1 \chapter{Performance and other Tips}%
2 \label{cha:performance_tips}
4 Performance of \CGG{} is related to the software and video format being used in conjunction with your computer system hardware -- the number of CPUs and its speed, I/O bus speed, graphics card, and amount of available memory. A basic, less powerful system will be sufficient for users working with audio only or lower resolution video formats. Higher end computers will be needed when playing and working with higher resolution formats, like 1080p or 4k. Adding effects and multiple tracks will require more cpu, memory, and various other resources to
5 perform at an acceptable level.
7 Perhaps the easiest method for determining if your performance could be improved is to look at the numerical value displayed as \textit{Framerate achieved}. Good performance means that when \textit{Play every frame} is set
8 in \texttt{Settings $\rightarrow$ Preferences, Playback A} tab, the frames/second (frames per second or fps) in playback might be almost always at the maximum rate of your project setting and/or video frame rate. You can check this in \texttt{Settings $\rightarrow$ Preferences, Playback A}, by watching \textit{Framerate achieved} while playing forward. A higher number is better, up to the format frame rate of the video.
10 Some computer hardware factors to consider for better performance are listed here:
12 \item Multi-core and more SMP processors greatly improve speed by making use of threads. \CGG{} automatically uses the available threads, but some processes are single-threaded anyway and these become the weak link in the chain. The \textit{Project SMP cpus} parameter is used to limit the use of threads in the deconding (playback) stage only. It is better to lower the number of threads to leave some for the system and for the plugins in use.
13 \item RAM: In video editing in general, the more RAM the better. A large amount of free memory available can help speed up operations by avoiding unnecessary disk swaps and keeping videos easily accessible in memory.
14 You can optimize RAM utilization with \textit{Cache size} and \textit{Seconds to preroll render} parameters. See section \ref{sec:cache_and_preroll}. If there is limited RAM you must necessarily have a large swap partition. For system swap, (2 x RAM) GB seems to be more than sufficient. If the amount of memory being used by the program is close, then swap might save you but often if swapping becomes necessary, it presents more problems and you end up killing the \CGG{} process anyway.
15 \item Video editing is almost always I/O intensive. To create longer running videos at high resolution
16 you will want to have a lot of disk space available on fast access disks. The higher the transfer rate, the less slowdowns and problems. So a modern SSD is better than an old HDD.
17 \item \CGG{} benefits from OpenGL hardware acceleration. A good graphics card is worthwhile to have.
18 \item Multiple monitors really come in handy to increase productivity as you can see more information and
19 in bigger windows so you do not have to keep moving windows around.
21 Besides the above hardware recommendations, this section covers tips for performance improvements and tips on how to perform some specific tasks, often for older media.
23 \section{Hardware video acceleration}%
24 \label{sec:hardware_video_acceleration}
25 \index{hardware!acceleration}
27 With certain newer, more powerful graphics boards and newer device drivers, there is the potential for enhanced \textit{decode} and \textit{encode} performance. Decode refers to loading and playing video in \CGG{}. The GPU, Graphics Processing Unit, on the graphics board is accessed via one of the following libraries: vdpau or vaapi. The hardware acceleration done by the graphics card increases performance by activating certain functions in connection with a few of the FFmpeg decoders. This use makes it possible for the graphics card to decode video, thus offloading the CPU. Decode operations are described here next.
28 Encode refers to rendering video and is described at the end of this section
29 under \hyperref[sub:gpu_hardware_encoding]{GPU hardware encoding}.
31 VDPAU, Video Decode and Presentation API for Unix, is an open source library to offload portions of the video decoding process and video post-processing to the GPU of graphics boards, such as Nvidia. It may also apply to Nouveau and Amdgpu boards (by wrapper), but that has not been verified.
33 VA-API, Video Acceleration API, is an open source library which provides both hardware accelerated video encoding and decoding for use mostly with Intel (and AMD) graphics boards.
35 AppImage will probably not allow for either VDPAU or VA-API hardware acceleration because the
36 computer where AppImage is created will not have the same graphics card and/or vaapi/vdpau
38 It is recommended for best results that you build \CGG{} on your specific computer as described in \ref{sec:single-user-build}.
41 \item Hardware acceleration, vaapi or vdpau, works the same running either the created binary or the created AppImage from that binary on that same computer.
42 \item Hardware acceleration using an AppImage created on another computer running a different Operating System or even just a different level of Operating System, most likely will not work unless the user's computer just happens to have the same characteristics with respect to vaapi/vdpau as the computer where the AppImage was created.
45 Currently only the most common codecs, such as MPEG-1, MPEG-2, MPEG-4, H.264 /MPEG-4 and h265 (hevc), are accelerated/optimized by the graphics card to play these particular video formats efficiently. The other formats are not optimized so you will see no performance improvement since the CPU will handle them as before, just as if no hardware acceleration was activated. There are many different graphics cards and computer systems setup, so you will have to test which specific settings work best for you. So far this has been tested at least with Nvidia, Radeon, and Broadwell graphics boards on some AMD and Intel computers; depending on the graphics card, two to ten times higher processing speeds can be achieved. However, most graphic operations are single-threaded so that performing the operations in the hardware may actually be slower than in software making use of multiple CPUs, which frequently multi-thread many operations simultaneously.
47 \subsection{GPU hardware decoding}%
48 \label{sub:gpu_hardware_decoding}
49 \index{hardware!decoding}
52 \item Verify that you have installed \textit{libva-dev} or \textit{libva} on your operating system.
53 \item Verify that you have installed \textit{libva-intel-driver} on your operating system
54 \item Verify that you also have \textit{libvdpau-dev} or \textit{libvdpau} installed.
55 \item Verify \texttt{Settings $\rightarrow$ Preferences, Playback} tab, Video Driver is set to\textit{ X11} -- or \textit{X11-OpenGL} if that produces better results for your configuration.
56 \item Before starting \CGG{}, you can set an environment variable that can be easily reversed and
57 then, to run from the \CGG{} installed directory, key in:
60 \begin{lstlisting}[numbers=none]
61 CIN_HW_DEV=vdpau ./cin # for computers with Nvidia and some other graphics cards
62 CIN_HW_DEV=vaapi ./cin # mostly for computers with Intel or AMD specific graphics hardware
65 If you find that the environment variable setting is advantageous for your CinGG usage and you want to always use it, you can add it in your \texttt{\$HOME} directory \texttt{.profile} file which takes effect every time you log in. The line you would add would look something like this:
67 \begin{lstlisting}[numbers=none]
68 export CIN_HW_DEV=vdpau
70 export CIN_HW_DEV=vaapi
73 In addition for a computer with dual graphics cards, you can select your specific device with an environment variable
74 to decode and encode. For example:
75 \begin{lstlisting}[numbers=none]
76 CIN_DRM_DEC=/dev/dri/renderD129 CIN_DRM_ENC=/dev/dri/renderD128 ./cin
78 and substituting your specific location instead of /dev/dri/renderD129 or D128.
