latency calculation. Quoting the original 4front OSS documentation re. SNDCTL_DSP_GETOPTR: "This ioctl call should only be used when using mmap(). In normal applications it's practically useless." As it does in fact not work for playback through the Linux ioctl() interface, go to the more complex GETODELAY route, which does work. Long ago GETODELAY wasn't supported everywhere (eg, the BSDs' independent implementation of OSS circa 2000) but I believe that's no longer true. Also add locking to latency calculation, as the latency calculation process is not atomic. This uses a new mutex rather than blindly reusing one of the preexisting mutexes.	Monty	13 years ago
1036	Fix videos with row length not multiple of 16 Fix videos with row length not multiple of 16 This fixes bug 524: videos with row lengths that are not a multiple of 16 were completely garbled. This pads the row lengths to the next multiple of 16.	Einar Rünkaru	13 years ago
1035	Fix build with libx264 version 76 and later. Fix build with libx264 version 76 and later. Alexis Ballier points us to this announcement in the libx264 API: http://mailman.videolan.org/pipermail/x264-devel/2009-September/006337.html	Monty	13 years ago
1034	Correct a rounding bug in the compositor fade engine. Correct a rounding bug in the compositor fade engine. This bug caused occasional skipped frames in the fade sequence, resulting in jerky fades. Original commit by Monty.	Simeon Voelkel	13 years ago
1033	Add [trivial] support for all non-8-bit-RGB png types in the... Add [trivial] support for all non-8-bit-RGB png types in the VFrame abstraction.	Monty	13 years ago
1032	Add return code checking to ALSA backend. Add return code checking to ALSA backend. This avoids that failed HW setup no longer crashes / blows the speakers out.	Monty	13 years ago
1031	Do not use outdated pixel format names in ffmpeg code. Do not use outdated pixel format names in ffmpeg code. This fixes compilation with modern ffmpeg. Patch by Alexis Ballier (Gentoo).	Simeon Voelkel	13 years ago
1030	Initialized edit_font_color in S.U.V theme Initialized edit_font_color in S.U.V theme This variable was not initialized in S.U.V theme Initialized it as Yellow.	Einar Rünkaru	13 years ago
1029	Check OpenGL server version at runtime. Check OpenGL server version at runtime. According to http://www.opengl.org/sdk/docs/man/xhtml/glXCreatePbuffer.xml "glXCreatePbuffer is available only if the GLX version is 1.3 or greater." Note that this is about GLX version. This means, that on the client side a version of OpenGL greater than 1.3 can be installed (and in fact must be otherwise, Cinelerra would not even start), but on the server side the GLX version can still be lower. If OpenGL version is less than 1.3 don't show OpenGl selection in preferences..playback.	Einar Rünkaru	13 years ago
1028	fix background rendering crash on >=jpeg-7 (patch adopted fr... fix background rendering crash on >=jpeg-7 (patch adopted from gentoo) This commit fixes crashes of the background rendering threads due to problems with too recent (lib)jpeg versions (only the first frame got rendered before a backtrace on the terminal indicated the crash of the renderfarm). On versions of jpeg more recent than 6b have an automatically enabled fancy up/downsampling feature that has to be disabled manually if the calling application does not support that feature, as cinelerra does. jpeg versions before 7 do not have these features which prohibits compilation without version checks.	Simeon Voelkel	13 years ago
1027	Correct a calling error of sws_scale(). Correct a calling error of sws_scale(). cinelerra/ffmpeg.C passed sws_scale an incorrect starting slice y, causing most output pixel format conversions to fail.	Monty	13 years ago
1026	Change wording of the option that disables hardware synchron... Change wording of the option that disables hardware synchronization. The option description for disabling hardware playback/record sync information is exceedingly confusing; I thought it mean thte exact opposite before reading the code, and several others have said the same thing. Make it clear the option DISABLES hardware playback sync and instead relies on a blind software timer. The current description makes it sound as if it enables sync tracking.	Monty	13 years ago
1025	apply gentoo pkgconfig x264 patch to configure.in apply gentoo pkgconfig x264 patch to configure.in	Simeon Voelkel	13 years ago

Older »