Check data quality
After you have loaded your captured data files into Vicon Shogun Post, you can use its diagnostic tools to help you to identify any issues with the data.
See also the Vicon videos:
and
Load mocap data files into Shogun Post
Shogun Live can record processed real-time data (.mcp files) along with raw 2D camera data (.x2d files). You can use .mcp files as a starting point for processing in Shogun Post. Because the frame rate during real time can vary, .mcp files may contain missing (dropped) frames. When you open an .mcp in Shogun Post, if dropped frames are detected, a dialog box prompts you to confirm whether you want to reprocess the .mcp. The fixing operation requires an .x2d with the same name as the .mcp file to exist in the same folder.
- To load mocap data files (.mcpthat contains everything captured in Shogun Live, or .vdf saved in Shogun Post) you can:
- Open Windows Explorer, locate the files you saved and drag and drop them into the Shogun Post view pane; or
- On the the File menu, click Open and locate your files; or
- In the Data Management panel, locate the required file icon (an .mcp has a purple icon) and double-click it.
Although the recommended workflow is to load an .mcp file (processed real-time data), you can instead import an .x2d (raw 2D data), but in this case, you must run reconstruct, label, and occlusion-fixing operations before proceeding with any required cleanup. If you chose an .mcp file and if dropped frames are detected, a dialog box informs you of the percentage of dropped frames in the file and asks whether you want to reprocess to fix them. - To automatically reprocess the file to fix the dropped frames, click Yes.
The.mcp file is opened into the scene and the entire take is reprocessed.
The dropped frames are reconstructed, labeled and solved,smoothly integrating data from the related .x2d file into the existing data.
(If you click No, processing is quicker as Shogun Post just interpolates between the dropped frames, but note that no keys are created and you will still need to fix the dropped frames.) - You can now decide if you want to save the scene.
Fast MCP reprocessing
With Shogun 1.4 and later, you can use the following fast and easy ways to remove dropped frames from .mcp files in Shogun Post. This enables you to save time by reprocessing the .mcp files without having to load them into the scene.
- Fast MCP reprocessing with HSL
- Fast MCP reprocessing with ShogunPostCL
- Fast MCP reprocessing in batch mode
Fast MCP reprocessing with HSL
You can now use new options added to the quickPost command to perform drop frame fixing directly to .mcp files. For details, see quickPost in the HSL Scripting with Vicon Shogun.
Fast MCP reprocessing with ShogunPostCL
You can also run the quickpost command to reprocess .mcp files directly from ShogunPostCL, for example:
- To open ShogunPostCL, navigate to: C:\Program Files\Vicon\ShogunPost1.6 and then double-click ShogunPostCL.exe to open the terminal.
-
In the ShogunPostCL terminal window, enter:
quickPost processingLevel -readMCP "C:\\myPath\\myInputFile.mcp" -writeMCP "C:\\myPath\\myOutputFile.mcp";
Where:
-
processingLevel
= Any one of:reconstruct
,label
,solve
orretarget
(Note that if you specifyretarget
, the relevant retargeting setup must exist.) -
The path after
-readMCP
= The path to the .mcp file on which you want to run drop frame fixing. Ensure that the related .x2d file is located in the same folder. -
myInputFile.mcp
= The name of the .mcp file on which you want to run drop frame fixing -
The path after
-writeMCP
= The path to the fully processed output .mcp file -
myOutputFile.mcp
= A name for the fully processed .mcp file
-
Fast MCP reprocessing in batch mode
If required, you can also fix dropped frames directly on multiple files in batch mode, as follows.
To directly reprocess MCP files in batch mode:
- In the Batching panel, set the mode to Post Process.
- In the Settings section, ensure Reprocess MCP is selected.
- Set the required file path and naming options.
- In the Processing panel, set the required processing settings.
- In the Batching panel, expland the Files and Procession section, click Add Files and add the required .mcp files to the batch.
- Click Start.
The.mcp files are processed without being loaded into Shogun Post.
When you load the new .mcp file into Post, it is fully processed, with no dropped frames.
Review occlusion fixing
When you import an .mcp file or perform any processing that includes occlusion fixing, your first step is to review it to ensure that it is labeled correctly and that the automatic occlusion fixing worked well.
To review occlusion fixing in Shogun Post, both of the following (default) settings must be selected:
-
In Shogun Live: Before capture, in the General section of the Processing panel, Occlusion Fixing must be selected.
This normally produces higher quality results, especially during occlusion-heavy moves such as interactions and when a subject is on the edge of the volume. Occlusion fixing is stored as a layer when the .mcp data is recorded. - In Shogun Post: Before importing the .mcp file that was captured with Occlusion Fixing selected as described above, in the Preferences dialog box (General > Preferences), on the File Import tab, in the File Format field, select MCP File and ensure Apply Fixed Markers is selected.
You can then compare the data before and after occlusion fixing in a 3D Scene view (or any other 3D view) in Post.
Note that, by default, in the 3D Scene view Show All Clips is selected:
This is necessary to enable you to view data from the backup clip, which contains the data before occlusion fixing, as well as the current (occlusion-fixed) clip.