80 It might be more difficult to analyze problems as a result of using the GPU because of the wide variation in hardware. When you do not set the \texttt{CIN\_HW\_DEV} environment variable, the code will work exactly as before since this feature is self-contained.
82 There is also a \texttt{Settings $\rightarrow$ Preferences, Performance} tab, \textit{Use HW device} flag
83 with a pulldown to set up \textit{none, vdpau, vaapi,} or \textit{cuda}. To ensure it takes effect, it is best to set it the way you want, quit out of \CGG{} and then restart. Its current purpose is for flexibility, but there is a possibility that it might eventually take the place of \texttt{CIN\_HW\_DEV} -- both are not needed.
85 Precedence of the decode hardware acceleration settings are:
87 -- \texttt{yourfile.opts} is checked first so is of the highest priority; special .opts usage is described below
89 -- environment variable \texttt{CIN\_HW\_DEV} is checked next
91 -- preferences \textit{Use HW device} settings is of the lowest priority
93 \subsubsection*{Hardware decoding in \CGG{}}%
94 \label{ssub:hardware_decoding_cinelerra}
95 \index{appimage!vaapi/vdpau}
96 There are 4 phases during \CGG{}’s handling of hardware acceleration. These first 2 steps occur just \textit{before} the first read.
99 \item Check to see if Hardware Acceleration is enabled, usually indicated by \texttt{CIN\_HW\_ \_DEV} being set to
100 vaapi or vdpau. If enabled, try to activate the decode device, and if that fails revert to software.
101 \item The next step is to send some data to decode to see if that works. If this does not work, you will see
102 an error message of \textit{HW device init failed, using SW decode}.
105 \noindent These next 2 steps occur \textit{during} any read. Now there is no turning back to software so if the hardware gets an error, that video will not be handled correctly.
107 \begin{enumerate} [resume]
108 \item Read the media and send the raw stream data to the device for processing.
109 \item Read the device to receive the decoded data, and convert it to the session color model. If the GPU
110 can not convert the data, you will see the error message of \textit{Error retrieving data from GPU to CPU}.
113 AppImage will probably not allow for either VDPAU or VA-API hardware acceler-
114 ation because the computer where AppImage is created will not have the same
115 graphics card and/or vaapi/vdpau libraries as yours. In other words:
118 \item Hardware acceleration, vaapi or vdpau, works the same running either the created binary or the created AppImage from that binary on that same computer. See \nameref{sub:built_appimage_scratch}.
119 \item Hardware acceleration using an AppImage created on another computer running a different Operating System or even just a different level of Operating System, most likely will not work unless the user’s computer just happens to have the same characteristics with respect to vaapi/vdpau as the computer where the AppImage was created.
122 Due to these variations in user’s computer hardware configuration, it is often suggested that you refer to your startup window to check for error messages. Since your situation is unique, the error may not have been seen by anyone else and is probably unknown/undocumented.
123 \textcolor{red}{For debugging problems, modify in the \CGG{} ffmpeg subdirectory, the file: \texttt{decode.opts} by temporarily changing the line from \textit{loglevel =fatal} to \textit{loglevel =verbose} and restarting \CGG{}}.
125 \subsubsection*{Possible improvements or differences}%
126 \label{ssub:possible_improvements_differences}
129 \item The Frames Per Second (FPS) in playback might usually be at the maximum rate. You can check
130 this in \texttt{Settings $\rightarrow$ Preferences, Playback A}, looking at \textit{Framerate achieved}; the higher, the better.
131 \item Percent of the CPU used should be less, thus saving more CPU for other operations.
132 \item Some users get the the impression that playing seems smoother.
133 \item The CPU fan noise may go down as the CPU is used less.
134 \item The GPU graphics fan noise may go up as the GPU is used more.
137 Using the GPU is going to react differently depending on your hardware, software, and the number of files loaded. A good way to determine how well it is performing, is to watch the CPU load from another window running the command line routine \texttt{top}. Consider the following possibilities:
140 \item If you only use smaller videos occasionally and then use other codecs than the ones mentioned
141 previously as being the optimized set, it is usually a better idea to stay with the default settings
142 without all the hardware tests.
143 \item If you have 4 cores or less, but a really good \textit{gaming card}, using vaapi/vdpau could be a big help.
144 \item If you load only a couple of files, the GPU vaapi/vdpau should be faster depending on your graphics
145 board and how capable it is.
146 \item If you load 10 camera Mixers of H.264 format, it always seems to work a lot better playing them.
147 \item If you have an epyc AMD chip with 128 CPUs and load 50 files, without vaapi/vdpau may be better.
150 \subsubsection*{Special .opts file}%
151 \label{ssub:special_opts_file}
152 \index{hardware!special opts files}
154 The situation may arise where you have enabled hardware acceleration and after loading several files for a project, you find that a file had some kind of error resulting in a black video instead of an image or you see an error message pop up which states something like \textit{Error retrieving data from GPU to CPU} or \textit{err: Unknown error occurred}. Because the \texttt{CIN\_HW\_DEV} environment variable is either all or none, ordinarily in order to correct the non-working video you would have to turn off hardware acceleration for the entire project/session. However, there is a way to continue working on your project without having to reload all of your files. You still use the environment variable and it will be effective for all of the formats it is able to handle, but you make an exception for any of the files that erred out. To do this you simply create a file in the same directory with the same name as the erring file with the different extension of .opts. The contents of this .opts file would just be the one line of:
156 \begin{lstlisting}[numbers=none]
159 Conversely, if you have a bunch of files in your project, like dnxhd format, that are not hardware accelerated, but you have an accompanying large file of type .mp4 for which you would like the hardware acceleration, you can leave the \texttt{CIN\_HW\_DEV} variable unset (that is, do not use it) and just create an .opts file containing the line:
162 \begin{lstlisting}[numbers=none]
166 For example your file, \texttt{test.mp4}, would have a side-kick called \texttt{test.opts} that will use the GPU for decoding/playing and the other files will just use the software. This is of some advantage because the ones that can not use the GPU if the environment variable is enabled, will not have to even check which saves a minuscule bit of time.
168 It is important to note that if using the .opts file to override the default \texttt{ffmpeg / decode.opts} file, you will most likely see more warnings (not really errors) in the \CGG{} startup window because the standard \texttt{decode.opts} file has \textit{loglevel = fatal} whereas the default is \textit{loglevel = error}. To avoid seeing all of the extra warnings, you can simply add the line \textit{loglevel=fatal} to your .opts file.
