Merging 2 Scans together

Hey Guys 

I am new to Reality Capture and come from other software. Other software has a feature to merge two chunks together. I use that quite often in Other software and I wanted to know if this is possible in Reality Capture too? I experimented a bit with the options but couldn’t find the right way yet. 

Detail of my Scan. 

It is a turntable scan of a scan as you can see on the pictures. I did a capture of the top surface and the bottom surface. In Other software I can align both scans in different chunks and then merge these chunks together the results look like this in Other software:

Top

Bottem

I tried aligning both sets of photos in reality capture but I expected results like this:

 

Reality capture used the control field as reference for the alignment so this happened. 

Would be nice if anyone could describe me a way how to combine the scans. 

Thanks for your help 

I guess one way of dealing with the problem is to mask the stone completely and aligning the pictures normally. But I was curious if reality capture has the feature merging 2 scans together. I always find black masking very time consuming since I use photoshop and manually creating 160 black masks takes very long. What do you use to automatically create black masks for a large number of photos?  

Hi Leon,

if you search a bit here in the forum, you will find some threads where people suggest to use unicolor backgrounds to make it easier for RC. Also, it has been recommended to scan from more than 2 sides - I guess then RC would rather latch onto the stone than the grid. Furthermore, you could use a moderate amount of Control Points.

Also you might want to look in the F A Q (hint-hint) https://support.capturingreality.com/hc/en-us/articles/115001485012-How-to-put-together-more-components-  It doesn’t cover everything but it’s a starting point.

I agree with you that masking is too laborious to make it worthwhile.

What Gotz said! 

Try it without the color grid.  For turn table I find it works best with a featureless background and matching color surface on the turn table. You should be able to shoot a full turn sequence and then rotate the object for another turn sequence to capture the underside.   I usually do three sequences.  One original orientation, one rotated 90, and one rotated 180.  Trick is to make RC think the camera is moving when in reality only the object is moving.  That will be hard to accomplish with it sitting on a grid.  If you get it right you should get near 100% alignment every time with very few stray points to clean up.

Might also try adding a couple more light sources positioned lower to kill the shadows underneath.

Hi all, are there any news about this ticket?

 

 

thank you,

Sure you already know but in recents versions you can make the mask automatic with RC

https://support.capturingreality.com/hc/en-us/articles/360018481659-Scan-Objects-Using-Masks-Image-Masking