Resources for Understanding Photogrammetry in 3D Scanning

[[Mi...]]

I am looking to understand the science/technical-workings behind photogrammetry a little better. In particular, as it relates to use in a Scan Box and the Zeiss/GOM Inspect software.

I came across this video earlier today and learned somethings I did not know, but it only scratches at the surface of the concept of photogrammetry and seems to be more orientated towards the application of photogrammetry in a field like photography: https://www.youtube.com/watch?v=rh6xZ5sb31M 

I have a few questions:

• Does any one have any useful links / videos / resources on photogrammetry for learning purposes?
• Is it correct to say that coded reference markers are required to run photogrammetry in the Zeiss Inspect software? Otherwise, the uncoded reference markers cannot be located (in photogrammetry), and you would instead need to run a regular ATOS measurement series with uncoded reference markers placed on your part?
• How much 'accuracy' is lost by not using coded refence markers and only using uncoded reference markers?

[[Ja...]]

Hi mike to answer part of this , yes coded markers are required for photogrammetry to function in gom/zeiss solutions.

It is hard to articulate the accuracy beneift it brings but it provides a more stable generalised base accuracy to around 10/15 microns per meter .

If just using atos it is a general rule/advisable to always use photog when the part is over twice the measurement volume of the atos system.

[[Mi...]]

Please sign in to view this username.

Thank you for the response, James. I came across another post from about a year ago and to my understanding Photogrammetry is more stable because you don't need overlapping images (which can accumulate small bits of error with every image).

[[Ja...]]

The way i used to explain it was to imagine a long table and start scanning at one end , overlapping as you say increments the error to a point where in the end the data from the table will be bent /warped .  Hence why the general rule of thumb on using it as the use of phog g eliminates this stacking issue .I

The other minor benefit of this is needing less markers overall.

My other advice was if you have access to use photog just use it ! It doesnt take long for the benefits it brings.

May i ask what your application is or are you just curious how it works?

[[Mi...]]

Please sign in to view this username.

Sorry for my slow response. I got busy with some work tasks and life stuff. I've heard your analogy in a class that I took previously and it makes sense. It's kind like tolerance stacking.

Can you speak at all to the longevity of reference markers (i.e.: how long they last)? I see another minor benefit in that with photogrammetry, you're not stickering every part that comes through the door. You just need to sticker your fixturing once.

Aside from needing to replace damaged reference markers as they are found, is there a "prescribed" timeline for how long you can go before needing to re-sticker an entire fixture? I figure you'll know if when looking at your photogrammetry results, you notice that you have much less coded and uncoded reference markers than expected.

I work in aerospace and usually on airfoils, but I am only asking questions on Photogrammetry to improve my knowledge. I like being to confidently understand a topic and to be able to explain it well to others.

 

[[Aa...]]

We perform Photogrammetry on our cage's once a year. We are also in Aerospace.

[[Ja...]]

Hi , i lost my login so sorry for the delay in responding.

There i no prescribed timeline for replacement of points. You are absolutely correct that having points on the fixture and not on the part is a major bonus and critical to proper automation solutions.

You could consider making recesses in fixtures to place points which will reduce their likelyhood of damage , also there are (expensive) ceramic markers that could be used that are more resiliant , but best utilised when recessed to avoid damage.

As you say you can sense check the total number of points captured over time and evaluate whether points need replacing, in the end if it gets too many the bundle doesnt work, or poor result or your atos images will not be orientated .  Its been a while but i think it is possible to set up automated checks around such kpvs that can flag such potential problems.

If you have a photogrammetry system id use it as often as you can, it doesnt take long to do, particularly automated .  If automated the frequency of measurement depends on your frame stability and setup, which you can characterise.  In some applications it can make sense to complete before every run of digitising . However,   It could be once a day, once a batch , but once a year is far too long a gap in my opinion/experience particularly for higher end applications like in aerospace .

Please sign in to view this username.

 has written some nice tools for understanding fixture stability and has many years of experience of setting up high end aerospace setups , some of which with me !

I hope this is helpful

cheers

James

[[Mi...]]

Please sign in to view this username.

No problem on the delay! Most of our fixtures have actually been designed with the recesses that you mention, and we run photogrammetry with every production part. We do this due to the setup of our scanbox/fixtures, and because the actual scan series that capture the part tend to be much, much, longer than the ~five minutes takes to run photogrammetry.

I like the idea of setting up automated checks that will flag potential problems. What exactly is kpvs? I don't recognize the acronym. Is it something related to python scripts?

I appreciate all the help/advice you offered and references to other work. It's taken a while, but with the more I dig into these things I think I'm slowly beginning to under the capability and functionality of the software.

[[Ja...]]

Please sign in to view this username.

and

No problem at all.  If i was taking a guess your application it sounds in the direction of blisk measurement which if so is important to get an understanding of the influencing factors!

Kpvs = key process variables.    In this case it could be number of points found , bundle adjustment /alignment values etc.

However im afraid its been a bit too long to remember exactly how or if there are tools available for kiosk mode to handle this without you writing yourself.

Im hoping Chris sees this as hes absolutely the right person who can help with the detail!