170 \subsubsection*{To verify hardware acceleration}%
171 \label{ssub:verify_hardware_acceleration}
173 Probably the easiest way to tell if hardware acceleration is working, is just to look at the messages in the window from where you started Cin (not available if start using the application icon). For example load a png, dnxhd, or some other non-supported format file and you will see messages similar to those below. The line \textit{HW device init failed, using SW decode} indicates that the vdpau/vaapi HW (hardware) decode is not available so will use SW (software) decode instead.
175 \begin{lstlisting}[numbers=none]
176 Failed to get HW surface format.
177 HW device init failed, using SW decode.
178 file:/tmp/media/aer_zypr.mp4
183 Decoder dnxhd does not support device type vdpau.
184 HW device init failed, using SW decode.
185 File:/tmp/media/test.gxf
190 HEVC with NVIDIA, VDPAU driver is buggy, skipping
193 AppImage does not provide this capability. If you would like to see more information on what is occurring, you can modify in the \CGG{} ffmpeg subdirectory, the file: \texttt{decode.opts} by temporarily changing the line from \textit{loglevel =fatal} to \textit{loglevel =verbose} and restarting \CGG{}. Then you will see messages in the startup window like:
195 \begin{lstlisting}[numbers=none]
196 [AVHWDeviceContext @ 0x7fc9540be940] Successfully created a VDPAU device
197 (NVIDIA VDPAU Driver Shared Library 390.116 Sun Jan 27 06:28:58 PST 2019) on X11 display :0
198 [h264 @ 0x7fc950159380] Reinit context to 1920x1088, pix_fmt: vdpau
199 [h264 @ 0x7fc92c3dd4c0] Reinit context to 1920x1088, pix_fmt: yuv420p
202 Again, to measure the performance run \texttt{top} from another window to check the CPU usage which will go lower as more work is offloaded to the GPU graphics card instead (it may go down by 2 to 10 times) or check the \textit{Framerate achieved} while playing.
204 Some mixed preliminary results that have been reported are provided below.
206 \subsubsection*{Case 1:}%
209 \noindent X11 Video Driver set in \texttt{Settings $\rightarrow$ Preferences, Playback A} tab:
213 CIN\_HW\_DEV=off ./cin & &CPU $58\%$ \\
214 CIN\_HW\_DEV=vdpau ./cin & &CPU $32\%$ \\
215 CIN\_HW\_DEV=vaapi ./cin & &CPU $82\%$ \\
219 \noindent X11-OpenGL Video Driver set in \texttt{Settings $\rightarrow$ Preferences, Playback A} tab:
223 CIN\_HW\_DEV=off ./cin & &CPU $48\%$ \\
224 CIN\_HW\_DEV=vdpau ./cin & &CPU $12\%$ \\
225 CIN\_HW\_DEV=vaapi ./cin & &CPU $80\%$ \\
229 \noindent Best is the least amount of CPU usage. Note that in this Case 1, using the X11-OpenGL video driver is
230 better than using the X11 video driver.
232 \subsubsection*{Case 2:}%
235 \noindent X11 Video Driver set in \texttt{Settings $\rightarrow$ Preferences, Playback A} tab:
239 CIN\_HW\_DEV=off ./cin & &CPU $60\%$ \\
240 CIN\_HW\_DEV=vdpau ./cin & &CPU $11\%$ \\
241 CIN\_HW\_DEV=vaapi ./cin & &CPU $60\%$ \\
245 \noindent X11-OpenGL Video Driver set in \texttt{Settings $\rightarrow$ Preferences, Playback A} tab:
249 CIN\_HW\_DEV=off ./cin & &CPU $67\%$ \\
250 CIN\_HW\_DEV=vdpau ./cin & &CPU $60\%$ \\
251 CIN\_HW\_DEV=vaapi ./cin & &CPU $67\%$ \\
255 \noindent Best is the least amount of CPU usage. Note that in this Case 2, using the X11 video driver is better
256 than using the X11-OpenGL video driver.
258 Older graphics cards or non-performing graphics cards will probably bring only a small amount of improvement or no speed advantage at all. You can check to see if vdpau is implemented for your specific Nvidia board at:
260 {\small \url{https://download.nvidia.com/XFree86/Linux-x86_64/304.137/README/supportedchips.html}}
262 And, you can see what your specific hardware and software might support by running either \texttt{vainfo}
263 (available in the libva-utils package) or \texttt{vdpauinfo} (available in the vdpauinfo package) from the
264 command line. For example which "Profiles" your hardware can handle is listed in the output such as Main
265 and High. Partial examples of each are shown below.
267 \begin{lstlisting}[numbers=none]
269 vainfo: VA-API version: 1.4 (libva 2.4.0)
270 vainfo: Driver version: Intel i965 driver for Intel(R) Broadwell - 2.4.0.pre1 (2.3.0-11-g881e67a)
271 vainfo: Supported profile and entrypoints
277 VAProfileH264MultiviewHigh
278 VAProfileH264StereoHigh
282 VAProfileVP8Version0_3
285 \begin{lstlisting}[numbers=none]
287 display: :0 screen: 0
289 Information string: G3DVL VDPAU Driver Shared Library version 1.0
291 Decoder capabilities:
293 name level macbs width height
294 ----------------------------------------------------
295 MPEG1 --- not supported ---
296 MPEG2_SIMPLE 3 65536 4096 4096
297 MPEG2_MAIN 3 65536 4096 4096
298 H264_BASELINE 52 65536 4096 4096
299 H264_MAIN 52 65536 4096 4096
300 H264_HIGH 52 65536 4096 4096
301 VC1_SIMPLE 1 65536 4096 4096
302 VC1_MAIN 2 65536 4096 4096
303 VC1_ADVANCED 4 65536 4096 4096
306 One last item of note, \textit{nvdec} is also enabled in the ffmpeg build, but at this time it is not known how this decode option on Nvidia graphics boards works or does not work in conjunction with vdpau.
308 \subsection{GPU hardware encoding}%
309 \label{sub:gpu_hardware_encoding}
310 \index{hardware!encoding}
312 Encoding using hardware acceleration of your graphics board GPU is
313 included in \CGG{} but it is of limited availability and works only
314 with a specific set of hardware graphics boards, a certain level of
315 graphics driver versions and only with certain ffmpeg formats. The
316 encoding is done via vaapi (libva installed), which is known to work
317 with Intel HD graphics boards and some others or via nvenc as
318 developed by Nvidia for Nvidia graphics boards.