If required, you can revert back to the raw data after loading the .mcp file into Shogun Post by using the Restore section of the Marker Editing panel.
Note that the occlusion fixing algorithm affects all the markers on a subject. For this reason you should run occlusion fixing only once: either during capture in Shogun Live, or at the end of data processing in Shogun Post.
Before you start your review, ensure you can recognize the various types of data that may be displayed in the 3D Scene view, as listed below.
Symbol in 3D view | Description | Data type |
---|---|---|
Circle | Original labeled markers from the backup clip. Displayed by default if the difference between the position of the original marker and the occlusion-fixed marker is greater than 1 cm. | |
Sphere | Markers without gaps | |
Sphere with a dot in the center | Markers that have been occlusion-fixed during a gap | |
Sphere with a cross in the center | Markers that have had a gap manually filled | |
Red sphere | Markers that are missing at the current frame | |
Wireframe box | Constraint, showing expected location of marker |
If circles are displayed for some markers during some time ranges, it indicates that the marker position was changed by more than 1 cm by occlusion fixing.
In this case, check the position of the occlusion-fixed marker. If it doesn’t look right, the cause is likely to be either a labeling mistake or an occlusion-fixing error. To fix these problems, see Correcting labeling mistakes.
Correcting labeling mistakes
If you notice any swaps, mis-labels or other issues with marker data and the file has already been occlusion fixed, the recommended workflow is to restore the data back to its original state before it was occlusion-fixed.
You can do this on a per subject basis using the Restore section of the Marker Editing panel. Normally, restore the data across the whole take, because occlusion fixing across a range leads to jumps in the data.
The recommended workflow for dealing with problematic occlusion-fixed data is:
- From the Current Subject chooser, select the required subject, or if you want to work on everything, select All.
- In the Restore section of the Marker Editing panel, with selected subject active, click Restore All Data.
- Fix any labeling issues that are present in your data. For more information, see Use the Labeling panel and Fix common labeling issues.
- Re-run occlusion fixing from the Processing panel.
- Re-run Solve Solving to update the solving skeletons.
Check for swaps and other errors
Before filling gaps, you must find and fix any labeling swaps or errors in your data. Gap-filling in Shogun relies on the data before and after the gaps to predict marker location, so any labeling errors will cause gap-filling to fail.
To identify swaps and other errors, on the time bar, click the Play button
to view your data in the 3D Scene view. For a closer look, slowly scrub through the whole take (drag the current time indicator along the timeline, or for finer control, press the A or S key), carefully checking for any 'popping', unexpected movement or obvious misalignment.For a more detailed view, use the normal mouse actions to move around the view: drag to rotate the view, right-click and drag to zoom in or out, left- and right-click and drag to pan. If you need to view the image from a different angle, you can use one of the orthogonal views.
Use the View Filters options in the 3D Scene view to make it easy to recognize any issues. For more information on recognizing and correcting swaps and other errors, see Fix common labeling issues.
To snap a 3D view camera to selected object(s), press C or click the Snap button in the 3D Scene view toolbar.
Use video overlay to check accuracy
If your system includes Vicon video cameras, you can check your data accuracy by comparing optical capture data with supported video data. You do this by overlaying the 3D data onto the video (.vvid or .mov) data.
Shogun Post supports the following types of video files:
- .vvid files captured from Vicon Shogun Liveusing Vicon Bonita and Vicon Vue cameras
- .mov files that have been transcoded using Shogun Post or the standalone ViconVideoConverter tool (a command line tool that is installed with Vicon Video Viewer, by default to C:\Program Files\Vicon\ViconVideoViewer).
To overlay 3D data onto video files in Shogun Post:
- Ensure that the video files are in the same folder as the required 3D data (.x2d or .mcp) file. If a different path was defined during capture (often an SSD per pair of cameras), use the batch transcoding feature of Shogun Post to ensure the files are moved to the correct location. For more information, see Transcode video files.
Tip: If video files are kept in a different folder from the 3D data files, when you load an . x2d or . mcp file, Shogun Post warns you that the video files cannot be located. To manually specify the video file path, in the Selection panel (or 3D Scene view) select the video camera, then in the Attributes panel, expand the Video (Offline) section and set its Video_File attribute to the path of the file. This change is saved when you save the scene in Shogun Post.
- Open the .x2d or .mcp file (either from the Data Management panel, or click Import on the File menu, or drag and drop the relevant file from Windows Explorer, as described in Load mocap data files into Shogun Post).
The video files are automatically loaded. - In the Selection panel, expand the System node and click the required camera.
- At the top left of the view pane, click the current View type button, and then click Cameras.
- At the top of the Cameras view, ensure View Filters is selected.
- Select the 3D check box.
- The video for each Vicon video camera is displayed in a 3D overlay.
Video is in black and white or color, depending on the video camera used. - To see the image plane of the camera corrected for lens distortion, select Distort 3D.
The 3D overlay should line up with the subject in the Cameras view and the grid should line up with the floor. You can zoom and pan the video using normal Cameras view controls. You can select or clear the View Filters to control which 3D elements are displayed.
