Messages

1

General / Camera correction and pitch roll yaw input values used for alignment yet?

« on: June 21, 2013, 03:16:14 PM »

Are the values of pitch, roll, yaw, and X Y Z (offset?) input in the Camera Correction screen of alignment enabled in 0.9.1?
I have precise values for these parameters (quality GPS/IMU) and seem to be getting mixed results in enabling camera correction for the alignment output when using them. Does anyone know if this is enabled, and if so, best strategies to use it?
For example, if the pitch error from input values to output values in a good project is about a 1 degree offset, ( 1 degree plus or minus about 1 degree)  is it useful to enter a pitch correction of 1 degree (or 179 degrees since the initial is always 180 degress?) in the same project or subsequent projects using the same GPS/IMU geometry?
This post is somewhat an extension of post:
http://www.agisoft.ru/forum/index.php?topic=370.0

which indicated camera correction was not enabled back in 0.8.5

Thanks

2

Python and Java API / Re: align chunks and build depth failures, ground control preselection?

« on: June 21, 2013, 06:07:30 AM »

The twin processes of matching points and then aligning in the scripted environment seem to take 2x as long as performing the single nonscripted task from the gui application? Does anyone else find this to be true? If so, is their some way to step directly into aligning from matching to avoid the double step process?

3

Python and Java API / Re: align chunks and build depth failures, ground control preselection?

« on: June 21, 2013, 12:19:26 AM »

Alexey and James,

Thanks again for the great help, here is the simple working script in case it helps anyone else:

import PhotoScan
app = PhotoScan.Application

pathfile = PhotoScan.app.getOpenFileNames()

doc = PhotoScan.app.document

#first loop is just QC to make sure all the files exist and filenames are correct:
for files in pathfile:
        doc.clear()
        doc.open(files)

for files in pathfile:
        doc.clear()
        doc.open(files) 
        doc.activeChunk.matchPhotos(accuracy="high", preselection="ground control")
        doc.activeChunk.alignPhotos()
        doc.save(files[:-4] + "a.psz")

4

Python and Java API / Re: align chunks and build depth failures, ground control preselection?

« on: June 15, 2013, 08:18:39 PM »

James and Alexey,

Thanks very much for the suggestions. I rebuilt the paths very carefully by hand typing rather than copying the path, and the opening of the files works properly when I delimit out the matchphotos and alignphotos, so that is not a problem with the file and pathnames. but when I undelimit the matching and aligning, I get the following errors:

Traceback (most recent call last):
  File "J:/alignmodels.py", line 15, in <module>
    doc.activeChunk.matchPhotos(accuracy="high", preselection="ground control")
AttributeError: 'NoneType' object has no attribute 'matchPhotos'

The 0.9.1 documentation does not mention that ground control preselection is implemented, can this be the problem?

Has anyone started up a repository of some example or useful scripts? That might be quite helpful, I have not found quite relevent examples in this forum yet for the multi file batch processing I am trying to undertake.

I thought I might query the user for a file name and then print that out with something like this:

import PhotoScan
app = PhotoScan.Application
testfiles = app.getOpenFileName()

but got errors that the getopenFileName object needed a photoscan application object, but tried many but no luck.  If anyone has example scripts that do any of these sort of things that would be extremely helpful to review.

Thanks,

5

Python and Java API / Re: align chunks and build depth failures, ground control preselection?

« on: June 14, 2013, 03:45:24 AM »

Alexey,

Yes, using absolute paths, e.g:

            "J:\\Midwest\\2013_06_10_DC\\LenoxHi\\Ortho\\LenoxHi",
            "J:\\Midwest\\2013_06_10_DC\\SchuylerHi\\Ortho\\SchuylerHi",
            "J:\\Midwest\\2013_06_10_DC\\Imperial\Ortho\\Imperial",

and the error returned is:


Traceback (most recent call last):
  File "J:/Midwest/2013_06_11_LS/alignmodels.py", line 12, in <module>
    doc.activeChunk.matchPhotos(accuracy="high", preselection="ground control")
AttributeError: 'NoneType' object has no attribute 'matchPhotos'

6

Python and Java API / align chunks and build depth failures, ground control preselection?

« on: June 13, 2013, 11:03:07 PM »

I am unable to get some basic scripting operations to execute, seemingly due to parameters not being properly specified. Is 'ground control' preselection enabled in scripting?  If anyone can suggest the correct scripting to execute the following it would be much appreciated, in both the align and the build opertions return errors:

import PhotoScan
doc = PhotoScan.app.document
pathfile = ('c', .... )

for files in pathfile:
    doc.clear()
    doc.open(files + ".psz")
    doc.activeChunk.matchPhotos(accuracy="high", preselection="ground control")
    doc.activeChunk.alignPhotos()
    doc.save(files + "a.psz")


    doc.activeChunk.buildDepth(quality="lowest")
    doc.activeChunk.buildModel(object="height field", geometry="point cloud", faces=50000)
 

7

Bug Reports / Unable to override groups camera calibration

« on: May 23, 2013, 04:05:08 PM »

In the 0.9.0 build 1586 version of photoscan, I have a single camera used for an aerial photo project, and at sometime the calibration was set to groups and permitted the groups to vary. I am trying to import a high precision calibration for all cameras, and setting the calibration to fixed. But every time, the resulting refined calibration shows new groups orientated by flight line. The fixed calibration does not seem to work to stop the regrouping of the calibration.

8

Camera Calibration / Re: iwitness calibration erroneous focal length?

« on: April 21, 2013, 04:23:53 PM »

Alex,

That was it, Thanks! it was actually 0.017 for a pixel size. If you update the help section for lens, you might toss in a screenshot of example numbers to help users and cut down questions.

Thanks again for the fast solution.

9

Camera Calibration / iwitness calibration erroneous focal length?

« on: April 20, 2013, 10:04:10 PM »

I am trying to import an iwitness calibration into photoscan and lens, but getting unusual values.
Attached is the calibration report from iwitness, and the pixels are 17um and the camera is 640 x 512 yielding a sensor size of 10.88 x 8.704
However, when I convert, I get an impossible focal length?