320 \subsubsection*{Broadcom, Intel HD, AMD}%
321 \label{ssub:broadcom_intel_amd}
322 \index{hardware!vaapi}
324 To use hardware acceleration for rendering (that is, encoding) you do not have to set a preference or an environment variable, as was required for decoding. To use this feature you use an ffmpeg render options file which specifies a vaapi codec, such as h264\_vaapi. You must include this line in that options file to trigger the hardware probe: \qquad \texttt{CIN\_HW\_DEV=vaapi}
326 There are currently 4 options files available in the Render menu already set up for you that you see when you select the Video wrench and use the down arrow on the first line in the menu. These are:
329 \item[h264\_vaapi.mp4] known to work on Intel computer with Intel Broadwell graphics driver
330 \item[mpeg2\_vaapi.mp4] known to work on Intel computer with Intel Broadwell graphics driver
331 \item[mjpeg\_vaapi.mp4] error message of \textit{open failed with mjpeg\_vaapi$\dots$} on above computer
332 \item[hevc\_vaapi.mp4] error message of \textit{open failed with hevc\_vaapi$\dots$} on above computer
334 Other option files can be added as needed for your specific hardware if it is known to work for you, such as VP8 and VP9. An example of the included \CGG{}’s \texttt{ffmpeg/video/h264\_vaapi.mp4} file (figure~\ref{fig:render-vaapi}):
338 \includegraphics[width=1.0\linewidth]{render-vaapi.png}
339 \caption{Render menu setup to encode using GPU with vaapi}
340 \label{fig:render-vaapi}
343 \begin{lstlisting}[numbers=none]
349 According to an online wiki, hardware encoders usually create output of lower quality than some software encoders like x264, but are much faster and use less CPU. Keep this in mind as you might want to set a higher bitrate to get output of similar visual quality.
351 Results of a particular test case performed on a Intel, 4-core computer, with Broadwell Graphics using an mp4 input video/audio file with dimensions of
352 $1440x1080 / 29.97fps$ is shown next (note, filename is \texttt{tutorial.mp4}). This may very well be a \textit{best case} scenario! But clearly, at least on this computer with only 4 cores, the hardware acceleration seems to be quite advantageous. A comparison of the 2 output files
353 using \texttt{ydiff} as described in the Appendix (\ref{sec:ydiff_check_results}) shows no obvious defects.
356 \begin{tabular}{l|cccc}
357 &CPU usage & Render Time & File Size & File \\
359 none & 388\% &100 secs &36,862,542 & h264.mp4 \\
360 vaapi & 150\% & 19 secs & 74,522,736 & h264\_vaapi.mp4 \\
364 \subsubsection*{Nvidia graphics boards}%
365 \label{ssub:nvidia_graphics_card}
366 \index{hardware!nvenc}
368 To use hardware acceleration for rendering (that is, encoding) you do not have to set a preference or an environment variable, as was required for decoding. To use this feature you use an ffmpeg render options file which specifies the nvenc codec, either \texttt{h264\_nvenc.mp4} or \texttt{nvenc.mp4}. There are several requirements in order for this to work on your computer as listed here:
371 \item Nvidia graphics board at or above a certain hardware level. For h265, newer boards are required.
372 \item Software drivers for your graphics board must be installed on your computer.
373 \item The driver must support at least API version 9.0 -- minimum required Nvidia driver for nvenc is
374 390.25 or newer. You will see error messages on the startup window if you are on lower versions.
377 If you try to render using\texttt{ h264/h265\_nvenc.mp4} formats and do not have an Nvidia graphics card or this feature was not built in, you will see in the window from where you started \CGG{}, the error message: \qquad \textit{Cannot load libcuda.so.1}
379 A small test using 2 minutes from the 4k version of Big Buck Bunny shows using nvenc can be about 4 times faster. The test was done on a 4 core Intel laptop with an Nvidia 950M graphics board.
382 \begin{tabular}{l|cccl}
383 &CPU usage & Render Time & File Size & File \\
385 none & 388\% &20 mins 18 secs & 156,517,069 & h264.mp4 \\
386 nvenc & 252\% & 5 mins 44 secs & 42,052,920 & h264\_nvenc.mp4 \\
390 Of note in this test, 388\% CPU usage with only 4 cores shows that there is probably slow down because there is no more CPU power available. Therefore, using the GPU hardware acceleration with nvenc provides a significant speed-up. Also, note the larger file size without making use of the GPU – this probably indicates that there is a big difference in bitrate or quality parameter settings used in the options file and this should be taken into consideration.
392 \subsubsection*{Important Tip}%
393 \label{ssub:important_tip}
395 There is one last potentially significant graphics speedup when using the X11-OpenGL driver for users with Nvidia graphics boards who are seeing frames/sec achieved lower than what the video format is set to. You may want to disable \textit{sync to vblank} (an option for OpenGL) in NVIDIA X Server Settings for the proprietary drivers. This could increase your frames per second on playback.
397 \subsection{Effects (OpenCL, Cuda)}%
398 \label{sub:effects_opencl_cuda}
402 CUDA® is a parallel computing platform / programming model developed by Nvidia that provides big increases in computing performance through use of the GPU. It was first introduced in about 2006 for applications in computationally intense fields such as astronomy, biology, chemistry, and physics.
404 At the time this was written, the use of Cuda is not going to improve the playing and rendering of video in \CGG{} except in the case where you use a specific Cuda-enabled plugin that is computationally intense -- sadly, most of what Cin does, Cuda will not help. Cuda is mostly a \textit{block oriented algorithm} which works well for such things as \textit{a flock of birds all flying next to each other}.
406 The same as for vaapi and vdpau, you can enable Cuda in the:\\
407 \texttt{ Settings $\rightarrow$ Preferences, Performance} tab, \textit{Use HW Device}\\
408 but it will not affect anything unless you have Cuda installed on your system and have built \CGG{} yourself with Cuda build enabled. To install it on your computer, you will need to do the following:
411 \item Make sure you have the Nvidia proprietary library drivers for your graphics board already installed and up to date.
412 \item Go to the Nvidia Cuda development website and choose one of the available operating system’s
413 such as Fedora, OpenSuse, CentOS, Ubuntu, $\dots$ at {\small \url{https://developer.nvidia.com/}}
414 \item You will be installing repositories by package -- this will be around 3 GB.
415 \item Also, install the Fusion repo, although it is unknown if necessary or not.
