Reuse of depth maps

Development Asset Creation
,
1

How exactly does the reuse of depth maps manifest itself?
Because I am rerunning reconstruction with idenical images and registration several times, but there is no gain whatsoever durcing depth map calculation. Does it have to bee in the same project and/or with identical combination of cameras?

2

Hi Götz
If you run alignment, then depth maps must be recomputed because ALIGNMENT changes camera positions, so new depth maps must be calculated. Mostly the cache is reused in the ALIGNMENT step, depth maps are reused if you have say multiple reconstruction regions ( whichever workflow reason for that )

3

I did NOT realign - same registration aka alignment!
With multiple you mean reconstruct different parts of the same component?
That’s the thing - its exactly what I did but it recomputes every time…

4

Hi Götz
Tested many times, the depth map calculation is working as designed. I can take a look directly at your PC to see what is happening there in your case.

5

Hi Milos,

maybe I expect too much - what will the difference be roughly?
Will it skip the depth map process entirely if they have been calculated before?
Because if so, then it definitely doesn’t work here.
If it is only a gain of lets say 25-50%, then it could be.
Will prepare a test scenario with only few images and get back to you.
Thanks so far!

6

Ok, NOW it worked and it was considerably faster!
Is it also dependent on the bounding box?

7

Hi Götz.

Yes, if you will change Reconstruction region, this also required depth maps re calculation.

8

Thanks mate!
That explains a lot! :slight_smile:

9

Hey Vladlen,

is there any way I can contact you other than PM here?
My outbox seems to be jammed but you could still try to get in touch by PMing me…
Thanks!

10

Sorry. Answer in PM.
Facebook or Twitter is faster way contact me.
Also several messengers but for friends/family only.

11

Al right, no problem.
I’m a dinosaur prefering email… :wink:
But you answered my question in the right thread already.
Thanks again!

12

I have this project with 6600+ inputs (close range photos + laser scans). At this point, reconstruction is in progress with image scale for depth map calc set to 1, and Default grouping factor set also to 1.

Depth map calculation is done, and now it is in stage of creating model (7250 parts :o),

If I stop reconstruction now, change grouping factor to 2, and re-run the reconstruction, will depth maps be re-used? Can I tweak some other setting to speed up process, and still have satisfactory model geometry?

Final result of reconstruction should be highres ortophoto…