Video Parameters

VideoOutput (/VO=0) ⇧ Range of values [0-8] With the parameter VideoOutput you can create MP4 or AVI files, if you are in Videomode (InputMode = 0 or InputMode = 3). The creation of the single images (with parmeters LeftRightOutput, StereoOutput, AnaglyphOutput and ExtraOutput) is independent of that and can be switched on or off completely. 0: No videos are created. 1: Creation of two separated videos in the cl- and cr-directories. The audio is only embedded into the left video. 2: Creation of a side-by-side (SBS) video in the cs-directory and -lr indicator. 3: Creation of an anagylph video in the ca-directory. 4: Creation of a 1:2 squeezed side-by-side mounted (SBS-H) video and -lrq indicator. 5: Creation of a top-and-bottom (TAB) mounted video with left on the top and -abq indicator. 6: Creation of a 1:2 squeezed top-and-bottom mounted (TAB-H) video and -abq indicator. 7: Creation of an interlaced video (first line left, e.g for ACER 5740DG), indicator -oe or -eo. 8: Creation of a dualstream video (two embedded streams). 9: Creation of two separated videos in one single clr-directory. 10: Creation of a 360°-3D-panorama image from two roundshot videos according to flat stitching method. Remark: up to version 0.9i3 this parameter was named AviOutput.

VideoContainer (/VC=0) ⇧ Range of values [0,1] Selection of AVI or MP4 videos. 0: Creation of a MP4 video container: only VideoCodec = X264 is possible (internally stored, without further installation)! 1: Creation of a AVI video container: one of the VfW codecs installed in the system is selected as VideoCodec (e.g. x264vfw64).

VideoCodec (/CC=????) ⇧ Range of values: characters If VideoContainer = MP4 there is no choice, only X264 is possible. If VideoContainer = AVI, this parameter is used to select the Vfw video codec to be used for creating the avi. There are four different forms to set this parameter (I, II, III or IV): I: ????: If you set this parameter to ????, a dialog window will be opened, where you can choose one of your installed video codecs and set the quality values for it. The settings of the dialog box will be stored into a file cosima_codec_CCCC.bin. II: CCCC: Four character code (four letter code), which definitely identifies the codec. Example: "LAGS": Lagraith, "DIVX": DivX, "X264": H264-Clone, "DIB ": uncompressed. In that case the following default values are used: quality = 95%, bitrate = 0.3*bit*pixel*framerate and one keyframe every 100 frames. III: CCCC|file: The quality values will be read from the file cosima_codec_CCCC.bin. IV: CCCC|Quality|kBitrate|Keyframes: Explicite form of this parameter with the following detailed meaning: CCCC: Four letter code of the codec, Quality [0-100]: quality value in percent, default 95% kBitrate [0-100000]: bitrate in kBit/s, default 0.3*pixel*frames/s Keyframes: amount of frames between 2 keyframes, default 100 Example: VideoCodec = IV50|95|30000|100: Ligos Indeo Codec with 95% quality, bitrate 30MBit/s, and one keyframe every 100 frames. Remark 1: Some codecs use in III their own defaults and ignore in VI the settings concerning the quality. These codecs can be configured only by the dialog window or with the file cosima_codec_CCCC.bin. Remark 2: The encoding takes place either with a variable bitrate (VBR) or with a constant bitrate (CBR). With VBR encoding, the encoding quality is given by the quality value and with CBR encoding with the bitrate. It depends on the codec, which of these two methods will be adopted and which of the two parameters is therefore relevant. With some codecs, the encoding method itself can also be choosen. Remark 3: The X264 codec is available as VfW codec (in the AVI container) or implicitly in the MP4 container. In both cases, the encoding performance is identical. In the AVI container the codec is completely configurable (via the codec selection window), in the MP4 container six quality levels are preconfigured. The frame packing parameter is set automatically.