418 There is a very good set of directions on the website to just follow. Once you have installed the Cuda software on your computer, you must build \CGG{} yourself -- the default flag for a build in the configure script for cuda is \textit{auto}. For Arch, and possibly other distros, you may have to supply the CUDA\_PATH if it is not in the standard place. This would be
422 \begin{lstlisting}[numbers=none]
423 env CUDA_PATH=/opt/cuda
424 OR if using the bash shell:
425 export CUDA_PATH=/opt/cuda
427 For all distros you will need to add to the ./configure line in your build script, the following:
430 \begin{lstlisting}[numbers=none]
432 so your configure line will look something like this:
433 ./configure --with-single-user --with-booby --with-cuda
436 There are currently 2 available plugins for \textit{show and tell} that take advantage of the hardware acceleration of Cuda -- \textit{N\_Body} and \textit{Mandelbrot} (see \nameref{sec:cuda_plugins}).
438 An error you may see on your \CGG{} startup window when you have Cuda installed and try to run one of the 2 plugins is \textit{cudaErrorInsufficientDriver}. This indicates CUDA 10 (the current version at the time of this writing) is not compatible with the driver version on your computer. You can either:
441 \item Upgrade the driver if your board supports newer nvidia builds.
442 \item downgrade the cuda development package to a version that works for your board.
445 \subsection{Final Note}%
446 \label{sub:final_note_on_acceleration}
448 In wrapping up this Hardware Acceleration section, you may want to refer to the following to determine the current supported formats:
450 {\small \url{https://wiki.archlinux.org/index.php/Hardware_video_acceleration}}
452 \section{Optimized Playback -- X11 Direct}%
453 \label{sec:optimized_playback}
454 \index{playback -X11 direct}
456 Normally, when \CGG{} reads a video frame, it is copied into a \textit{Vframe}. This frame may also need other actions performed on it, such as a color model change. In addition, ffmpeg and libzmpeg \textit{can\_scale\_input}. So the read can be color transformed and scaled just by asking the library to do that. This means that if the compositor is in a \textit{good} state with no zoom, the VFrame read can be done in the fastest render color model, and already scaled to the correct size for the compositor. In reality, this is not what you need for editing, but quite often the \textit{virtualconsole} is not used because the render is media only -- that is \textit{just data}. If no data transforms are needed and the input scaling can be done, the vrender program detects this, and tells the codec to transmit the data in a compatible way to the compositor canvas. This is the \textit{X11 direct} data path.
458 With the X11 video driver choice, large format files such as 4K, will playback faster than either X11-XV or X11-OpenGL. However, you still have the option to turn off the X11 direct data path if you use
459 \texttt{Settings $\rightarrow$ Preferences, Playback A} tab set video driver to X11 and uncheck \textit{use direct X11 render if possible}.
462 \section{Proxy Settings and Transcode}%
463 \label{sec:proxy_settings}
465 Information on proxies to reduce required CPU for less powerful computers is in the section \nameref{sec:proxy}.
467 Information on transcode which is used to provide keyframes to make the video seekable is in the section \nameref{sec:transcode}.
471 \section{Cache size and Seconds to preroll render}%
472 \label{sec:cache_and_preroll}
475 \textit{Cache size} in \texttt{Settings $\rightarrow$ Preferences, Performance} tab is used to store images on the timeline. One 1080p frame uses about 10 MB. The default setting is 256 and this is enough for testing and running. However, why not use more memory if it is available. To experiment for testing a good number tuned to the way you use your computer, set the cache to 0, start up \CGG{}, load a typical media file, play it and run \textit{top} on the command line in another window to see how much memory is being used. In the \textit{top} display, look at \textit{free} memory. Whatever your computer is not using, is a good number to use for cache. If you start other programs, or change the design of the session so that it uses a lot of frame storage, you may need to experiment again later and resize accordingly. The system keeps all requested data cached until it is replaced by other data or you reboot your PC. Reboot if you want to clear the cache.
477 \textit{Seconds to preroll render} in \texttt{Settings $\rightarrow$ Preferences, Performance} tab are used to increase the amount of frames (which are calculated in advance) at the time of their use. The default setting is 0.5 sec but you can increase it to 1.0 sec to improve smoothness in the timeline. Higher values are always less effective.
479 \section{Tips for Improving Smaller Computers Use}%
480 \label{sec:tips_improving_smaller_computers}
482 A list of items to check for smaller computers that will help to use less cpu/memory/resources follows:
485 \item For large media files, use proxy to do your main editing.
486 \item In \texttt{Settings $\rightarrow$ Preferences, Appearance} tab, uncheck \textit{Use thumbnails in resource window}.
487 \item In \texttt{Settings $\rightarrow$ Preferences, Appearance} tab, uncheck \textit{Autocolor assets}.
488 \item Speed-up certain time-consuming FFmpeg plugins through use of a carefully selected \texttt{.opts} file. See \nameref{sub:speedup_ffmpeg_plugin_opts}
489 \item For large media files, in \texttt{Settings $\rightarrow$ Preferences, Playback A}, Video Driver set \textit{use direct X11 render if possible}.
490 \item For the Video Driver in \texttt{Settings $\rightarrow$ Preferences, Playback A}, if using a good graphics card, choose \textit{X11-OpenGL}.
491 \item Set \textit{CIN\_HW\_DEV=vdpau} or \textit{vaapi} to use the graphics GPU for certain ffmpeg media decoding.
492 \item If you have multiple cpus or multiple computers, even if slow, take advantage of using \textit{Render Farm}.
493 \item When editing, \textit{background rendering} causes temporary output to be rendered constantly while the
494 timeline is being modified. The temporary output is displayed during playback whenever possible so
495 it does not have to be recalculated -- very useful for transitions and previewing effects that are slow.
496 \item In \texttt{Settings $\rightarrow$ Preferences, Playback A}, uncheck \textit{Play every frame} which means frames will be skipped as playback of the video falls behind.
497 \item Adjust \textit{Cache size} in \texttt{Settings $\rightarrow$ Preferences, Performance} tab, to not exhaust the memory and yet still provide decent playback.
500 \section{General Crash Handling Tips}%
501 \label{sec:general_crash_tips}
502 \index{crash handling tips}
504 This section is a handy guide for describing various kinds of software computer system failures. Only some of these various lockups or crashes can be dealt with. Hopefully, it will help to have some hints to know what kind of failure it is, or to save your work or to avoid future problems. For most of this, your user name must be root, although you can certainly try to see if it works for you when not root.
507 \item[System lockups:] When the system locks up, it is usually a system problem. Normally an application program cannot lock up the system. It is a major goal of system design to prevent an application (app) from failing a system interface. This does not mean an app can not cause a system lockup, but it is unusual.
508 \item[\CGG{} crash:] This is covered in \hyperref[cha:crash_dumps_analysis]{Crash Dumps for Analysis}. Just a reminder that for best results you should be root and by providing a crash dump and as much other information as possible, you will be helping the developer to analyze the problem and fix it so that it can be avoided in the future.
