Topics

1

Bug Reports / chunk.importShapes results in crash

« on: October 11, 2016, 04:42:16 PM »

When i try to load a shapefile or KML through the following code:

app = PhotoScan.Application()
doc = PhotoScan.app.document
chunk=doc.chunks[0]  #this is a chunk with an orthomosaic, sparse point cloud, mesh and cameras
chunk.importShapes(path="pathtoshapefile.shp",replace=False,boundary=PhotoScan.Shape.OuterBoundary)

then the system crashes. Since I saw nothing like this issue on the forum i would like to know if anyone succesfully loaded a shapefile through python. Loading the shapefile works when i do it manually.

2

General / Using Global (ASTERDEM) as ground control points?

« on: August 10, 2016, 11:22:33 AM »

I am having problems with my DEM. The details are mostly correct but i get convex/concave dems and large rifts when using photoscan for dem generation from aerial imagery. Would it be valid to use the global dem (Aster DEM or worldDEM V3) as ground control points without manual correction? If so what would be the correct processing method?

What i have done before is extract a ground control point for every 30 meters and loaded these values as ground control points. However, when i optimized camera's based on the ground control points my whole sparse point cloud would tilt way too much and neither the camera's nor the ground control points would still be in the bounding box.

3

General / convex DEM and rifts in DEM

« on: August 03, 2016, 10:47:21 AM »

I am using Photoscan to build DEM's from aerial imagery and at moments i get the following problems:

The DEM that photoscan produces is convex (see attachment) or tilted.

The DEM shows rifts (height can drop over 60 meters within one meter.

Did anyone else experience these issues and what is your solution? Note: ground control points are not possible for me.

4

Python and Java API / autoIt python script start

« on: June 09, 2016, 11:18:21 AM »

Dear All,

In this forum I found that some people have a workaround solution to automate running a script for photoscan. I would like to implement this but cannot seem to get it to work (Especially the run python script window is not very controllable). Does any one of you have a script for this? What I need is:

-Click the run python script icon in the console tab.
-select the Arguments line
- Type in customized text
- Press ok (or enter)

5

Python and Java API / Zero resolution after region resize in python

« on: April 01, 2016, 09:45:46 AM »

I've written a script that sets the rotation, size and center of the region of a chunk. I run this script on an aligned chunk and if i then want to create a dense point cloud i can see that 0 cameras have been selected which leads to an error of Zero Resolution. If i then manually change the chunk boundaries only even a tiny bit then the dense point cloud creation will work. With a tiny bit I mean that the boundaries do not change in any significant way. This leads me to think that when you manually resize the region photoscan updates in some way but I have already tried the following things and nothing seems to work except for manually resizing:

- update the project
- copied the chunk and made the new one active
- saved the project
- saved, closed and opened the project
-optimize cameras after resizing of the region

Does anyone have any ideas about this?

6

Python and Java API / rotate from true north and resize

« on: March 24, 2016, 07:26:00 PM »

I'm trying to resize my region with a new center, size and rotation. I read all topics that are on this forum but could not get a conclusive answer on this. Setting the center is going okay, but changing the size or doing the rotation proves a problem.

about the size: my images are referenced in epsg:4326 so in LatLon. my chunk also has this reference system. i set the size to half the bounding box width and half the bounding box height, which are:

         x                     y
0.001210391   0.00532877

These do not correspond at all with the x and y sizes of the chunk region, which are about 20 and 35. Which function do I need to apply to get to the correct values? I calculated s with the script as given in http://wiki.agisoft.com/wiki/Bounding_Box_to_Coordinate_System.py

but in this s is 44 so still no way to get to the right value.

about the rotation matrix:
I want to rotate my bounding box with +4 degrees from true north but i have no idea how to first set the bounding box to true north and second how to rotate it 4 degrees.

7

General / Error: can't read image

« on: January 27, 2016, 10:46:30 AM »

I'm currently trying to process a large orthomosaic. the whole thing was processed (the second to last processing step stated 'Finished Processing' and right after that I got the error 'Error: can't read image: 2688'. How can this error come about and how can i prevent it from happening again. This were the two final lines of build orthomosaic. The chunk in which i build this orthomosaic contains a suitable DEM on which the orthomosaic is based. The error occurred both with and without color correction. The disk space did increase during the building of the orthomosaic so something was made in the meantime.