VideoQuality (/VQ=3) ⇧ Range of values: [0-5] Preconfigured quality level for X264 codec in MP4 container. 0: very low: creates a video with a constant bitrate ~ 4MBit/s for FullHD 1: low quality: creates a video with a constant bitrate ~ 7MBit/s for FullHD 2: presentation: creates a video with a constant bitrate ~ 10MBit/s for FullHD, recommended for theater projection 3: enhanced quality: creates a video with a constant bitrate ~ 15MBit/s for FullHD, recommended for home projection 4: further processing: creates a video with a constant bitrate ~ 20MBit/s for FullHD, recommended for videos for further processing in presentation software, e.g. m.objects 5: uncompressed (only 4:2:0 subsampling)

VideoFramerate (/AF=0.0) ⇧ Range of values [0-100.0] Amount of images per second. Attention: If this is value deviates from the frame rate of the original video, the video will be created without sound. VideoFramerate = 0.0: If this value is set to zero, the framerate of the original video will be used for the videos to be created (0.9h3 and later).

AudioSource (/AS=-1) ⇧ Range of values [-1,0,1,2] Selects the source for the audio stream. For left-right separation the audio stream is embedded in both videos. -1: The audio stream will be selected from the left original. 0: No audio stream will be included into the resulting video. 1: The audio stream will be selected from the right original. 2: For the resulting video silence will be included.

FirstFrameIndex (/FX=0) ⇧ Range of values [0-262143] Index of the first frame or image, which is used from a stream or a series from single images in the VideoMode. This index starts normally with zero. werden soll. Der Index startet normalerweise bei Null.

FrameIncrement (/FK=1) ⇧ Range of values [1-32767] Index increment for the frames of the output images and output videos.

LastFrameIndex (/LA=262143) ⇧ Range of values [0-262143] Index of the last frame or image, which is used from a stream or a series from single images in the VideoMode.

StartforAnalyse (/SR=0) ⇧ Range of values [0-262143] This is the index of the first image, which is used for the analysis in the VideoMode (InputMode = 3). So you can skip glued joints of movies for the analysis. This index starts normally with zero.

AnalyseIncrement (/AI=25) ⇧ Range of values [1-32767] During the first run of the video mode not every image pair has to be analyzed. This parameter tells Cosima the index incrementation for the image analysis.

EndforAnalyse (/EA=262143) ⇧ Range of values [0-262143] Index of the last image, which is used for the analysis in the VideoMode (InputMode = 3).

FrameOffset (/FO=0) ⇧ Range of values [-128 ... 127] Frame offset. That is the number of frames to shift the two streams against each other (right frame index = left frame index + FrameOffset). It should be positiv, when at a 1:1 comparison the left stream show a delay. In that case, the left camera has started the record too late.

AutoFrameOffset (/AE=0) ⇧ Range of values [0,1] With the tick in the box "Automatically", the frame offset is automatically searched using the embedded audio information. The prerequisite for this is a sufficient sound as, for example, the sound of a clapperboard at the beginning of the video.

SmoothGeometryCorr (/SG=3) ⇧ Range of values [0-5] Selection of the averaging method for the correction parameters and, if existing, the floating windows: 0: Constant mean value. Recommended for shootings without zoom. 1: Linear interpolation between two analysis values. Recommended only for shootings with aprupt changes of the shooting parameters. 2: Median filtering with 3 keypoints 3: Median filtering with 5 keypoints 4: Median filtering with 7 keypoints 5: Median filtering with 9 keypoints Hint 1: The effective time constant is dependent on AnalyseIncrement. With AnalyseIncrement = 25, SmoothGeometryCorr = 2 yields a time constant of 1 second, SmoothGeometryCorr = 3 yields a time constant of 2 seconds, SmoothGeometryCorr = 4 yields a time constant of 3 seconds and SmoothGeometryCorr = 5 yields a time constant of 4 seconds. Hint 2: SmoothGeometryCorr has no influence to the setting of the stereo window.