509 \item[X Server crash:] Keyboard does not respond, screen is frozen, caps lock may operate LED light. Sometimes using Ctrl-Alt-F1 $\dots$ Ctrl-Alt-F7 (etc.) will allow you to regain control of a VT console. You can use this to login and check logs: eg. \texttt{/var/log/Xorg.0.log}, \textit{dmesg}, \textit{journalctl} $\dots$ etc. If you have another computer, make sure a terminal server is configured (for example: rsh, ssh, or telnet), then remote login via this other computer and check the logs. Most important is to immediately note the current software state, and the very last thing that preceded the crash, i.e. last button click, last keystroke, $\dots$ or whatever.
510 \item[Kernel crash:] The machine goes completely dead. The keyboard caps lock LED will probably be flashing. Most likely the only way to see anything after the kernel crashes is to use a serial port console log and usually kdb, the kernel debugger, and special cabling. This requires a lot of setup, and is normally reserved for experts. Login from another computer will not work. Pinging the ip address will not respond since the network stack is part of the kernel. There are some virtual machine setups that will let you debug a \textit{guest} kernel, but this also requires a lot of setup, and affects which kernel is currently under test. The kdb route is preferable.
511 \item[Keyboard grabs, Server grabs, and Deadlocks:] A grab is an X-server state where all events are forced to just one window event stream. This forces the user to respond to the dialog. Things seems to be working, but no keypresses do anything useful. The system clock and other programs will still be working. The network will work for remote logins. Grabs can be canceled if the \texttt{/etc/X11/xorg.conf} X config contains special setup as shown below:
514 \begin{lstlisting}[numbers=none]
515 Section "ServerFlags"
516 Option "HandleSpecialKeys" "Always"
517 Option "AllowDeactivateGrabs" "True"
518 Option "AllowClosedownGrabs" "True"
521 Section "InputDevice"
522 Identifier "Keyboard"
525 Option "XkbOptions" "terminate:ctrl_alt_bksp"
526 Option "XkbOptions" "grab:break_actions"
530 or to \texttt{\$HOME/.xinitrc}, add:
532 \begin{lstlisting}[numbers=none]
533 # xkb terminate/grab actions disabled in xorg.conf, use:
534 setxkbmap -option "grab:break_actions"
535 setxkbmap -option "terminate:ctrl_alt_bksp"
536 ctrl-alt-bksp = terminate the X-server, may restart automatically
540 Modal forms (always on top, and usually ptr/kbd grab) dialog boxes can lock a system by putting a form over another form holding a grab. This means the form that needs input may never get any because you can not get to it, and the result is a deadlock. Usually you will have to restart X (Ctrl-Alt-Bksp).
543 \item[Window Manager issues:] The \textit{desktop} window manager can intercept and modify all kinds of user input. Mostly, this is a good thing, but can be a nuisance. If user keypresses can be programmed to trigger actions, then they may be useful to send \texttt{KILL} or \texttt{INTR} to an app that seems to be holding X's attention. For example:
544 \begin{lstlisting}[numbers=none]
545 killall -INTR cinelerra,
546 killall -9 cinelerra,
548 # but you must run as root to be able do this
550 The ALT and META keys may be intercepted by the window manager, and this can cause unexpected interface operations.
553 \section{Tips for Specific Operations}%
554 \label{sec:tips_specific_operations}
556 \subsection{Generating a 440 Hz tone}%
557 \label{sub:generating_440_tone}
559 To create a specific 440 Hz tone, follow these steps. You can vary the length, use more channels, or change the frequency to a different desired value (figure~\ref{fig:aeval}).
563 \includegraphics[width=1.0\linewidth]{aeval.png}
564 \caption{Use Audio$\rightarrow$Render effect to set render parameter values and then that effect can be varied.}
569 \item Make sure there is an armed audio track on the timeline, get into Cut and Paste mode, and highlight
570 a selection or define In/Out points where you want to insert the audio tone.
571 \item Go to \texttt{Audio $\rightarrow$ Render effect}. Rendered effect usage is described in \hyperref[sec:rendered_effects]{Effect Plugins}. This brings up a menu where you will select the desired effect which in this case is \textit{F\_aeval}. Also Select a file to render to, a File Format, and Insertion strategy of Paste at insertion point.
572 \item Click on the green OK checkmark which will popup the F\_aeval effect so that you can set the
574 \item Highlight the \textit{exprs} option and key in a specific audio filter expression which for 440 Hz would be:
575 $\sin(2\pi t\times440)$. Then hit the Apply button.
576 \item Next when you hit the green OK checkmark on the \CGG{}: Effect Prompt popup, you will have
577 your 440 Hz tone on the timeline plus in the select file that you chose to render it to.
578 \item To use 2 channels instead of 1, in the F\_aeval menu highlight the \textit{channel\_layout} option and change
579 that to 1C|2C instead of the usual default of 1C.
582 \subsection{Camera supplied LUTs}%
583 \label{sub:camera_supplied_luts}
586 A LUT, acronym for Look-Up Table, is a mathematically precise way of taking specific RGB image values from a source image and modifying them to new RGB values by changing the hue, saturation and brightness values of that source image. In other words, LUTs are used to map one color space to another. Some high-end cameras supply a .cube file to use as input. There are several different ffmpeg plugins included with CinGG for using Lut's. These are:
589 \item[F\_lut:] Compute and apply a lookup table to the RGB/YUV input video.
590 \item[F\_lut1d:] Adjust colors using a 1D LUT "file" input.
591 \item[F\_lut3d:] Apply a 3D LUT "file" to an input video.
592 \item[F\_lutrgb:] Compute and apply a lookup table to the RGB input video.
593 \item[F\_lutyuv:] Compute and apply a lookup table to the YUV input video.
596 For example, to use a 3dlut simply load your video, drop the F\_lut3d plugin on that track, and bring up the lut3d controls window, highlight the \textit{file} option, key in your file name (with path), and hit Apply to have the lut take effect. To easily adjust, move the \textit{fader} slider in the patchbay for that video track. Only F\_lut1d and F\_lut3d allow for a file input and these usually are files with the .cube extension.
598 \subsection{Encoding into Dolby Pro Logic}%
599 \label{sub:encoding_dolby_pro_logic}
600 \index{dolby pro logic}
602 Dolby pro logic is an easy way to output 6 channel audio from a 2-channel soundcard with degraded but useful results. Rudimentary Dolby pro logic encoding can be achieved with usage of some effects.
605 \item First, create the front left and right channels. Create 2 audio tracks, each carrying either the left or
606 right channel. Pan the left channel to the left and the right channel to the right with the pan control.
