-
|
Hi, I try to calculate camera rig, but with no susses. I cannot find some description how to start it. and tried to: I have made sfm file with StructurFromMotion (and changed path for V1): And started it with (tried it with gui to): C:\Users\Xma1\Desktop\Meshroom-2021.1.0\aliceVision\bin\aliceVision_rigCalibration.exe --sfmdata "C:/Users/Xma1/Desktop/RIGCHECK/cameras.sfm" --mediapath "C:/Users/Xma1/Desktop/RIGCHECK/rig2" --cameraIntrinsics "" --export "trackedcameras.abc" --descriptorPath "C:/Users/Xma1/Desktop/RIGCHECK/MeshroomCache/FeatureExtraction/54f7cb26c30018d5c72e877dd2e64093752516a7" --matchDescTypes sift --preset normal --resectionEstimator acransac --matchingEstimator acransac --refineIntrinsics "" --reprojectionError 4.0 --maxInputFrames 0 --voctree "C:\Users\Xma1\Desktop\Meshroom-2021.1.0\aliceVision\share\aliceVision\vlfeat_K80L3.SIFT.tree" --voctreeWeights "" --algorithm AllResults --nbImageMatch 4 --maxResults 10 --matchingError 4.0 --nNearestKeyFrames 5 --outfile "C:/Users/Xma1/Desktop/RIGCHECK/MeshroomCache/CameraRigCalibration/b1a2e820962e78196cda5068fe72cd3b64b0502e/cameraRigCalibration.rigCal" C:\Users\Xma1\Desktop\Meshroom-2021.1.0\aliceVision\bin\aliceVision_rigCalibration.exe --sfmdata "C:/Users/Xma1/Desktop/RIGCHECK/MeshroomCache/StructureFromMotion/6690c080eef5b22e67b337953828c95ce98fa92b/cameras.sfm" --mediapath "C:/Users/Xma1/Desktop/RIGCHECK/rig2" --cameraIntrinsics "" --export "trackedcameras.abc" --descriptorPath "C:/Users/Xma1/Desktop/RIGCHECK/MeshroomCache/FeatureExtraction/54f7cb26c30018d5c72e877dd2e64093752516a7" --matchDescTypes sift --preset normal --resectionEstimator acransac --matchingEstimator acransac --refineIntrinsics "" --reprojectionError 4.0 --maxInputFrames 0 --voctree "C:\Users\Xma1\Desktop\Meshroom-2021.1.0\aliceVision\share\aliceVision\vlfeat_K80L3.SIFT.tree" --voctreeWeights "" --algorithm AllResults --nbImageMatch 4 --maxResults 10 --matchingError 4.0 --nNearestKeyFrames 5 --outfile "C:/Users/Xma1/Desktop/RIGCHECK/MeshroomCache/CameraRigCalibration/b1a2e820962e78196cda5068fe72cd3b64b0502e/cameraRigCalibration.rigCal" but I get:
Calibrating sequence using the voctree localizer
[2021-04-16 15:02:15.071640] [0x00002984] [debug] folder without expression: C:/Users/Xma1/Desktop/RIGCHECK/rig2 Processed 0 images for camera 0 Camera 0 localized 0/0 C:\Users\Xma1\Desktop\Meshroom-2021.1.0\aliceVision\bin\aliceVision_rigCalibration.exe --sfmdata "C:/Users/Xma1/Desktop/RIGCHECK/cameras.sfm" --mediapath "C:/Users/Xma1/Desktop/RIGCHECK/rig2" --cameraIntrinsics "" --export "trackedcameras.abc" --descriptorPath "C:/Users/Xma1/Desktop/RIGCHECK/MeshroomCache/FeatureExtraction/54f7cb26c30018d5c72e877dd2e64093752516a7" --matchDescTypes sift --preset normal --resectionEstimator acransac --matchingEstimator acransac --refineIntrinsics "" --reprojectionError 4.0 --maxInputFrames 0 --voctree "C:\Users\Xma1\Desktop\Meshroom-2021.1.0\aliceVision\share\aliceVision\vlfeat_K80L3.SIFT.tree" --voctreeWeights "" --algorithm AllResults --nbImageMatch 4 --maxResults 10 --matchingError 4.0 --nNearestKeyFrames 5 --outfile "C:/Users/Xma1/Desktop/RIGCHECK/MeshroomCache/CameraRigCalibration/b1a2e820962e78196cda5068fe72cd3b64b0502e/cameraRigCalibration.rigCal"
Calibrating sequence using the voctree localizer
[2021-04-16 16:35:46.398371] [0x0000620c] [debug] folder without expression: C:/Users/Xma1/Desktop/RIGCHECK/rig2 Processed 0 images for camera 0 Camera 0 localized 0/0 |
Beta Was this translation helpful? Give feedback.
Replies: 5 comments 5 replies
-
At the moment the Rig support is highly sensitive to configuration / metadata issues. So it is possible the rig is not defined properly due to the missing metadata. |
Beta Was this translation helpful? Give feedback.
-
|
thank for answer, but injecting metadata and converting into jpg haven't worked. I have added: do I need to add extra camera Intrinsics? |
Beta Was this translation helpful? Give feedback.