SmoothWindowCorr (/SW=3) ⇧ Range of values [0-5] Selection of the averaging method for setting the stereo window: 0: Constant mean stereo window. No reaction of changes during the scene. 1: Linear interpolation between two analysis values. (Only for special cases.) 2: Median filtering with 3 keypoints 3: Median filtering with 5 keypoints 4: Median filtering with 7 keypoints 5: Median filtering with 9 keypoints

SmoothFlicker (/FF=-1) ⇧ Range of values [-1:5] This parameter activates a time-domain filter that specifically compensates for exposure jumps, which is caused, for example, by aperture flicker during time-lapse photography. -1: Filter is off. 0: Creates a constant brightness throughout the video. 1: Linear interpolation of brightness between two analysis values. (Only for probe cases.) 2: Median filtering of the brightness at the analysis base points over 3 values. 3: Median filtering of the brightness at the analysis base points over 5 values. 4: Median filtering of the brightness at the analysis base points over 7 values. 5: Median filtering of the brightness at the analysis base points over 8 values. Note: In practice, SmoothFlicker = 0 is especially important. Then the brightness is kept constant in the whole video. Dynamic changes are possible for SmoothFlicker = 1,2,3,4,5, the higher the value, the smoother the brightness curve.

SetWindowbyZoom (/SZ=0) ⇧ Range of values [-100, 100] The Video Previewer enables a dynamic post zooming into the image. This will enlarge the deviation proportional to the zoom factor and objects behind the the stereo window will wander undesirably into the background. This parameter SetWindowbyZoom enables you to compensate this effect by including an additional horizontal shift, which depends on the zoom factor. The parameter value is the window shift, measured in promille of the image width for an zoom factor of 2.0. Example: Setting SetWindowbyZoom = 10, the stereo window will be shifted 10 promille into the background, if a zoom with a zoom factor of 2.0 is applied.

InterlacedMode (/IL=0) ⇧ Range of values [0,1,2,3,4,5,6] If the images to be processed origin from PAL or NTSC sources, the even and the odd lines emerge from different sampling times (interlaced). Cosima will separate these image information and calculate from it two complete separate images. Hereby the missing parts of the images are estimated, thus doubling the total number of images (50 frames/s for PAL or 60 frames/s for NTSC). 0: InterlacedMode off, each pair of images is considered as independent (progressive film). 1: InterlacedMode on. To each input image pair <name_l.jpg> and <name_r.jpg>, two output image pairs are generated, for example: <name_0cl.jpg> and <name_1cl.jpg>, and <name_0cr.jpg> and <name_1cr.jpg>. To interpolate the missing pixel information only images of the same time point are used (field interpolation). 2: As 1, in addition, one frame delay on the right channel (1/50 second for PAL). 3: As 1, in addition, one frame delay on the left channel (1/50 second for PAL). 4: Optimal linear deinterlacing: As 1, but for image estimation, both, image information at the same time sample (field interpolation) and image information of earlier and later frames (frame interpolation ) are used. 5: As 4, in addition, one frame delay on the right channel (1/50 second for PAL). 6: As 4, in addition, one frame delay on the left channel (1/50 second for PAL). Hint: With internal deinterlacing there is no deshaking possible. So if one of the above mentioned options is selected, DeshakeMethod = 0 must be set.

FirstLineIndex (/FI=0) ⇧ Range of values [0,1] Specifies the index of the first line for interlaced frames: 0: Straight lines first (PAL Camcorder). 1: Odd line first.

DeshakeMethode (/DM=0) ⇧ Range of values [0-4] With this parameter, the video will be deshaked simultaneously with the correction. The values ​​used for the deshaking will be expected in the file <StereoFileName>.log for stereo videos and in the files <LeftFileName>.log and <RightFileName>.log for left/right separated videos. These files are generated by the menu entry Parameter - Deshaker Analysis Run. 0 : No Deshaking will take place. 1: Deshaking enabled: The strength of the stabilisation is set with the DeshakeFilterX and ~Y parameters. The necessary compensation of the image loss is calculated and applied automatically. The parameter DeshakeZoomMax serves as an upper limit for the scaling factor. 2: Same as 1, but the deshaking is reduced to such an extent that no black edges appear when DeshakeZoomMax is applied. 3: Deshaking with pan detection. Like '1' with the following additional automatic: In areas with strong video pans, an alternative filter method is used, which reacts much faster to changes in the image content. Also this method should avoid the occurrence of temporary black edges and delivers in most cases a better overall result than method '2'. 4: A completely different approach is pursued with this method '4'. When panning, an attempt is made to remove missing image content with previous frames. This works a bit better with calm monochromatic image content, but with fast movements and with many image contents less good. The Previewer offers a preview of this method, also for setting manual masks. To to enable the preview in Previewer, two prerequisites are necessary: The scroll cache must be activated and the deshake method '4' must be set before the Previewer is started. For performance reasons, only the left image is processed in the Previewer, in the right image the corresponding parts of the image remain black. Attention: This method only works with FrameIncrement = +1 and is very slow! Attention: The calculation of the maximum necessary zoom factor assumes DeshakeMethod = 1 and is therefore to be pessimistic. Note-1: For the deshaking analysis run, you need to have the AviSynth plugin Deshaker The file deshaker.vdf must be located in the Cosima directory. Note-2: There is no internal deinterlacing during deshaking. So if one of the above options is selected, InterlacedMode = 0 must be set.

