Copyright Vicon Motion Systems Ltd. All rights reserved. Information in this document is subject to change without notice. The software described in this document is furnished under a license agreement or nondisclosure agreement. The software may be used or copied within the terms of those agreements.
|Published (Last):||14 October 2010|
|PDF File Size:||6.91 Mb|
|ePub File Size:||18.54 Mb|
|Price:||Free* [*Free Regsitration Required]|
Quickstart Guide Expre NetApp Basic Concepts Portfolio QuickStart G You can now edit individual frames in a camera solution or create the framework of a solve by editing the camera position manually.
An approximate camera path can be used as the starting point for a more accurate solution. These new tools require a greater level of understanding of the inner workings of boujou and this document attempts to explain some of the new concepts and mechanisms. For more information about the embedded scripting contact 2d3support 2d3. The scripting will be covered in more detail in separate documentation. Image display controls — these can be found in the Zoom Tool tab of the right sidebar click on the button with a magnifying glass icon — see Appendix 1 for an explanation of the new interface icons.
Masks with holes Orthographic views. Top, front and side views can now be displayed to help with lining up objects and structure. Create the first mask 2. Create the second mask which will be the hole 3. Select the first mask and choose Manage Holes from the right-click menu The Subtract a Mask dialog will appear. Select second mask then click on the Make a Hole button. The second mask will be extracted from the first. The hole will move with the first mask, but you can also keyframe it separately.
When this button is pressed you will define a selection area from the path of the mouse cursor. If you press this button again you will deactivate lasso selection and return to the usual rectangular selection mode. You can add and subtract from the selection set by using the Ctrl key in the usual way. You have the option of deleting all existing solves when you run complete camera solving, or you can keep them in your project for reference or backup.
The Advanced solve refinement section allows you to optimize the camera path smoothness or correct for radial lens distortion as part of the solving process. Make sure that you have got the correct solve selected when you use the workspace views. You can switch between solves using the Solve dropdown list in the right sidebar.
You will be asked to select the solve that you want to adjust from the dropdown list because multiple solves can exist in the same project.
The Optimize radial lens distortion and Optimize camera path smoothness options can be selected when running an adjust solve the radial option will automatically evaluate the value of lens distortion in your shot. Camera Solving Process boujou 4 opens up its internal camera solving tools to the user for the first time to give direct control for solving problem shots.
Camera solving can be broken down into three main stages: Camera Initialization Obtain an initial estimate of the parameters of the camera, i. This is most difficult stage and the most likely to fail for problem shots where the feature tracking information is low quality or there is not enough parallax. Structure Generation For each track a 3D point will be computed using the solved camera. The reprojection error is examined and all points where this error is above a threshold will be discarded.
The reprojection error for a point in a frame is the distance between the position of the track in that frame and the position of the tracks 3D point projected into the frame using the cameras parameters on that frame. Reprojection Error To visualize the reprojection error, open a camera tracked project and make the following changes to a 2D view window 1.
You should see the reprojection errors increase as you fly the camera away from its starting position. This changes the camera parameters in order to minimize these reprojection errors. For example, cameras can now be generated in a large variety of ways including?
Regenerate Structure — this generates 3D points for all possible tracks. Filter Structure — removes the least accurate predictions based on their covariance.
Interpolate Views — this allows you to delete a range of camera views and replace them using a straight line interpolation. Copy Solve — creates a copy of the current active solve. Several solves can co-exist in a project file and copying a solve is a good way of creating a backup of your solve before you start editing it. Create User Solve — this will create a solve that contains no camera views or structure. This is the starting point for creating your own camera views. Default View from Known 3D — creates a new camera view based on the coordinate frame of the imported model.
The camera will be positioned so that the model is in the centre of its field of view, pointing down one of the axes. Delete View — deletes a single frame from the current solve.
Copy View — copies a view from a specified frame to the current frame. Optimize View — optimizes the camera position and orientation in the current frame by minimising the error between the measured features and the 3D predictions. Regenerate View — recreate the view on the current frame using the existing 3D structure.
Before boujou does any camera tracking it examines the feature tracks and breaks up the shot into sections which are linked by a certain number of tracks.
These breaks are called solve keyframes and their distribution is called the key schedule. Shots with plenty of long, consistent tracks may just have a solve keyframe at the start and the end.
Shorter, more sporadic tracks may result in lots of keyframes. When boujou calculates a camera solution it will create views at the keyframes and then fill in the gaps. Feature tracks that span less than 3 keyframes will be discarded. If the key schedule is very sparse then certain sections of the shot may have very few 3D predictions, even though they seem to lots of feature tracks.
You can add extra solve keyframes by adding a locator each locator keyframe will create a new solve keyframe. The cloud of 3D points calculated by boujou after camera solving. As a test object As survey data To help with the approximate positioning of the camera For face-based object tracking Objects can only help the camera tracking if there are enough feature tracks to link the object to the image.
