[multiview] camera tracking pipeline: update edges

- DistortionCalibration in input of the FeatureMatching (instead of SfM)
- Add sfmDataFilter in input of ExportAnimatedCamera
This commit is contained in:
Fabien Castan 2021-06-01 16:32:23 +02:00
parent 14579f766d
commit 64d27158f5

View file

@ -493,8 +493,8 @@ def cameraTrackingPipeline(graph, sourceSfm=None):
distortionCalibrationT = graph.addNewNode('DistortionCalibration',
input=cameraInitT.output)
graph.removeEdge(structureFromMotionT.input)
graph.addEdge(distortionCalibrationT.outSfMData, structureFromMotionT.input)
graph.removeEdge(featureMatchingT.input)
graph.addEdge(distortionCalibrationT.outSfMData, featureMatchingT.input)
imageMatchingT.attribute("nbMatches").value = 5 # voctree nb matches
imageMatchingT.attribute("nbNeighbors").value = 10
@ -506,6 +506,8 @@ def cameraTrackingPipeline(graph, sourceSfm=None):
structureFromMotionT.attribute("minAngleForLandmark").value = 0.5
exportAnimatedCameraT = graph.addNewNode('ExportAnimatedCamera', input=structureFromMotionT.output)
if sourceSfm:
graph.addEdge(sourceSfm.output, exportAnimatedCameraT.sfmDataFilter)
# store current pipeline version in graph header
graph.header.update({'pipelineVersion': __version__})