Closed jrjdavidson closed 3 years ago
Hi, it look really strange. It looks like some problem with conversion. Would it be possible to send me the source SLPK file to investigate?
I sent a link to your @melowntech.com address.
Vaclav, just generated another dataset with a similar problem, I wasn't expecting this as i thought it was a relatively simpler geometry.I had also generated a quick test case earlier and had no issues with it.
here is the problem dataset: https://vtsbackend.canterbury.ac.nz/store/resources/tilesets/kakanuiDroneGCP here is the test without problems: https://vtsbackend.canterbury.ac.nz/store/resources/tilesets/kakanuiDroneTest
It appears the problem crops up when there is an overhang? again, happy to send you the original slpk files
Have you been able to re-produce the image with the other dataset I sent you?
Let me know if there might be another way, i.e. exporting and converting in another format rather than slpk? metashape can export in various formats, see screenshot below
Sorry, unfortunately, I had not found time to investigate yet. Hopefully next week.
Just an update on this- it appears that the issue is caused by how the data is created in metashape. The problem occurs in vts if I generate the tiled model straight from depth maps in metashape. If i create a dense point cloud and then generated the tiled model, it works fine. cheers
Pls, what OS (distro) and what vts-tools
version did you use? I tried to convert it myself and there's no broken mesh.
Ubuntu 18.04.5 LTS slpk2vts 1.89 (built on Jan 29 2020 16:32:20 at bionic)
Tested. Freshly built tag for 1.89 generates broken meshes. Current version is OK. IIRC there were some fixes with properly splitting meshes with too much faces/vertices but I cannot find it in the log.
Anyway, I've made a new vts-tools
package for bionic, v. 1.93. It's not from the HEAD because there were some very recent updates in metatiles that I do not want to publish into general public yet.
Finally was able to test and it works fine now! thanks heaps.
Hello,
I created a highly detailed true 3d dataset from drone and handheld photos using photogrammetry ( sfm in Agisoft metashape). It was a pretty extreme model to start off with given the technical challenges, but it worked quite well so I thought I'd try it in vts. I exported as a slpk and converted using
slpk2vts
The result is quite good except for the odd behaviour in parts. I attached an image, and here is a live link so you can explore: https://vtsbackend.canterbury.ac.nz/store/resources/tilesets/caveRock/?pos=obj,172.759753,-43.566130,fix,12.29,176.92,-28.76,0.00,20.00,45.00
Happy to share the data if you're curious