Lee D, post: 333937, member: 7971 wrote: I'm not completely sold on it, that's why I was hoping to hear about some real-world experiences from people using it for production. But I think that once quality point clouds can be produced using photogrammetry with similar accuracy to a scanner, expensive scanners may go the way of the dinosaur other than for certain specific applications.
Interesting observation. It's very possible that I'll retire before I need to add either photogrammetry or scanning to my tool array, but I'll continue to follow these developments with interest.
Yes, we use realworks. It has come a long way since we first started using it. You don't feel so bad about paying for new releases (via the support agreement) when they keep improving it and coming out with new features. Our partner company (photogrammetric) uses topodot to extract information that they can then use.
But, I am not the person who does any of that processing, I am pretty much a novice at that. Todd, who works for me, has been using Realworks since 2008. We are doing a presentation next week to a DOT, so he is out scanning a nearby bridge right now to get some data to show them. We already did a nearby intersection as well, and also did it with the V10.
We have been using the GX until now, but of course it was VERY slow. But, I did like the survey workflow. A lot of the projects we plan on using the TX8 on have plenty of control points, so we want to be able to give it coordinates for the scanner station and the targets rather than just try to resect.
beuckie, lee, john, and jim,
now this is ripening into a nice dialog! i look forward to gavin schrock weighing in on some of these elements.
a couple more thoughts:
yes, i did pay a price and a learning curve on the 80% lap on photos as well as the need for the photos to be in some 'linear fashion'. you need to wrap your head around that right away or most CRTP software will give you troubles
in defense of CRTP's accuracy and precision as a whole: (paraphrased) several journals have been written by what appear to be heavy hitters. said heavy hitters often made the point that relative accuracy from CRTP is much more defensible, as the comparisons are made from one pixel to another in the same image, compared to 3D scanning, where each scanned measurement is 'stand alone'. i will try to look these up while the iron is still hot, so stand by
I'm 28 minutes into watching the video Jim Frame linked. My first impression is that some of those business card sized target stickers placed at strategic location around the building walls would enhance the work flow dramatically. And maybe a few lath driven into the ground here and there, to provide known locations in good geometry.
Moe Shetty, post: 333951, member: 138 wrote: where each scanned measurement is 'stand alone'
I think that statement exaggerates the situation. The angular and distance errors for each scanner measurement are quantifiable, and errors resulting from scanner settlement would normally be minor and in any case only of concern between points separated by a long time span. Scanned positions aren't like RTK shots, in which the two nearby positions separated by only a short span of time can have a lot of relative error due to the very large ranges used in deriving each position solution. I would expect temporally-close scanned positions to correlate very well.
OK i agree, in general, with you jim. not trying to make a fight with you or split millimeters, simply backing a new product's viability and utility
any point in a scanned image is affected by the systematic error +/- environmental and condition errors, every error computed between the instrument and the surface measured. errors are computed similar to IR or reflectorless
any point/pixel in a CRTP image is correlated to its adjacent pixels (relative accuracy)
let's think about this in a scenario of a factory or petroleum plant, some place where accuracy and precision needs to be high, and the place looks like the intro to the monty python show. pipes growing out of everywhere, and say you can't use a scanner because of traffic or vibrations. CRTP can be
CRTP seems a viable alternate to me.
CRTP
overall cost +
faster in the field +
processing labor -
risk of faulty data collection could mean subsequent field visit -
SCANNING
overall cost -
slower in the field -
processing labor +
risk of faulty data collection could mean subsequent field visit -
what other factors am i missing?
After watching the video Jim posted I would have to agree that the particular kind of work demonstrated would have been better completed by simply getting reflectorless shots. But I can see loads of applications where Datugram would be very useful. I see it as a way to fill in details in design topos and ALTAs.
Can't figure why they used photos taken in such poor lighting for their demo.
[QUOTE="
let's think about this in a scenario of a factory or petroleum plant, some place where accuracy and precision needs to be high, and the place looks like the intro to the monty python show. pipes growing out of everywhere, and say you can't use a scanner because of traffic or vibrations.
Mobile mapper - interior and or exterior. Get a good IMU or utilize SLAM.
Moe Shetty, post: 333964, member: 138 wrote: not trying to make a fight with you or split millimeters, simply backing a new product's viability and utility
Understood. I'm looking at this entirely from the perspective of my particular operation, and am not suggesting that others wouldn't be able to make good use of it.
I've looked at the video. The example off the chruch can be done with Kubit Photoplan. It is a bit cheaper than datagram software.
http://faro-3d-software.com/CAD/Products/PhoToPlan_Camera/CAD_photogrammetry.php
Maybe to some interest for some of you guys.
Lee D, post: 333937, member: 7971 wrote: Jim we did a two week trial of Datugram...
I'm not completely sold on it, that's why I was hoping to hear about some real-world experiences from people using it for production. But I think that once quality point clouds can be produced using photogrammetry with similar accuracy to a scanner, expensive scanners may go the way of the dinosaur other than for certain specific applications.
I have also demo'd a product called PhotoModeler, and am looking into a product called iWitness. These are MUCH cheaper than Datumate. Also, many other products do not care about ground control. They can process the images without scale or orientation.
As I have dived into this, I am starting to feel that there are aspects not unlike handing a surveyor a GPS receiver and telling them to go and survey. There are potential pitfalls. However, my initial work has verified the accuracy claims of Datumate, and I have found this to be the case with all the products I have used.
Datumate seems to be specifically geared to Surveyors. This is a good thing. My impression is that it is not mature when compared to other products that are available. However, they seem to be adding functionality at a fairly good pace.
I do not believe that the ACAD 360 is currently a good solution for us. There are many cloud options out there, but they lack certain functionality. Firstly, they generally create a point cloud or a mesh. This is fine if you are looking to replace a laser scanner, but most of us need planametric features. We need points and linework.
We are moving to a 3D world, but are not there yet. Datugram, in my opinion, is attempting to essentially allow surveyors to use the 3D world in their work.
-David
Lee D, post: 332853, member: 7971 wrote: Is anyone using this? I'd be interested to hear about your experience with it, I'm evaluating it.
FYI,
Right now, I am a recommend buy on software of this sort. This software has been around for a long time, Topcon Imagemaster has been doing this for decades. And...there are third party non-surveyors using this type of imagery right now, creating maps left and right and making money at it.
Lastly, this will not reach the full potential until it is combined with UAV's. Until the FAA lets us fly...
On laser scanning, no one I know of scans an object with a single scan. The typical scenario involves multiple overlapping scans so there is redundancy on the scanned item, whatever it may be.
Registration is pretty easy at this point especially with targets on known control. Cyclone has greatly improved the cloud-to-cloud registration routine too so that it is much faster to do. If two clouds are registered together using cloud-to-cloud registration the software has indicators of quality.
One advantage of a photogrammetric method or a laser scanner method is that it captures things you may not know you need with respect to civil or criminal evidence mapping which is different from the typical engineering surveying project. The Investigator often hasn't interviewed all of the witnesses when we are out there (and they want us in and out of there pronto for various reasons) so they don't necessarily know that a particular item is important. If we take the prism pole around and map things they have marked there could be things missed because they didn't know to mark them. We still use a total station or RTK for certain things because they are hard to see or find in the scans such as evidence hats or pin flags.
