Hasta La Motion!

This is a major new release of Ocula with one new plug-in and many improvements and bug fixes.

Release Date
July 2010

Requirements
•Nuke 6.0v7 or a version of Nuke 6.1 on

•Windows XP SP2, XP64
•Mac OS X 10.5 “Leopard” and 10.6 "Snow Leopard" (x86 only)
•Linux CentOS 4.5 (32- and 64-bit)

•Foundry FLEXlm Tools (FFT 5.0v1 or later) for floating licenses.

New Features
•There is a new O_DisparityViewer plug-in, which lets you visualise the disparity vectors in your node tree. You can add it after any node in the Node Graph. As long as there is a disparity channel at that point in the tree, O_DisparityViewer produces a Viewer overlay with arrows showing the disparity vectors at regular intervals. For more information, see “DisparityViewer” on page 76.

Improvements

•O_Solver

•You can now add your own feature matches to the automatically detected ones by right-clicking on the image and selecting add feature. This allows you to get a good solve when there aren’t many good automatically detected matches. For more information, see “Solver” on page 15 and “Visualising and Editing the Results” on page 18.
•There is a new Luminance Correct control, under Features. This matches the luminance between the two views before searching for feature matches. Where there are large differences in in luminancebetween the two views, this can increase the number and quality of feature matches found.

•O_DisparityGenerator

•You can now sample over multiple disparity field detail levels (different resolutions of the images). This can help to reduce errors in the calculated disparity field. There are controls for the number of samples (Number of Samples), the minimum disparity field detail to go down to (Minimum Detail Level), and the sample spacing (Sample Spacing).
•There is a new Image Alignment menu with the following options:
Rectification – This is the default alignment method. The two images are resampled so that all matching pixels are on the same horizontal scanline in the second image as they are in the first.
Vertical Alignment – The two images are aligned along the y axis using a skew, but not moved along the x axis.
None – If the disparity between your stereo views is horizontal only (corresponding pixels lie on the same scanline in both images), you can select this option for faster processing. This is the same as Horizontal Shift Only in previous versions of Ocula.
•The Occlusions menu has been renamed Disparity Method. A third method, Unconstrained Motion, has been added to the menu. This calculates the disparity using unconstrained local motion estimation.
•There is a new Median Filter Size control. Increasing this value should reduce the noise on the disparity field. This control is only available when Disparity Method has been set to Normal Occlusions.

For more information on the new controls, see “Controls” on page 30.

•O_ColourMatcher

•There is a new Pre-blur Disparity control, which allows you to blur the incoming disparity map before using it. If the disparity map is imperfect, this can help to reduce artefacts in the colour correction.
•In the Block-based Matching Mode, you can now calculate the colour correction for multiple block sizes and then blend the results together. This can help to reduce errors and make the results more temporally consistent. There are controls for the number of samples (Number of Samples), the maximum block size to go up to (Max Block Size), the sample spacing (Sample Spacing), and the method of combining the samples (Colour Correction Type).
•There is a new control, Halo Correct, aimed at reducing the halo effect you can get around high-contrast edges in the Block-based Matching mode. Note that where occlusions occur this correction can introduce artefacts around edges. Another new control, Occlusion Compensate, is designed to correct these and probably needs to be tuned to your particular footage.

For more information on the new controls, see “Controls” on page 49.

•O_VerticalAligner

•If you have a pretracked Nuke stereo camera that describes the camera setup used to shoot the Source images, you can now use O_VerticalAligner to analyse the sequence and output a vertically aligned camera pair. This allows you to continue using pretracked cameras once your footage has been vertically aligned. This works in all vertical alignment modes except Vertical Skew (which can’t be represented by a camera transform).
Once O_VerticalAligner has analysed the sequence, you can see the analysis data in a Transform Matrix on the Output tab of the node controls. This represents a 2D corner pin that can be applied to the input image to create the same result as O_VerticalAligner. If necessary, you can take the matrix to a third-party application, such as Baselight, and align the image or camera there. There is one matrix for each view in the source.
For more information, see “Creating a Vertically Aligned Camera Pair” on page 43.

Fixed Bugs
•BUG ID 9188 – O_ColourMatcher: In block-based matching mode, colour fringing sometimes occurred where there were high-contrast regions in the input images. You can now reduce this by using Halo Correct, Occlusion Compensate, and the sampling controls.

etc.

Full Release Notes

Product Website