607 \item Now create the rear left and right channels. Create another 2 audio tracks as above -- the left channel
608 panned left and the right channel panned right. Then apply invert audio to both new channels and
609 the signals will come out of the rear speakers.
610 \item Next, create the center channel by creating a single audio track with monaural audio from a
611 different source. Center it with the pan control and the signal will come out of the center speaker.
615 \item If a copy of the signal in the back speakers is desired in any single front speaker, the signal in the back
616 speakers must be delayed by at least $0.05$ seconds and a single new track should be created. Pan the
617 new track to orient the signal in the front speakers.
618 \item If the same signal is desired in all the speakers except the center speaker, delay the back speakers by
619 $0.5$ seconds and delay either the front left or front right by $0.2$ seconds.
620 \item If you want to hear something from the subwoofer, create a new track, select a range, drop a
621 synthesizer effect, and set the frequency below $60 Hz$. The subwoofer merely plays anything below
625 Other tricks you can perform to separate the speakers are parametric equalization to play only selected ranges of frequencies through different speakers and lowpass filtering to play signals through the subwoofer.
627 \subsection{Improving Analog TV Quality}%
628 \label{sub:improving_tv_quality}
630 The picture quality on analog TV is not always good but you can modify parameters in \CGG{} to make it look more like it did in the studio.
632 First, when capturing the video, capture it in the highest resolution possible. For Europeans this would be $720\times576$ and for North Americans, $720\times480$. Do not adjust the brightness or contrast in the recording monitor, but you might want to max out the color. Capture the video using MJPEG or uncompressed Component Video if possible; if not possible, then capture it using JPEG preferably or RGB if that is all that will work. Now on the timeline use \texttt{Settings $\rightarrow$ Format} to set a YUV colorspace, drop a \textit{Downsample} effect on the footage and set it as follows:
634 \begin{lstlisting}[numbers=none]
645 Use the Camera in the compositor to shift the picture up or down a line to remove the most color interference from the image. If you have vertical blanking information or crawls which constantly change in each frame, block them out by using a Mask. This improves compression ratios. More invasive quality improvement techniques involve removing the interlace via deinterlacing.
647 \subsection{Remove Interlacing}%
648 \label{sub:remove_interlacing}
651 Interlacing often exists on older video sources, such as camcorders, and was previously used in broadcast television. Playing this video results in jagged images on a computer monitor, but with \CGG{} you can use deinterlacing effects to solve this. After some experimentation, it has been determined that the FFmpeg \textit{F\_kerndeint} plugin seems to produce the best results with the least amount of fiddling. But some of the parameters described next are pertinent to other potential plugin usage.
654 \item[Line Doubling:] done by the \textit{Deinterlace} effect when set to \textit{Odd} lines or \textit{Even} lines. When applied to a track it reduces the vertical resolution by $\frac{1}{2}$ and gives you progressive frames with stairstepping. This is only useful when followed by a scale effect which reduces the image to half its size.
655 \item[Line averaging:] the \textit{Deinterlace} effect when set to \textit{Average even} lines or \textit{Average odd} lines does exactly what line doubling does except instead of making straight copies of the lines, it makes averages of the lines. This is actually useful for all scaling.
656 \item[Inverse Telecine:] this is the most effective deinterlacing tool when the footage is an NTSC TV broadcast of a film. It is described in Effect Plugins (\ref{sub:inverse_telecine}).
657 \item[Time base correction:] the previously discussed three tools either destroy footage irreversibly or do not work at times. Time base correction may be a better tool to use because it leaves the footage intact. It does not reduce resolution, perceptually at least, and does not cause jittery timing.
658 \item[Frames to Fields effect:] converts each frame to two frames, so it must be used on a timeline whose project frame rate is twice the footage's frame rate. In the first frame it puts a line-averaged copy of the even lines. In the second frame it puts a line-averaged copy of the odd lines. When played back at full framerate it gives the illusion of progressive video with no loss of detail. This effect can be reversed with the \textit{Fields to Frames} effect, which combines two frames of footage back into the one original interlaced frame at half the framerate. However, keep in mind that Frames to Fields inputs frames at half the framerate as the project. Effects before Frames to Fields process at the reduced framerate. The output of Frames to Fields can not be compressed as efficiently as the original because it introduces vertical twitter and a super high framerate. Interlaced $29.97$ fps footage can be made to look like film by applying Frames to Fields and then reducing the project frame rate of the resulting $59.94$ fps footage to $23.97$ fps. This produces no timing jitter and the occasional odd field gives the illusion of more detail than there would be if you just line averaged the original. It is described in Effect Plugins (\ref{sub:frames_to_fields}).
659 \item[HDTV exceptions:] $1920\times1080$ HDTV is encoded in a special way. If it is a broadcast of original HDTV film, an inverse telecine works. But if it is a rebroadcast of a $720\times480$ source, you need to use a time base and line doubling algorithm to deinterlace it.
662 \subsection{Making video look like film}%
663 \label{sub:making_video_look_film}
665 With an older camcorder video which has low quality video, you can improve the results by turning it into progressive 24 fps output as close as possible. Only do this for low quality video.
668 \item Set project framerate to twice the video framerate.
669 \item Apply a \textit{Sharpen} effect. Set it to sharpness: 25, no interlacing, and horizontal only.
670 \item Drop a \textit{Frame to Fields} effect on the same track. Set Average Empty Rows to on and play through
671 the video a few times to figure out which field is first. If the wrong field is first, the motion is shaky.
672 Any editing in the doubled frame rate may now damage the field order. It is not clear which is the
673 easiest way to support warnings for field glitches but you should go back to the normal framerate to
674 do editing or play test to make sure the fields are right.
675 \item Render just the video to the highest quality file possible.
676 \item Import the video back to a new track. Set the project framerate to 24. The new track should now
677 look more like a file with sharper images than the original footage.
680 This entire procedure could be implemented in one non-realtime effect, but the problem with that is you will most often want to keep the field based output and the 24 fps output for historical purposes. A non-realtime effect would require all that processing just for the 24 fps copy.
682 \subsection{Clearing out haze}%
683 \label{sub:clearing_out_haze}
685 If you photograph a lot of haze instead of blue sky, these horizon shots will usually need more depth. You can use the \texttt{gradient} effect to improve your video. Drop the gradient effect on hazy tracks and set the following parameters:
687 \begin{lstlisting}[numbers=none]
691 Inner color: blue 100% alpha
692 Outer color: blue 0% alpha
695 It is important to set the $0\%$ alpha color to blue even though it is $0\%$ alpha. The color of the outer alpha is still interpolated with the inner color. This is a generally applicable setting for gradient. Some scenes may work better with orange or brown for an evening feel.
