Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - dcm39

Pages: 1 [2] 3
Indeed they are, but the precision to which they are aligned is not the same as that achieved by feature matching unless the ground control is very accurate.

Feature Requests / Ground control based pre-selection for chunk alignment
« on: September 08, 2014, 02:24:56 PM »
Hi, not sure if this exists some how already, but I don't think it does.

When running chunk alignment, I have ground control for each chunk separately, but it seems that the "preselection" option for the point based alignment is just the simple "generic preselection".

For a nice linear survey with three chunks, the software really shouldn't need to try and match all of the photos, only the nearby ones. Can we have a "ground control" preselection for chunk alignment?

Feature Requests / Re: Marker errors for each photo
« on: September 08, 2014, 02:06:30 PM »
Awesome, thanks Alexey.

Bug Reports / Re: path structure failiure in Linux
« on: September 08, 2014, 01:33:24 PM »
Thanks very much Alexey, that has resolved the problem.  It would be good if it didn't happen in the first place though!

Feature Requests / Marker errors for each photo
« on: September 08, 2014, 01:29:45 PM »
It would be REALLY useful to have some measure of the error pixel error for each photo that a marker is picked in. Maybe this could be included in the photos pane when the details view is chosen (when the photos are filtered by maker)

At present, if you pick a marker in a photo that is poorly positioned, it really skews the location of the marker, but it is extremely difficult to go back and work out which photo is the problem!

Feature Requests / Re: Aerial Footprints
« on: September 08, 2014, 01:26:30 PM »

Feature Requests / Re: Run photo align -> multiple chunks detected
« on: September 08, 2014, 01:24:37 PM »

Bug Reports / path structure failiure in Linux
« on: September 08, 2014, 01:19:31 PM »
Somewhere around the dense cloud reconstruction step, the path to the photos gets mixed up within the project.

When the photos were loaded into the project, they were navigated to via a soft link. I suspect this problem is due to a mistake in treating soft links in the linux version of the software, but it only became a problem later after resaving the project.

Question: If the paths to all the photos in the project have been corrupted in this way (4000 photos for this project), is there any way to relink them? I really don't want to spend several days rebuilding the dense point cloud.

General / Re: TUTORIAL - Request
« on: April 22, 2014, 07:56:53 PM »
A good tutorial about the basics of photogrammetry and understanding some of the algorithms behind the functioning of Agisoft would be amazing as well.  ;)

I agree... some details on what algorithms PS uses would be both useful and interesting. It is very black box at the moment.

Bug Reports / Re: Bug in "Export Ground Control data to file"
« on: February 27, 2014, 07:53:17 PM »
Brilliant, thanks very much Alexey, you are quite right!

It might be useful however to add another tick box to the export dialog for the source values - so that you don't have to export them...


Bug Reports / Bug in "Export Ground Control data to file"
« on: February 21, 2014, 06:37:03 PM »

There seems to be a bug in using the "export ground control data to file" tool in the ground control panel. I have run as follows:

- Import photos
- Import ground control data (the rough guess positions and orientations of cameras)
- Align photos
- Export ground control data (I want to export the newly estimated values.)

I chose the column based text format for output, then in the given dialog box I tick the box "save estimated values" and the "save orientation" boxes.

The resultant text file is NOT the new estimated values, but instead the original source camera locations... This doesn't seem right! I need the new positions, not the old ones.

I am running build 1812


Feature Requests / Re: Request: Dense Cloud Filtering
« on: February 20, 2014, 09:38:41 PM »
+1 more for dense cloud filtering/decimating.

At present, if I want to use the point cloud I filter it with the GMT blockmean command. But it would be really useful to be able to do this before the meshing in the photoscan workflow.

Other programs also aren't really able to maintain the vertex colour  etc... again, would be useful!

Feature Requests / Re: Bundler-compatible image list
« on: February 20, 2014, 09:33:49 PM »
I agree that a bundler_list.txt type export would be pretty useful...

It is just a list of the photo filenames, so for now I use a short grep and sed command to do the job if you export the cameras as an .xml file:

grep "<camera "   cameras.xml  |  sed 's/\"/ /g'  |  awk '{print $3, $5}'   >  cameras_list.txt

Hope that helps

Feature Requests / Re: Adding new photos to existing chunk
« on: February 20, 2014, 09:29:06 PM »
I would second this - being able to match a new photo to an existing chunk would be an extremely useful tool. It is really clunky to have to completely re-align all the photos just to a single new photo (that wasn't included when the matching was done).

Maybe the best way to do this would be to split the align photos into two steps: "Matching" then "Aligning"? In particular, setting the matching such that the user can find the extra matches for the new photo...

Bug Reports / Re: Memory leak in point matching
« on: February 20, 2014, 09:16:02 PM »
Thanks Alexey, the problem certainly seems to be solved now.

The time cost dependency on the number of features per photo seems to have changed now though... The time required used to scale really badly with the set maximum points limit, whereas now doesn't have much effect.

Has the algorithm changed somewhat? Have you introduced some filtering to reduce the number of points which actually get matched?


Pages: 1 [2] 3