Top row not taken into account

Hi there!

I’ve been making quite some tests with a turntable set-up. And the same scenario happens to me quite a lot. I’m taking about 60 pictures of my subject in three different view (bottom/front/top) so a total of about 180 images, which seems to give me good alignment. The front and bottom pictures are well registered and aligned, but the top row is not. I understand this is due to an angle difference being to high, but I wonder if there is a way to add these unregistered pictures to the mix.

Thanks a lot,

Jb

What are we seeing in the bottom half of the top shots? I can’t see that in other shots. The material of the object seems to be difficult and if RC finds more tie points on this weird background, then it might be mislead somehow. Do you get a component with the missing top shots or are they just left away entirely by RC?

hello, have you tried using control points?

@Gotz

Hello Gotz and thank you for your reply. What we see in the bottom half of the top shots is the background I didn’t clean properly, as it was a quick test. I did get a component with the top shots, but the camera is being placed at the same place for all the shots by RC (probably because of this problem). As I understand it, the key for good turntable is first to make sure your object is properly masked out. I’ll carry on with tests to find the easiest way of masking the background, removing it manually for all shots is tedious.

 

@lubenko

Hello and thanks also, this was my 1st idea. I hoped that setting control points would make RC prioritise the subject over the background, but it seems that even with control points it’s not the case. These are my 1st tests… I’ve got a question about control points for my next test: should I place them on all the top row pictures, or should I just concentrate on the 1st few images of the top row and than RC will deduct the rest of the row automatically, (after understanding the link between the front and the top row)? I’ve read that placing at least 4 point is a good start, would you have other advices about this?

 

I will make sure we have immaculate background in any case, and hope for the best.

For turntable shots the background must rotate with the object or you will get exactly what you saw. All photos in one position.

I use a lightbox with white background and a white turntable to eliminate the background as much as possible.

What remains in the photos is not defined enough for RC to pick it up as a feature (see image below).

Excellent capture, thanks for the input.


We are now getting much better results. The problem was indeed the background not being even enough. Not the camera angle. Thank you!

It would be fantastic to get a “turntable mode” though.

Where RC analyses the fix pixels in a series of pictures, and strips then out considering them as background.

Alignment settings => Image Overlap high

That will force RC to only use the center part of the input images for its feature detection.

That is basically “turntable mode” :slight_smile:

Excellent, thanks so much.

Thanks for the feedback, jb!

I think the idea with the filtering of static pixels is quite excellent though.

You could post a feature request for that…

Done!