Edit: the result was that the entire orthomosaic was not showing up in my chunk and seeing from the save-file size it was entirely removed.

8

General / depth filtering processing time

« on: January 19, 2016, 02:08:10 PM »

Hi All,

I am currently running quite big datasets on my computer and noticed that quite often there is a big time lag in processing time during the depth filtering step. In the attachment you can see a plot of time related to camera count and I would like to know three things:

1: can I expect that with, say,  600 camera's my system will never finish?
2: what parameters can I set to control the number of camera's used in a substep in the depth filtering step?
3: how can i further decrease this processing time?

I'd like to note that my system is not swapping and the RAM is only 30% used. the 32 CPU cores are running full-time (XEON E-5 Dual @ 3.5GHz)

9

General / crashing and slowing down when building orthomosaic

« on: January 09, 2016, 11:50:51 AM »

I am building an orthomosaic of 550 36 MP images with medium settings. I have tried all different kinds of configurations but the system keeps crashing or becoming immensely slow. At the moment there are two scenarios:

I select the entire area and everything processes very smoothly (aligning photos, building dense cloud, building dem). Then I do build orthomosaic (based on DEM), my memory of 64GB fills up in a steady pace, it starts swapping memory and then at some point freezes completely in the georectifying images stage. Then I  have to kill the computer.

In the second method i select only a small region of the whole area (say 10%) after doing all the steps up till build dem. When i build the orthomosaic it then takes over an hour to update the region and update orthomosaic, which is ridiculous and unnecessary and photoscan is not using my cpu, memory usage is low and gpu usage is also low. What is going on here?

Correction: Though I only selected a small region before doing the second method it processed the entire image. that took an hour and nothing else was done then the numerous times that the system just froze. So a solution would still be great!

10

Python and Java API / class names or orders of dense cloud in Python

« on: January 05, 2016, 07:40:10 PM »

Hi All,

I'm trying to create a DEM based on a classified dense point cloud in Photoscan python. The manual only states that I should state the classes as a list of int but i have no idea what integer value corresponds to what class and how this should be called. The manual only states the following:

buildDem(source=DenseCloudData, interpolation=EnabledInterpolation[, projection ][, region ][,
classes])

So what should i put after classes= to only build the dem based on ground points? The step before builddem is classifygroundpoints.

11

General / combining infrared and rgb camera

« on: December 29, 2015, 06:48:58 PM »

Hi All,

Our company is flying with a 36MP RGB camera and a similar NIR converted camera (infracutfilter removed, RGB-filter added). We use these camera's for for aerial photography but the images are not well aligned, because of large distance and inperfect alignment of cameras on mount.

Because of this I have two datasets, one RGB and one NIR. The images have identical gps positions because they are triggered at the same time and receive GPS data at the same moment. Does any one of you have the same setup and what did you do to align the images? I want a well overlapping orthomosaic but normal alignment does not work since the correlation between infrared and RGB is too low.

12

General / cannot export orthomosaic

« on: December 28, 2015, 03:33:01 PM »

Hi All,

I am trying to export my orthomosaic. I have done the alignment, build dense point cloud, build mesh, build texture, and previously this was no problem to then export as orthomosaic. For some reason I cannot get it to work now. The mosaic looks good in the interface itself, the options to export as any datatype just remain gray and when i do batch process export it just says failed.

13

General / Huge GIS dataset (10000 images, 36MP) SSD and RAM requirements

« on: December 23, 2015, 01:22:21 PM »

Hi All,

I've bought a computer with 32GB RAM and 2TB SSD and am fully aware that this will not do for my datasets. I work with 5000-10000 16-BIT 36MP images, which amounts to 1.5-2.5TB per dataset. I am going to upgrade my RAM but at the same time need to increase my ROM.

I am doubting between 128GB or 256GB RAM and would like to know the added value (decrease of processing time).

Also, since swapping will be necessary with these kinds of datasets, should I buy 2-3 2TB SSD's (540MB/s read/write) and put them in RAID 0 or should i go with one 2TB SSD with 2700/1800MB/s read/write?

In any case, should I process in chunks anyway?

Thanks so much for your help!