If you have used the survey points tool in boujou three then you will be familiar with the idea of locator keyframes having to be linked by feature tracks. The locators are connected to the survey points and each frame of the sequence that has sufficient locator keyframes will define a camera view.
It may not be clear which of the model-based tracking methods to use. If the vertices of your 3D model correspond to 2D features in the image then you should try the survey points solve from known 3D. If your model vertices do not correspond to vertices in the image then you can use facebased tracking. Survey data To use a model as survey data you will need to have locators or target tracks in your scene to identify the features that you have survey data for, and you will need to have feature tracks linking the keyframes.
This is exactly the same as the survey data tool in boujou three. However, you can now link the locators to the model vertices, which makes setting up your scene far more straightforward. As before, you need at least 7 locator keyframes on a frame and at least 10 feature tracks joining this frame to the next set of locator keyframes. Face-based tracking Face-based tracking links the feature tracks and the polygon faces in the model to calculate the camera motion.
If you have too few features tracks, or if they are poor quality, then the accuracy of the tracking will be low and the model will appear to drift. Vertex alignment Vertex alignment is a method of approximating the position of the camera. You can align the model with using the Fly Align method using the workspace navigation tools , or the object manipulators, or the Vertex Drag method. Adding subsequent springs will change the camera position until the model appears to be aligned with the image.
If you have the optimize focal length option switched on then the focal length will also be adjusted to match the model alignment. You can go in and out of Fly Align mode or Vertex Drag mode by clicking on the buttons in the toolbar. A large circle with a horizontal line through it will appear when you are in Fly Align mode to indicate that you are moving the camera rather than moving an object. It also warns you that the workspace navigation tools will not behave in the usual way e.
Solve from Existing Cameras Once you have set up approximate camera positions on various frames throughout the shot you can use the Solve from Existing Cameras function from the 3D Tasks menu. At this stage you can start using it as a test object. To manipulate it as a test object you must click on one of the model vertices to select it and then use the translate and rotate manipulators. Manipulating the model When you manipulate the model in the 2D view prior to camera solving you are actually moving the camera.
Vertex alignment is a method of approximating the position of the camera. You can align the model using the Fly Align method using the workspace navigation tools , or the object manipulators, or the Vertex Drag method.
The mesh is generated by projecting selected 3D points into 2D and then creating connectivity information in 2D which will then be used to mesh the 3D structure. The 3D Tasks menu gives you three different projection methods: in the direction of the current camera view, perpendicular to the ground plane, or in a user specified direction using the 3D view.
Only the selected predictions will be used to generate the mesh. Multiple meshes can be created and exported. It may be useful to view only the active 3D points those whose track has a feature in the currently selected frame when generating meshes so that points arising from non-continuous surfaces are not connected together.
The 3D points used can either be from manual tracks locators and target tracks or from those automatic tracks that have had 3D points generated. When generating 3D models it may be useful to generate more 3D points than are produced by a default camera solve. When generating meshes it may also be useful to filter out those tracks whose 3D position is poorly constrained e.
It can only be used when you are in the 3D viewing mode. This will open the Export Meshes dialog. Multiple meshes can be exported. Go to the Scripts menu and choose Export Meshes to Maya. The workflow described below is the most effective way to use a model to create a camera solve when the model vertices correspond to features in the image. In boujou three you were able to input survey data and use it to improve the 3D structure of your camera solve.
Typing in the coordinates and linking them to locators was a time-consuming process and it was easy to make mistakes. It is now possible to bring in your survey data as an OBJ model and then connect the vertices interactively.
boujou 5 Review: Matchmoving Enters its Maturity
Developed for film and television post and VFX facilities, boujou has long been the industry standard for matchmoving, allowing users to add computer-generated effects seamlessly into film or video footage. When boujou was first introduced in , it provided a completely different approach to motion tracking. Before boujou, matchmoving was an expensive and highly skilled process, adding significantly to post-production costs. Everything had to be done by hand in key-frame animation, or with laborious point-based camera tracking. Despite its groundbreaking status, boujou was still very much a hit-or-miss proposition when it first came out.
2d3D, boujou on track.
Home Software Boujou. It has a unique way of solving which radically accelerates your results, solving more shots in less time. Take the headache out of matchmoving. Two clicks and the majority of your shots can be matchmoved automatically.
Quickstart Guide Expre NetApp Basic Concepts Portfolio QuickStart G You can now edit individual frames in a camera solution or create the framework of a solve by editing the camera position manually. An approximate camera path can be used as the starting point for a more accurate solution. These new tools require a greater level of understanding of the inner workings of boujou and this document attempts to explain some of the new concepts and mechanisms.