Forum

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 - kpauly

Pages: [1]
1
General / Re: AgiSoft vs. Dronemapper
« on: September 24, 2012, 10:59:14 AM »
Hi Andes,

Dronemapper is very much a one-man effort still in development. Last time I was in contact with Dronemapper (a few months ago) about a Gatewing dataset displayed on their website they were still working on improving accuracy and resolution, although it visually looks fine. However, unlike PhotoScan, no manual or semi-automatic intervention is possible in the processing and more importantly, no GCP could be used (although they are working on that). The latter drawback essentially restricts Dronemapper to visualization applications and makes it inapt for any mapping applications (especially multitemporal mapping as I suppose conservation applications would require).

Best regards
Klaas

2
Bug Reports / GPU not listed under OpenCL devices
« on: June 27, 2012, 04:42:21 PM »
Hi,

I realize this is a computer specific issue and not a Photoscan bug, but after discussing this elaborately with Alexey and not having solved the issue, I'd like to ask some more opinions:

I have a Dell Precision M4600 mobile workstation with i7-2860QM (running Windows 7 64bit), 16GB RAM and Nvidia Quadro2000M (http://www.notebookcheck.net/NVIDIA-Quadro-2000M.47316.0.html). When I first started using this laptop in February and installed Photoscan (0.8.4 at that time), it recognized and listed the GPU under OpenCL devices (2GB RAM). However, after I have done a clean install of Windows after a hardware issue in my laptop in March, Photoscan still recognizes the GPU but doesn't list it under OpenCL devices anymore. Not in 0.8.5, not in 0.8.4, not in 64 or 32 bit installations. In the console window, I can see the following at startup in 0.8.5 64 bit:

OpenGL Vendor: NVIDIA Corporation
OpenGL Renderer: Quadro 2000M/PCIe/SSE2
OpenGL Version: 4.2.0
Maximum Texture Size: 16384
Quad Buffered Stereo: not supported
ARB_vertex_buffer_object: supported
ARB_texture_non_power_of_two: supported

I currently have the most recent driver installed (296.88 from 13 June) but the problem persists (I have not been able to find the driver which was installed on my system back in February when it still worked, but it is hard to imagine that later drivers support less functionality).
Attached are screenshots from GPU caps viewer.

This issue is not critical for me, but sometimes frustrating. Can anyone offer some insight or suggestions on how to solve it?

Thanks
Klaas

3
Bug Reports / settings in batch process module
« on: October 14, 2011, 10:20:25 AM »
Hi all,

There seems to be a bug in the batch process module under the menu workflow in version 0.8.3. (Windows 7 64bit).
By default, the settings displayed for the different job types are the ones last used directly in the steps under the workflow menu. When I change these settings, click ok, and then reopen the job type dialogue box by clicking edit, my settings are again changed to the initial ones. At first, I thought this was a display issue, but I have reasons to believe that it actually doesn't save the settings that I enter and uses the default instead.
For example, when processing a project of 750 photos (on a somewhat limited hardware setup), it crashed in the geometry building step while running the batch process module, in which geometry building was set by default to high quality. When I first ran the seperate step directly set to medium quality and then canceled and reran the batch module (this time actually displaying medium), the analysis did work.

Did someone else notice the changing settings when editing batch process jobs and can you verify if this is a display issue or a real processing bug?

Thanks in advance
Klaas Pauly

4
General / integration of small regions in an overall model?
« on: August 29, 2011, 01:09:09 PM »
Hi
I'm doing a terrain reconstruction of about 1km2 based on about 700 UAV photos. A first run with medium settings and 200k faces yielded generally acceptable results. However, I reconstructed a couple of parts in more detail (up to 2M faces) because of some flaws or shortcomings in this overall model. To do the detailed reconstructions, I selected the areas of interest by resizing the volume bounding box with the "resize region" toolbar button from the overall model.
I wondered if there is a way to subsequently reintegrate the detailed models in the overall model, i.e. by replacing or updating the flawed parts by the more detailed ones?

Thanks for any advice
Klaas

Pages: [1]