What's New in Ocula 4.0
This page provides an overview of the updates included in Ocula 4.0.
Note: See the navigation bar on the left for links to specific release notes by version.
O_DisparityGenerator has been rewritten from the ground up to provide cleaner, sharper vectors and use GPU acceleration to perform the necessary complex processing, where available. The new vector engine is designed to provide clean separation in depth so that there is no roll-off between foreground and background regions, ensuring that background corrections do not get dragged along with the foreground.
Ocula's O_DisparityGenerator no longer requires O_Solver alignment data by default. The Properties panel Alignment control is set to 0 by default, which performs better when detecting changes in vertical alignment with depth. Ocula also performs better with O_VerticalAligner's Local alignment mode and produces better results when picture building in O_NewView.
Occlusion detection has been rewritten to make best use of the new clean, sharp disparity vectors. Occlusion regions are now clearly defined in the mask_occlusion channel to identify where attention is required.
O_NewView now produces sharper images with greater fidelity to the original footage. O_NewView has been updated to just rebuild one view from the other, including controls to handle occlusion and edge treatment.
Colour matching has been rewritten to make best use of the new clean, sharp disparity vectors and to simplify the controls. The matched results now have greater fidelity to the original image, allowing for smaller block sizes to produce more detailed colour updates, while preserving the noise of the source.
Focus matching has been rewritten to explicitly match high-frequency edges between views. O_FocusMatcher now looks and feels more like O_ColourMatcher and no longer has issues with ringing, pixel shifts that change parallax, and correlated noise when compared to Ocula 3.
A new Key Sequence button has been added to automatically analyse the source and place keys where a change in camera alignment is detected.
Adding and aligning user matches in O_Solver is now more accurate with the addition of a zoom window and single click workflow to create user matches while viewing an image difference. Hotkeys have been added to toggle the display between the left view and right views or display a mix of both views. You can now click Delete Auto Matches and solve the camera alignment with only user matches.
If you have an interactive licence (ocula_i and nuke_i), O_Solver can now be executed from the terminal. You can execute multiple frames, using Nuke's standard expressions and Python scripts, to automatically set up Ocula trees and render corrections without the use of Nuke's interface. You can also set user matches using Python.
The O_VerticalAligner workflow has been streamlined, allowing you to perform several alignment calculations using a single node in a fixed order. New controls have been added to enable you to fix plate scale and offset, change the filter used to transform pixels, and output a vaSTMap channel containing uv data for use in an STMap node.
O_VectorGenerator has been rewritten from the ground up to provide cleaner, sharper vectors and use GPU acceleration to perform the necessary complex processing, where available. The new vector engine is designed to provide clean separation in depth so that there is no roll-off between foreground and background regions.
A brand new node, O_MultiSample, has been introduced. O_MultiSample uses the channels controls and Sample and Mask inputs to expand the sampled channels to fill the image or Mask. This node can be used for different tasks, such as:
• Expanding colours into black regions when aligning a plate.
• Removing and in-filling colour corrections at unreliable pixels.
• Painting corrections into disparity/motion vectors at unreliable pixels.
Several pre-built Node Graph templates have been added to automatically set up Nuke and Ocula to perform certain tasks. The toolsets, accessed from Ocula 4.0 > Toolsets, provide example node trees with Backdrops to isolate various parts of the workflow and StickyNotes containing important information.
• FullPipeline - creates an entire workflow pipeline from reading in your stereo source, through various precomp steps to add the required disparity and corrections, to reviewing corrections before handing off to the next stage of post.
• InputReview - a simple pre-processing Node Graph to read in your source and check colour and alignment.
• Output Review - a simple post-processing Node Graph using the StereoReviewGizmo to performing some depth correction review.
• Precomp_Alignment - creates a section of the FullPipeline toolset dealing with aligning views and rendering out the results.
• Precomp_Disparity - creates a section of the FullPipeline toolset dealing with creating disparity and rendering out the results.
• Precomp_PlateMatch - creates a section of the FullPipeline toolset dealing with colour and focus matching and rendering out the results.
Ocula now includes a new quality control gizmo, DisparityReviewGizmo, located in the Ocula tool bar. The existing StereoReviewGizmo has been rewritten to make it quick and easy to review colour and alignment at different depths, review flicker and changes in depth, and bake out transforms to line up views for detailed review.