-
Your camera is listed in the sensordb:
No, we don´t need a serial number for each image, but for each camera:
(Side note: having the same serial number for 101.png from camera 1 and 101.png from camera 2 would result in only one image being imported, as Meshroom would assume it to be the same image, since the timestamp and filename is identical. Error message at the end of the page -> https://github.com/alicevision/meshroom/wiki/Multi-Camera-Rig) When you drop the rig sub-folders into Meshroom, the rig constraint will be recognized and you will be able to see it in the CameraInit node (see Rig and Rig Sub-Pose of the Viewpoints parameter). Check if that is the case. |
Beta Was this translation helpful? Give feedback.
-
|
I will try to add timestamp to it, maby it will help somehow. Sorry for confusing it with the camera, I chose Canon because my camera (from Basler) is not in the list and I don't know the camera name, just the manufacturer. I'm now not near to the camera and cannot check parameter and name of it. Maybe I'll build small construction with ESP32 that I can take with me everywhere. I tried both options by (1) giving each camera the serial number and (2) each picture the serial number. Images have been loaded into Meshroom, that's no problem. It runs to StructurFromMotion (Dense and other steps was removed), but the cameras are not tied to a rig, and when I use CameraRigCalibration, it throws an error. names are normaly 1-9 (with leading zeros if more digits). „10x“ is only to separate names in other folders (cameraID0ImageID) without use serial number or to code camera and imageid into serial number. rig/ # "rig" folder If i use 101 and 201 names i can see rig and rig sub-pose to (and i can change to matching rig id) and in Structure from Motion is some checkbox with rig. I had tried it before I ascend in the forum, but I don't see any difference whether it is with the rig checkbox or without in this case. The cameras are not connected and have different positions to each other. I can not start CameraRigCalibration from Utils/CameraRigCalibration. Maybe I misunderstood something and the camera calibration is not there to improve the camera position in the camera rig housing (so that one camera improves the position of another camera to avoid a local minimum. So that the cameras have the same constant position to each other and only move in groups.) |
Beta Was this translation helpful? Give feedback.
-
|
Please try the default graph first and make sure it works before adding other nodes. CameraInit should detect the rig correctly and the SfM should compute without errors. You could share a portion of the (raw, unmodified) dataset if you want and I can take a look. CameraRigCalibration is optional - details: https://meshroom-manual.readthedocs.io/en/latest/feature-documentation/nodes/CameraRigCalibration.html |
Beta Was this translation helpful? Give feedback.
-
|
here are images and Blender file where i put cameras to first camera and moved first camera to Zero. After Metadata changes much better, but there is still difference. If i fill one camera with black color (to emit that camera have no points), camera is removed (on rig we can get position of that camera with help of another). https://www.dropbox.com/t/SBV7ZmZf8FE9Onci and where is example for /path/to/calib1.txt for CameraRigCalibration to get relative poses between cameras of the rig and choose the more stable value across the images? |
Beta Was this translation helpful? Give feedback.
-
|
I´ll take a look.
You can calibrate each camera with https://meshroom-manual.readthedocs.io/en/latest/feature-documentation/nodes/CameraCalibration.html This is how the correctly set up rig should look like in camerainit. If rig pose and/or sub pose is -1 something went wrong for two cameras there should also be only two intrinsics: (ignore the 2019.1 name in the path, it´s just where I put the image files)
Your dataset has 8 intrinsics declared, only two are expected:
Checking the metadata I noticed your PixelXDimension is changing between images! Images should always have the same,fixed dimensions.
The BodySerialNumber should be different for each camera but stay the same for images taken from the same camera. each of the two cameras chould have a unique serialnumber. The changing PixelXDimension results in Meshroom assuming you have 8 different cameras, as changing the x dimension is not expected for the same camera. Never modify (crop/distort) images prior import for photogrammetry use. |
Beta Was this translation helpful? Give feedback.
-
|
thanks for the reply i did not notice that the size changes with the standard parameters of the camera. I don't crop or resize image. I will set the capture script to the manual size parameter next week. And sorry for BodySerialNumber, I tried 1 for camera 1 and 2 for camera 2 first and then (because it doesn't worked) changed it to 101 and forgot to change it again. |
Beta Was this translation helpful? Give feedback.
-
it´s fine 1 and 2 were only placeholders. With the dataset you provided it changed within a camera dataset where it is expected to be the same for all files |
Beta Was this translation helpful? Give feedback.
-
|
orks, I need to set a rig ID for all rigs (I thought it is for each position). Now everything is combined into one rig. Many thanks for the help. |
Beta Was this translation helpful? Give feedback.
I´ll take a look.
You can calibrate each camera with https://meshroom-manual.readthedocs.io/en/latest/feature-documentation/nodes/CameraCalibration.html
As you don´t have access to your camera you can not use camera calibration, as it requires images/videos of a calibration pattern.
This is how the correctly set up rig should look like in camerainit. If rig pose and/or sub pose is -1 something went wrong for two cameras there should also be only two intrinsics:
(ignore the 2019.1 name in the path, it´s just where I put…