Get an overview with the time bar Issues map
To enable you to identify frames with issues, between the play range (green vertical) bars on the time bar, two horizontal colored bars are displayed. For each frame:
- The top bar indicates the percentage of markers that are labeled.
- As the top bar percentage may include markers that are incorrectly labeled, to help you identify issues further, the lower bar indicates the percentage of markers that have a high solving constraint error (that is, the number of markers whose solving constraint error is considered too large). This can indicate issues like mislabels, swaps, or a poor solve due to an incorrect fill, over filtering, etc.
For both bars, the color varies from yellow to red, depending on the number of markers with issues at each frame. In each bar, the longer the yellow part exists, the longer the issue exists, whereas the more red it turns, the more markers have issues. To show more information about the errors found at a particular frame, hover the mouse pointer over the relevant frame.
The issues map is all red if there is no labeled data in the current scene.
You can use the mouse to work with the time bar as follows:
To | Do this |
---|---|
Show a tooltip that explains the issues that are present at this frame | Hover the mouse pointer over the frame. |
Set the current time | Click on the time bar or drag the current time indicator. |
Set the current time (ie select a frame), highlight the markers that are missing labels or have high constraint errors, and snap the camera to them | Using the tooltip as a guide to where to click on the time bar, double-click on it. |
Select a time range | ALT+drag. See also Work with time ranges. |
After you have gained an overview of your data with the time bar, you can use the Data Health view, the Graph view and/or the Marker Editing and Labeling panels to further identify and correct any issues.
To enhance system performance when working on a scene that contains very large amounts of data, you may want to turn off the Issues map.
To turn the issues map on and off:
Click the Issues map button to the left of the time bar.
To set the issues map to show all issues or only those for a specific subject:
From the Current Subject list, select the required option.
Note that you can snap the 3D view to the selected subject by clicking the Snap 3D view button
to the right of the Current Subject list.Find issues using the Data Health view
The Data Health view enables you to view detailed information about the labeling and gaps in the markers in the current scene.
To open a Data Health view, in any view pane, click the View type button and then select Data Health.
By default, if a Backup clip exists, it is displayed, but if required, you can select a different clip by clicking the arrow on the Display Gaps button. (Clips contain keys for objects in your scene. The clip stores keys only; not markers, bones, or any other objects.)
To display the time ranges for the gaps in the markers on the selected clip as gray rectangles, ensure the Display Gaps button is selected.
To view unlabeled markers, from the Show current subjects' markers list, select Show Unlabeled.
To view the labeling for a different subject, from the Current Subject list at the top of the Shogun Post window, select the required name.
Use the marker list on the left of the Data Health view to find the required marker, then follow line along to see the gray rectangles that indicate gaps.
To select the range of a gap, in the Data Health view, double-click the relevant gray rectangle. The following image shows a Data Health view and 3D Scene view of a gap selected for a marker.
The Show all markers button shows every marker in your scene. Select this option only if you need to see all markers because if, for example, someone is standing to one side but wearing a full marker set, you’ll see all their markers listed, as well as the markers for the rest of the volume. Normally, it is more useful to have the Show current subject's markers button selected.
Find issues using the Graph view
In the Graph view, you can see a graphical representation of the X, Y and Z positions, as well as gaps, for selected marker(s). This is useful during gap-filling, as it gives you another way of checking that the gap-filling looks reasonable for the data.
To open a Graph view, in a view pane, click the View type button and then select Graph.
To select and fill gaps in a marker:
-
To keep the Data Health view open as well as a 3D Scene view, split the screen by clicking one of the three-way split buttons at the top right of the window, such as the Three Views Split Left button
. -
As in the Data Health view, if there are multiple clips in your scene, you can select the required clip by clicking the arrow on the Display Gaps on Specified Clip button
. To display X, Y and Z values of the selected marker at the current position of the time indicator in the graph, click the Show Display Channels button . - In the 3D Scene view, select the marker. In the Data Health view, double-click a gap to select its range.
The Graph view now displays all the keys for the selected marker.As in the Data Health view, any gaps are clearly visible.
-
Zoom in by right-click and dragging for a clearer view or by using one of the options under the Zoom button
. - You can now use the Labeling panel or one of the options from the extended context menu (press CTRL+ALT and right-click in the Graph view pane) as required.
For further tips on using the Graph view to identify and fix data issues, see Fix common labeling issues.
Work with time ranges
While identifying and fixing issues in your captured data, you will often need to select a particular time range on which to work.
To select and clear time ranges:
On the main time bar, and in the Data Health and Graph views, select a range (ALT+drag).Green and red triangles that indicate the start and end of a selected range are displayed.
- To adjust the start or end of the time range, on the time bar, or in the Data Health view or Graph, hover the mouse pointer over the required arrow and drag it.
Tip: You can also update the range handles using hot keys:
,
sets the start handle.
sets the end handle
To change these hot keys, on the General menu, click Preferences and then click the Hot Keys tab. - To clear a selected range, ALT+double-click within the range.