Messages

16

General / Re: Group or Split Calibration for Alignment of Historic Aerial Photos?

« on: September 08, 2014, 09:09:12 PM »

Hi Steven,

I was wondering about the image format, because it is not a quadratic. Commonly most Aerial Mapping cameras come with a quadratic image format. A camera which has not, is the Fairchild A-18, it has 9 by 18 inch. If the zig-zag mark at the bottom is reallly a remain of a fiducial -  the shape really suggests this - it is astonishing how large the mark is, compared to the size of the whole foto. Is this really the complete original photo or was it scanned in parts and this is only such a part?

I think too, that spending too much effort into the calibration will not necessarily lead to better results, that is what I found while working with the above mentioned images too.

Good luck!
Cheers

17

General / Re: Group or Split Calibration for Alignment of Historic Aerial Photos?

« on: September 08, 2014, 06:56:08 PM »

Hi Steve,

do you have any information about  the camera the images where originally  taken with? Are these military  aerial recon images? You may find something useful here:

http://www.airrecce.co.uk/

I was recently mosaicking some scans of German Luftwaffe recon images from WW2, and was lucky that most of the necssary information in the fotoframe was still present. Only one fiducial was cut of. Image and scan quality was quite good. Main problem with that kind of images in most case is lack of sufficiant overlap.They normally have only standard stereo overlap.

What do you have in mind with the fotographs? I think mosaicking may be possiible, but DSM generation probably not.

Cheers

18

General / Re: IMPORT dense point cloud

« on: September 01, 2014, 03:29:50 PM »

Hi all,

yes I think "point cloud cleaning" is the main thing people have in mind, when exporting the point clouds.

My suggestion is:
Why not implement a couple of algorithm directly within Photoscan?

For example:
1. A conditional point cloud filter, so that we can filter it according to specific attribute constraints, e.g. normal in z-direction not greater than a specific value etc.

2. Statistical outlier filter

For more ideas on possible filtering you may like to look here:

http://docs.pointclouds.org/trunk/group__filters.html

That would save the export/import steps and can maintain the integrity of the pointcloud within Photoscan.

Cheers

19

General / Re: IMPORT dense point cloud

« on: August 29, 2014, 10:14:05 PM »

Hi  alll,

you are right when you say  that importing a point cloud would be  a good feature.

When looking at the python api docu for ps it becomes clear that withtin ps the point cloud is NOT just a simple collection of 3D points with RGB colors and point normals. The point cloud class is a bit more because additionally it  holds the projection information for the points, or in other words. where the 3D points can be found in the 2dimensional  image space of the corrosponding camera.

So when you simply export the point cloud points as x,y,z,nx,ny,nz,r,g,b you will loose this connection. Therefore reimporting a pointcloud that was exported without storing the "connection-information" is not a trivial task.

Alexey, please can you give a comment regarding my interpretatiion, may be I am wrong with this.


Chreers

20

General / Re: Administrator privileges are required to activate Agisoft PhotoScan (Ubuntu 14)

« on: August 23, 2014, 12:31:44 AM »

Hi,

I am runing PS on Ubuntu 12.04 without problems. There is a folder Agisoft in /etc/xdg wich should hold a file like PhotoScan Pro.conf, That is the licence File. It is created by the Installation procedure. I assume the only reason why Agisoft needs admin priveledges during installation is the fact that this file is stored in Linux system dir, where the standard users is not allowed to write or store files.

cheers

21

General / Re: CPU and GPU benchmarks

« on: July 17, 2014, 07:14:45 PM »

Hi all,

thanks for publishing the benchmark-tests.

Having the most recent up-to-date and powerful hardware will make you really happy when working with agisoft, but it may also strain your budget nastily.

For those who are looking for computing power but having only restricted budget, I would like to report  my experiences  with second hand / used  older generation hardware:

D.... T7500 Precision Workstation, Dual hexacore Xeaon X5680 with 3.3 GHz, 48 GB 1333 MHz Cl 9 RAM, Dual nVidia Quadro 4000, 2 GB RAM., manufactured in 2010

Pricing EU / Germany: 1200 - 1600 Euro

First test run with agisoft showed, that processor cooling was absolutely insufficient for running under full load for a longer time. Core temp higher than 80 - 90 deg Celsius was common, but which I think is not healthy over a longer period of time.

I modified the cooling system in he following way:
- removed all the plastic stuff from inside
- removed both processor heatsinks (mainboard and riser-card, which is for the second CPU)
- installed two Noctua NH-U9DX 1366 tower cooler with fans, which sustainly solved the overheating problem

Core temp under full load over longer period of time does now seldom reach more than 60 deg C.