DeshakeZoomMax (/DZ=10) ⇧ Range of values [0-100] With this parameter, the maximum zoom factor to compensate the black edges can be given in percent. 0: No up-scaling at all, deshaking always causes black edges. 10: Maximum zoom factor 10%, this will scale-up the image by the maximum factor 1.1. 100: Maximum zoom factor 100%, this will scale-up the image by the maximum factor 2.0. Note-1: The deshaking result can be examined with the video previewer. There can be seen the deshaked raw image with the black edges. In order to cut off the black edges, individual dynamic masks can be set. Using of the dynamic masks have always precedence before the election of the parameter DeshakeZoomMax. Note-2: With DeshakeMethod = 4 the zoom factor is the same as with DeshakeMethod = 1 and is therefore too pessimistic. It is therefore recommended to set DeshakeZoomMax = 0 and to set the necessary masking manually in the Previewer.

DeshakeFilter (/SX=5.0) ⇧ Range of values [0.0-20.0] Deshaking in the x-direction: The deshaking raw values ​​are additionally filtered with a high pass cut-off frequency = 1.0/DeshakeFilterX. The parameter DeshakeFilterX corresponds to the time constant of this high-pass filter in seconds. Higher values ​​for DeshakeFilterX let the video react less to camera movement and with smaller values the video will shake more. With DeshakeFilterX = 0, in x-direction no longer deshaking takes place. The given filter value are only applied with the Use box checked. Without marking the Use box, the deshaking value are applied unfiltered (total deshaking) and the result corresponds to a fixed video camera on a tripod, at least almost.

DeshakeFilter (/SY=5.0) ⇧ Range of values [0.0-20.0] Deshaking in the y-direction: The deshaking raw values ​​are additionally filtered with a high pass cut-off frequency = 1.0/DeshakeFilterY. The parameter DeshakeFilterY corresponds to the time constant of this high-pass filter in seconds. Higher values ​​for DeshakeFilterY let the video react less to camera movement and with smaller values the video will shake more. With DeshakeFilterY = 0, in y-direction no longer deshaking takes place. The given filter value are only applied with the Use box checked. Without marking the Use box, the deshaking value are applied unfiltered (total deshaking) and the result corresponds to a fixed video camera on a tripod, at least almost.

DeshakeFilter (/DR=5.0) ⇧ Range of values [0.0-20.0] Deshaking of rotation: The deshaking raw values ​​are additionally filtered with a high pass cut-off frequency = 1.0/DeshakeFilterR. The parameter DeshakeFilterR corresponds to the time constant of this high-pass filter in seconds. Higher values ​​for DeshakeFilterR let the video react less to camera movement and with smaller values the video will shake more. With DeshakeFilterR = 0, no deshaking for rotation takes place. The given filter value are only applied with the Use box checked. Without marking the Use box, the deshaking value are applied unfiltered (total deshaking) and the result corresponds to a fixed video camera on a tripod, at least almost.

SynchronizeDeshake (/SD=0) ⇧ Range of values [0,1] Optional joint deshaking of left and right video. 0: Left and right deshaking values are independent. 1: Left and right deshaking values will be forced to be equal.