Try to open the multimediafile ‘moviefile’ for playback in onscreen window
‘windowPtr’ and return a handle ‘moviePtr’ on success.
This function requires the [GStreamer](GStreamer) multi-media framework to be installed on
The following movie properties are optionally returned: ‘duration’ Total
duration of movie in seconds. ‘fps’ Video playback framerate, assuming a linear
spacing of videoframes in time. There may exist exotic movie formats which don’t
have this linear spacing. In that case, ‘fps’ would return bogus values and the
check for skipped frames would report bogus values as well. ‘width’ Width of the
images contained in the movie. ‘height’ Height of the images.
‘count’ Total number of videoframes in the movie. Determined by counting, so
querying ‘count’ can significantly increase the execution time of this command.
‘aspectRatio’ Pixel aspect ratio of pixels in the video frames. Typically 1.0
for square pixels.
If you want to play multiple movies in succession with lowest possible delay
inbetween the movies then you can ask PTB to load a movie in the background
while another movie is still playing: Call this function with the ‘async’ flag
set to 1. This will initiate the background load operation. After some
sufficient time has passed, you can call the ‘OpenMovie’ function again, this
time with the ‘async’ flag set to zero. Now the function will return a valid
movie handle for playback.
If all your movies have exactly the same format and only differ in duration and
content, but not in image size, color depth, encoding format, or fps, then you
can also use an aync setting of 2 and provide the ‘moviePtr’ handle of an
already opened movie in the ‘preloadSecs’ parameter. This will queue the movie
‘moviefile’ as a successor to the currently playing moviefile in ‘moviePtr’.
Queuing movies this way is more efficient than async flag setting 1, although
also more restricted.
If the ‘async’ flag also contains the number 4 or is equal to 4, then movie
playback will not automatically drop video frames to preserve audio-video sync
in case fetching and display of video frames by your script is delayed or too
slow. This has the disadvantage that you’ll need to take care of audio-video
sync and framerate control yourself by proper comparison of movie presentation
timestamps and GetSecs or Screen(‘Flip’) timestamps. The advantage is, that
after start of playback the playback engine can internally predecode and buffer
up to ‘preloadSecs’ seconds worth of video and audio data. This may allow
complex movies to play more smoothly or at higher framerates.
‘preloadSecs’ This optional parameter allows to ask Screen() to buffer at least
up to ‘preloadSecs’ seconds of the movie. This potentially allows for more
stutter-free playback, but your mileage may vary, depending on movie format,
storage medium and lots of other factors. In most cases, the default setting is
perfectly sufficient. The special setting -1 means: Try to buffer the whole
movie. Caution: Long movies may cause your system to run low on memory or disc
space and have disastrous effects on playback performance! Also, the exact type
of buffering applied depends a lot on the movie playback engine and movie
format, but it usually affects the buffering behaviour and capacity of buffering
in some meaningful way.
‘specialFlags1’ Optional flags, numbers to be added together: 1 = Use YUV video
decoding instead of RGBA, if supported by movie codec and GPU - May be more
efficient. 2 = Don’t decode and use sound - May be more efficient. On Linux you
may need to specify a setting of 2 if you try to use movie playback at the same
time as PsychPortAudio sound output, otherwise movie playback may hang. A flag
of 4 will draw motion vectors on top of decoded video frames, for debugging or
entertainment. A flag of 8 will ask the video decoder to skip all B-Frames
during decoding to reduce processor load on very slow machines. Not all codecs
may support flags 4 or 8, in which case these flags are silently ignored. A flag
of 16 asks Screen to convert all video textures immediately into a format which
makes them useable as offscreen windows, and for the Screen(‘TransformTexture’)
function as well as for drawing them with your own custom GLSL shaders. Normally
this conversion would be deferred until needed, ie. it would get skipped if you
would just draw the texture regularly. If you know already that you want to use
the texture with one of the given functions, manually triggering the conversion
via this flag may be a bit more efficient - or convenient if you want to use
your own GLSL shaders.
The optional flags 32, 64 and 128 influence how looped playback is performed if
usercode requests such repetitive playback via Screen(‘PlayMovie’, …) with the
‘loop’ flag set to one. Different strategies exist to handle different quirks
with some movie file formats and encodings and some versions of [GStreamer](GStreamer): A
flag of 32 requests looped playback via gapless reloading of the movie instead
of rewinding it to the start. A flag of 64 uses so called segment seeks for
rewinding, a flag of 128 asks to flush the video pipeline during rewinding. Your
mileage with these looping strategies will differ, but usually the default
settings are good enough for most purposes.
A ‘specialFlags1’ setting of 256 will prevent automatic deinterlacing of video.
This is useful to prevent some internal color data conversions, e.g., of pure
grayscale data, which can cause slightly lossy decoding of lossless video data.