Here are the benchmark results for the  4 stages, settings and images as described in the other posts above (4 of 24 virtual cores disabled), Hypershreading and all Cores activated in Bios

1. 205.3 sec
2. 161.8 sec
3. 231.5 sec
4. 109.9 sec

Total of 708 sec or 11.8 min, Performances for step 2:   CPU  181.631 million samples/sec, GPU1 249.646 million samples/sec, GPU2 254.201 samples/sec, total 684.86 .

The machine is running Ubuntu 12.04.4 LTS with recent updates installed.


Cheers

22

General / Re: Problem with georeference system in ArcGIS

« on: June 25, 2014, 11:44:45 PM »

Hi,

Did you generate your point cloud originally from GPS-tagged images (WGSS84 CRS and spheroid) and applied the Amersfort ground control points in Photoscan in a second step using the "convert coordinate  system" from  ground control pane?

As far as I interpret the 28992 EPSG, Amersfort is a projected coordinate system,  using bessel speroid.
In case your original point cloud was  in WGS84  and you would like to convert to Amersfort a geodetic datum transformation will be necessary in order to transform between the different  spheroids. The projection engine in ArcGIS should recognize that and offer a suitable transformation.

I am not sure if Photoscan is able to do a correct datum transformation for Amersfort  28992 in a similar way.

May be the alignment error is caused by such a kind of  problem.

cheers

23

General / Re: calibrate underwater photo

« on: March 11, 2014, 04:52:57 PM »

Hi,

unfortunately I have no experience with underwater photography and photogrammetry. The first thought that came into my mind after reading your post was: What about the diffraction index of water? May be you have to take this into account (somehow)? The diffraction index may vary depending on the percentage of minerals / salt diluted. Calibrating charts / marker panels may help with this.


Cheers

24

General / Re: Linux-issue - Photoscan Pro

« on: February 16, 2014, 07:49:41 PM »

Hi Jens,

our experiences are, that it works to run PS on CentOS, ScientificLinux  or Fedora, though we did not test the latest versions of PS.  Crucial thing is always the driver for the graphics card, when you want to use OpenCL for GPU computing.

When you go for a test, please report, what your findings are.

Cheers

25

General / Re: Short report on reconstruction with WW2 air-recce images

« on: January 15, 2014, 03:06:39 AM »

Hi all,
there is a small typo:

Cameras were Zeis RB75/30 and 50/30.

first number is focal length in cm(!), second is image format (30x30cm)

here is a link for an image of RB75/30:
http://commons.wikimedia.org/wiki/File:Zeiss_RB-75_30_aerial_camera_1.JPG

Cheers

26

General / Short report on reconstruction with WW2 air-recce images

« on: January 15, 2014, 02:50:08 AM »

Hi all,
for a history project we tested PS reconstruction capabilities with 2 sets of  WW2 airial reconnaissence images:

Set 1: British Royal  Airforce recon images from a small town in Germany, late summer 1943, camera unknown, geometry etc. were unknown to us, poorly scanned by the vendor, low resolution, partly cloud cover, a minimum set of only two images.

Set 2: A couple of images from two flights over Volgograd (aka Stalingrad), August and Oktober 1942, German Luftwaffe (Source is an archive in the US, please send PM if you are interested), Images were of good quality and hr-scans provided by the archive were really good, 3.7k by 3.7 k pixel, fiducials were visible, so we were able to reduce some image distortion introduced by the scanning process. Camera was Zeiss RB 70/30 and RB 50/30.

Results:

Set 1: Even with this poor set PS was able to do a mosaicking.
Please see first image. no acceptable dsm.

Set 2: Very good mosaicand a more or less acceptable dsm, generation of textured 3 model, based on an 41 Mio point cloud. Please see image 2.

PS did a great job for this test.

27

General / Re: Softwear Verkauf und Schulung in der Schweiz?

« on: January 07, 2014, 10:00:03 AM »

Hallo und guten Tag,

ich habe Ihnen eine PN geschickt.

Mit freundlichen Gr??en

photogrammetrix

28

General / Re: How does PS Pro handle geodetic datum transormations?

« on: September 16, 2013, 07:31:39 PM »

Hi Alexey,

yes, of course, there are several ways how to deal with it. And, yes, this is only relevant when we need to transform between different datums. Conversions such as e.g. from WGS84 geographic coords  to WGS84/UTM32N metric coords are not touched by this, because they rely on the same spheroid.

Cheers

29

General / Re: How does PS Pro handle geodetic datum transormations?

« on: September 16, 2013, 05:57:22 PM »

Hi Alexej,

thanks for your response.

So it is good practice for the moment to check the resulting ortho carefully for positional errors, when performing a coordinate transformation that makes a datum shift necessary. Or to do the transformation in external application like ArcGIS.

I stumbled across this because exported KMZ did not fit correctly onto GE, while the one created with ArcGIS did.

Cheers

30

General / Re: How does PS Pro handle geodetic datum transormations?

« on: September 16, 2013, 12:02:44 PM »

Hi all,

here comes an update to my question. Please see attached image.

Cheers