Problems with CCTags and repeating textures #2677
Replies: 1 comment 6 replies
-
|
Try the Improve CCTAG detection pipeline modification as described here https://github.com/alicevision/Meshroom/wiki/CCTAG-scaling (scroll down). The image of the marker should be roughly no less than 30 pixel of radius for the external ring. The size of the actual marker to print can be computed considering the distance of the camera w.r.t the marker, the focal length and the resolution of the image. |
Beta Was this translation helpful? Give feedback.
-
|
Can you share the settings of the sfmtransform node? |
Beta Was this translation helpful? Give feedback.
-
|
|
Beta Was this translation helpful? Give feedback.
-
|
i've also tried the Apriltag tag16h5 option leading to the same error sometimes. |
Beta Was this translation helpful? Give feedback.
-
|
Ok, could you try scaling with two markers? You can also set the Verbose level to debug to get more details on what is going on. |
Beta Was this translation helpful? Give feedback.
-
|
this is what i get with debug: i think i didnt found the tag with ID 0. I could scale it with two markers, but i want to have a stable pipeline which works with 3 markers, cause i want it to run on a server per commandline and always use one pipeline. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hello,
I have the following problems. I want to use Meshroom for a project in which I scan larger things (e.g. streets/houses) in an app, maybe even with a drone. I have made a setup where I have 3 CCTags on DIN4 pages which I lay out at a certain distance (just next to each other). I want to achieve the correct scaling and alignment with the tags so that I can then measure distances accurately in the app.
My first problem is that the tags are not always recognized, e.g. when flying a drone (
[error] Alignment from markers: Invalid number of image describer types: 0). Is there a certain distance that the camera must be from the tags?The next problem is that when I scan e.g. a street (only the street without surroundings), some images are discarded because not enough feature points were found, perhaps because the street texture is too similar and the images look too similar. Is there anything that can be done to the pipeline so that it is better recognized?
I use the following pipeline: Default pipeline with CCTag3 active in the necessary nodes. Additionally after the SfM node two SfMTransform nodes: 1. from_markers for scale, rotation, translation; 2. auto_from_landmarks with rotation.
Beta Was this translation helpful? Give feedback.
All reactions