When alignment fails, allow me to manually place the scan in Capture

When scanning larger spaces, the Matterport tends to get lost and puts scan points in different locations on the model. It would be great to draw a box to tell the camera where you have moved the camera so it knows to put the scan point in the right spot. It may also cut down on alignment errors since it would know what to compare the new scan to.



  • Hi @josh, do you mind if I change the title of this, to make it a bit more descriptive? I like the idea but I'm afraid it won't get as much attention. Maybe something like, "When alignment fails, allow me to manually place the scan in Capture" or something like that?

  • Yes. Feel free.

  • This would be a huge timesaver,

  • Or just simply be able to drag the errant scan position to the approximate correct location

  • Yes. Yes. Yes. In my short time scanning I have run into this on multiple occasions. I'm using scanning to record construction process. But because of the nature and condition of the working environment the camera can sometimes be confused. Especially when I have to backtrack to scan 1 and it's expecting something around scan 80. (at least the alignment process seems to have some bias regarding what scans it has recently processed).

    To be able to manually give the camera a little bit of help would be excellent.

  • This could also be accomplished by giving us the ability to choose which scans are most nearby "Alignment Error; Please re-scan or select 2 or more scans around the area in question." I had this exact same idea on my 2nd scan.

  • the first gen capture software required you to overlap with the previous scan or select the scan you were next to when you skipped around.

  • Guys, the Matterport Camera aligns by using the 3D Data. Manual connection would be so difficult to pull off and would likely result in corrupting models completely.

    Perhaps Matterport might be able to work on their algorithm a bit to include things like doors that move or allowing us to see where the scan failed to connect and allow a certain level of interaction -- possibly present us with a choice -- ignore this area, etc. This would be VERY hard to do, but would open up the possibility of reliably allowing us to open a door when close to it. (The other method is to scan up to the door, then open the door and scan from inside the door frame -- this -sometimes- works)

  • I had the same problem with all the models I took. I had to move from one area to another and then when I got back to the first area it took me a long time for the camera to recognize the location and sometimes I even had to walk a lot of steps back. I was very happy if I could scan the next point and manually set it where it is Is on the map in the app model

Please sign in to leave a comment.