Forum

Author Topic: Weak objects texture differentiation.  (Read 6799 times)

adrian2608

  • Newbie
  • *
  • Posts: 9
    • View Profile
Weak objects texture differentiation.
« on: June 01, 2017, 06:58:25 PM »
Hi there,

I have a question about shooting objects which don't have that good texture differentiation. Like for example an orange, lemon, etc.
Basically I'm placing markers, then merge chunks. then export chunks to MeshLab, merge them better than Photoscan do, and reimport mesh to merged chunk. Everything good, working.

But is there any other hack to get it from one chunk? Generally I'm shooting on a tripod, where I just change the height and angle. Object is on a turntable board whit 10 degree increment.
Yesterday I get good scan from bottom 3/4 and front angles, but the same from top and 3/4 from top doesn't match the front one. Which I really don't know why.

Have anyone their own solutions to that type of scenario and can share? Other objects are perfect, it goes only about object where the texture is very similar. Like orange, lemon, tomato.
And another question. Is the photos order important?

For example from top:
So- TOP, 3/4 TOP, FRONT, 3/4 BOTTOM, BOTTOM. Or not?

Regards,
Kamil.

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 15710
    • View Profile
Re: Weak objects texture differentiation.
« Reply #1 on: June 01, 2017, 07:17:24 PM »
Hello Kamil,

You can check the suggested scanning way on the following page:
https://blog.sketchfab.com/monthly-3d-scanning-challenge-horn-plenty/

In case you are taking several subsets (top and bottom, for example), you need to apply masks to the background, in order to process all images in the same chunk. Also good lighting conditions would be helpful to prevent hard shadows and difference between the image sub-sets.
Best regards,
Alexey Pasumansky,
Agisoft LLC

adrian2608

  • Newbie
  • *
  • Posts: 9
    • View Profile
Re: Weak objects texture differentiation.
« Reply #2 on: June 01, 2017, 07:31:22 PM »
Hi Alexey,

Yes, I have good lighting conditions, and object is masked with "from background" option, then manually fixed with Photoscan tools.

Will test one more time with better angles, and post the results here to further investigation.

Kamil.

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 15710
    • View Profile
Re: Weak objects texture differentiation.
« Reply #3 on: June 01, 2017, 07:32:30 PM »
Hello Kamil,

If you still have issues with the reconstruction, you can share the dataset and project file with us. We'll check if anything can be improved.
Best regards,
Alexey Pasumansky,
Agisoft LLC

adrian2608

  • Newbie
  • *
  • Posts: 9
    • View Profile
Re: Weak objects texture differentiation.
« Reply #4 on: June 11, 2017, 03:00:37 PM »
I run to another situation, and need to know why Photoscan can't process all of the photos in one chunk.

Here are 3 image from top angle, front and bottom angle. Why bottom angle isn't take into consideration with photoscan?
Is that maybe because of the leaves? Which aren't well shown?

(images for scan are bigger in px size)