A ‘specialFlags1’ setting of 512 marks the movie as encoded in Psychtoolbox’s
own proprietary 16 bpc high precision format. Grayscale movies in this format
can be created by specifying the keyword UsePTB16BPC in Screen(‘CreateMovie’) or
in the firewire videocapture engine as part of the codec spec string. RGB movies
can also get created this way. Encoding or decoding of such 16 bpc movies with a
channel count other than 1 or 3 for gray or RGB is not supported.
A ‘specialFlags1’ setting of 1024 tells the movie playback that this movies
video frames are encoded as raw Bayer sensor data and that they should get
converted to RGB images during playback via software Bayer filtering. You must
set the ‘pixelFormat’ parameter to 1 for this to work. You can choose the Bayer
filtering method via ‘DebayerMethod’ setting and the color sensor filter pattern
via ‘OverrideBayerPattern’ setting in Screen(‘SetVideoCaptureParameter’, -1,
…). By default, fast nearest neighbour debayering with an assumed sensor image
layout of RGGB is performed.
‘pixelFormat’ optional argument specifying the pixel format of decoded video
frames. Not all possible valid values are supported by all video codecs,
graphics cards and operating systems. If an unsupported format is requested,
Screen() will try to choose the closest matching format that meets or exceeds
the specified format, at a performance or efficiency penalty. If no sufficiently
close match is possible without severely degraded performance or other
restrictions, the function will abort with an error. The following formats are
supported on some setups: 1 = Luminance/Greyscale image, 2 = Luminance+Alpha, 3
= RGB 8 bit per channel, 4 = RGBA8, 5 = YUV 4:2:2 packed pixel format on some
graphics hardware, 6 = YUV-I420 planar format, using GLSL shaders for color
space conversion on suitable graphics cards. 7 or 8 = Y8-Y800 planar format,
using GLSL shaders, 9 = 16 bit Luminance, 10 = 16 bpc RGBA image.The always
supported default is ‘4’ == RGBA8 format. A setting of 6 (for color) or 7/8 (for
grayscale) for selection of YUV-I420/Y8-Y800 format, as supported by at least
the H264 and HuffYUV video codecs on any GPU with shader support, can be
especially efficient for fast playback of high resolution video. As this format
uses shaders for post-processing, it should be fast for texture drawing, but can
incur significant overhead if you try to draw into a texture of this format, or
try to post-process it via Screen(‘TransformTexture’). If you try to attach your
own shaders to such a texture during Screen(‘DrawTexture’), you will need to
implement color conversion yourself in your shaders, as your shaders would
override Screen’s builtin color conversion shader.
‘maxNumberThreads’ Optional parameter which allows to set the maximum number of
parallel processing threads that should be used by multi-threaded video codecs
to decode the movie. The parameter has no effect on single threaded codecs and
default behaviour is to let the codec do whatever it wants. A setting of zero
tells the codec to use multi-threaded decoding with a number of threads that is
auto-selected to be optimal for your given computer. A number n greater zero
asks the codec to use at most n threads for decoding. The most safe choice is to
not specify this parameter - this should work even with problematic movie
formats. If you need higher playback performance, e.g., for high resolution
video or high framerate playback, you should set the parameter to zero to allow
the optimal choice to the video codec. This should work flawlessly with well
encoded high quality movie files and can provide a significant performance boost
on multi-core computers. Specify a discrete non-zero number of threads if you
want to benefit from multi-core decoding but want to prevent movie playback from
using up all available computation power, e.g., because you want to run some
other timing-sensitive tasks in parallel and want to make sure to leave some
processor cores dedicated to them.
‘movieOptions’ Optional text string which encodes additional options for
playback of the movie. Parameters are keyword=value pairs, separated by three
colons ::: if there are multiple parameters. Currently supported keywords:
AudioSink=GStreamerSinkSpec – GStreamerSinkSpec is a [GStreamer](GStreamer) gst-launch line
style specification for a audio sink plugin and its parameters. This allows to
customize where the audio of a movie is sent during playback and with which
parameters. By default, the autoaudiosink plugin is used, which automatically
chooses audio output and parameters, based on your system and user settings.
Most often this is what you want. Sometimes you may want to have more control
over outputs, e.g., if your system has multiple sound cards installed and you
want to route audio output to a specific card and output connector. Example use
of the parameter: ‘AudioSink=pulseaudiosink device=MyCardsOutput1’ would use the
Linux pulseaudiosink plugin to send sound data to the output named
‘MyCardsOutput1’ via the PulseAudio sound server commonly used on Linux desktop
If you set a Screen() verbosity level of 4 or higher, Screen() will print out
the actually used audio output at the end of movie playback on operating systems
which support this. This can help debugging issues with audio routing if you
don’t hear sound.