697 \subsection{Making a ringtone for a cell phone}%
698 \label{sub:make_ringtone_phone}
701 \item Go to \texttt{File $\rightarrow$ Load files$\dots$} and load a sound file with Insertion strategy: \textit{Replace current project}.
702 \item Go to \texttt{Settings $\rightarrow$ Format }change \textit{Channels} to 1 and \textit{Samplerate} to 16000 or 22050.
703 \item Highlight a region of the timeline to use for the ringtone. To improve sound quality on the cell
704 phone, you need the maximum amplitude in as many parts of the sound as possible.
705 \item Right click on track Audio 1 and select \textit{Attach effect}. Highlight the \textit{Compressor} effect and hit
706 Attach in the attachment popup.
707 \item Make sure the insertion point or highlighted area is in the region with the Compressor effect.
708 \item Right click on track Audio 2 and select \textit{Attach effect}.
709 \item Highlight Audio 1 Compressor and hit Attach.
710 \item Click the Audio 1 Compressor's magnifying glass to bring up the compressor GUI.
711 \item Set the following parameters:
712 \begin{lstlisting}[numbers=none]
719 \item Click Clear to clear the graph. Click anywhere in the grid area and drag a new point to 0 Output
720 and -50 Input. The graph should look similar to the figure~\ref{fig:ringtone}.
721 \item Go to \texttt{File $\rightarrow$ Render}. Specify the name of an mp3 file to output to. Set the file format to MPEG
722 Audio. Click the wrench for Audio and set Layer to III and Kbits per second to 24 or 32. Check
723 Render audio tracks and uncheck Render video tracks. Hit OK to render the file.
728 \includegraphics[width=0.8\linewidth]{ringtone.png}
729 \caption{Using the Compressor plugin graph to create a ringtone}
733 The resulting .mp3 file can be uploaded to a web server and then the phone's web browser can download the .mp3 file directly from the URL. There may be a size limit on the file.
735 \subsection{Time stretching audio}%
736 \label{sub:time_stretching_audio}
737 \index{audio!time stretching}
739 It may appear that time stretching audio is a matter of selecting a region of the audio tracks, enabling recording for the desired tracks, going to\texttt{ Audio $\rightarrow$ Render Effect}, and applying \textit{TimeStretch}. In actuality there are 3 audio effects for time stretching: Time Stretch, Resample, and Asset info dialog.
741 \textit{Time Stretch} applies a fast Fourier transform to try to change the duration without changing the pitch, but this introduces windowing artifacts to the audio. It is only useful for large changes in time because obvious changes in duration make windowing artifacts less obtrusive.
743 For smaller changes in duration, in the range of $5\%$, \textit{Resample} should be used. This changes the pitch of the audio but small enough changes are not noticeable. Resample does not introduce any windowing artifacts, so this is most useful for slight duration changes where the listener is not supposed to know what is going on.
745 Another way to change duration slightly is to go to the Resources window, highlight the media folder, right click on an audio file, click on \textit{Info}. Adjust the sample rate in the Info dialog to adjust the duration. This method also requires left clicking on the right boundary of the audio tracks and dragging left or right to correspond to the length changes.
747 \subsection{Pan and zoom: still images}%
748 \label{sub:pan_zoom_still_image}
750 \CGG{}'s powerful keyframe features allow you to use pan and zoom effects on still pictures.
753 \item Load and create a clip from a still image. Make the clip 10 seconds long.
754 \item Activate the automatic generation of keyframes.
755 \item Using the transport controls, go to the beginning of the clip.
756 \item Using the compositing camera control set the clip's initial position.
757 \item Using the transport controls, move forward a couple of seconds on the clip.
758 \item Dragging on the compositing camera, move the camera center to a new position further along.
759 \item Next go to the beginning of the clip and play it.
762 You can see that the camera smoothly flows from keyframe point to next keyframe point, as \CGG{} automatically adjusts the camera movement in straight lines from point to point.
764 \subsection{Video lagging behind Audio}%
765 \label{sub:video_lagging}
767 When there is a lag between the Audio and the Video, it can be because there are a lot of video frames
768 and the computer can not display them as fast as the Audio can play the samples. However, this does
769 not affect the rendered media in that the Audio and Video will be correctly synchronized. When playing
770 the original media, you can alleviate the lag between the Audio and Video by disabling
771 \textit{Play every frame} in \texttt{Settings $\rightarrow$ Preferences, Playback A} tab. Now frames
772 will be skipped in order to keep the audio/video in synch.
774 \subsection{How to remove letterbox/pillarbox bands}%
775 \label{sub:remove_letterbox}
777 To remove the horizontal black bands of the letterbox or the vertical
778 black bands of the pillarbox we need to change the \textit{size} and
779 \textit{aspect ratio} of the source by cropping.
780 For example, if we want to remove the letterbox from a $4:3$ frame to
781 leave only the content with aspect ratio $3:2$, We have to change the
782 project format by doing the following steps:
785 \item Check the size of the base W of the original frame in pixels:
787 \texttt{Resource} window $\rightarrow$ \texttt{RMB} on Asset
788 $\rightarrow$ \texttt{Info} $\rightarrow$ \texttt{Detail}; e.g. W =
790 \item Obtain the height of the figure in $3:2$, i.e., without the
791 black bands; H can be obtained from the formula:
793 $\frac{3}{2} = \frac{W}{H}$ \quad from which $H = \frac{768 \times
794 2}{3}$ \qquad e.g., H = 512 px
795 \item Note that $W \times H = 768 \times 512$ is just the crop we
796 are looking for to switch from $4:3$ frame to $3:2$ frame without
798 \item Open \textit{Set Format} window: \texttt{Settings
799 $\rightarrow$ Format}
800 \item Change $H = 512$ and set \textit{Display Aspect Ratio} to
801 $3:2$; press \texttt{Apply} and \texttt{OK}. Note that we leave W
802 unchanged, since the frame width does not change.
803 \item If needed, use the \textit{Camera} tool to get the
807 \paragraph{Note:} in complex situations, with multiple sources of
808 different sizes, it may be appropriate to perform an additional
809 step first: change the size of the track on the Timeline via
810 \texttt{RMB $\rightarrow$ Resize track}.
811 In this way we crop the track to match it to the project format that
812 we will change in the next step. Thus we avoid possible unwanted
815 In the case of the pillarbox, we will leave H unchanged while
816 calculating the new value of W. The formula $\frac{x}{y} =
817 \frac{W}{H}$ is valid for any aspect ratio ($4:3; 16:9; 2.35:1$; etc).