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 - stihl

Pages: 1 ... 24 25 [26] 27 28
General / Re: Dell Workstation Configuration help!
« on: November 03, 2014, 01:58:39 PM »
I definitely don't think you need a Titan to gain any noticable speed differences. The CPU is the main bottleneck so to say.
I also would not use a SSD with Photoscan as it writes a lot of small files. Look up any longer term tests with SSD and you'll find that they áll break down after 2-3 months if you use them 24/7 which I think you will with a setup like that...
Of course you can use a SSD to install windows and the software on etc but you'll definitely need a seperate data disk to write the data to.

I also don't see the use of a dual GPU with Photoscan as it barely uses just one.

Hello Alexey,

Thank you for your reply.

The sparse cloud was generated with 500 tie points max per photo. I'll try again with 10.000 points per photo and see if this resolves it.

General / Matched tie points in Sparse Cloud but no data in Dense Cloud
« on: October 29, 2014, 03:41:59 PM »

I'm trying to create an Orthomosaïc. The sparse dense cloud after the alignment doesn't show any holes in the pointcloud.
However after running a medium dense point cloud step, holes in the dense cloud appear.

Can anyone explain why this would happen and how I can resolve it?

I was under the impression that Photoscan uses the sparse cloud to densify the point cloud. So how would holes appear in the Dense Pointcloud when the Sparse cloud doesn't show any holes?


General / Re: Ortho distortion
« on: October 29, 2014, 01:17:43 AM »
What you're seeing are 'True Ortho' artifacts. They're caused when Photoscan is trying to fill in areas of the ortho photo that have no data, not enough data or irregular data because of noise (such as on building edges).

The only solution for this - in Photoscan at least - is to use the ground filter to make a mesh of only the ground points. If you then export that as an Ortho, any object that's above the ground level will have object lean. This means that the ortho is no longer a so called 'True Ortho' but simply an orthophoto.

Unfortunately Photoscan does not yet provide a way of editing the seamlines in an orthophoto. So the end result might still be not what you wanted. Which is a shame because the quality of the ortho's is fine, except for the horrible artifacts.

I know that Photoscan v0.9.1 does a better job at delimiting the artifacts, but that version has a harder time with color balancing the entire ortho. They've changed some stuff from v0.9.1 upwards which gets the colors more accurate but also produces a lot more artifacts.

General / Re: Gpu SLI?
« on: October 14, 2014, 01:17:34 PM »
As far as I know the current version of Photoscan doesn't utilize multi GPU's so for stability wise I'd suggest turning it off as the GPU itself barely gets used.

General / Re: Number of GCPs
« on: October 12, 2014, 11:25:29 PM »

General / Re: Reproducing buildings in aerial mosaics
« on: September 25, 2014, 01:49:30 PM »
Hi Tenboair,

The reason you're seeing the rippled edges on buildings is because of incomplete building geometry and noise.

What Nadar meant to say was that you can try to use the Photoscan ground filter to make a mesh of only the ground points which makes a digital terrain model.
This will cause the ortho image not to be a so called 'True ortho' because you will be able to see the sides of buildings.

I haven't found a proper way to make proper geomtry and no noise for a True Ortho in photoscan.

General / Re: Process images from different heights
« on: September 22, 2014, 06:27:02 PM »
Our UAV system constrains it's altitude by the ground elevation. So if the ground goes up, the uav goes up accordingly. If the ground goes down, the uav goes down accordingly. That way we keep our GSD the same over the entire dataset.

General / Re: Filtering water, vegetation etc.
« on: September 12, 2014, 10:53:14 AM »

Photoscan has a ground filter feature. You should play around with the parameters, it should give you what you want i.e. a DTM (not DEM) with the vegitation etc filtered out. As for the water, if it's the shoreline the easiest way to remove that is simply by manually deleting the water points from the dense cloud.

General / Re: Some of my early work.
« on: September 09, 2014, 12:29:35 AM »
Hi Martin,

Can you share some insight in the workflow you used? Such as what texture creation parameters in Photoscan?


As far as I know the ortho image is generated directly from the dense point cloud and that the mesh is just there to keep geometry in place of objects such as building edges and vegetation.

Generating a mesh from the sparse cloud creates a heavily triangulated mesh when compared to a mesh generated from a high or even ultra high dense cloud. This would explain the weird geometry on your trees.

When you say that the mosaic doesn't seem to be any better for the dense cloud, can you explain what you mean by this?

If your computer is up for it try the settings;
Build Dense cloud:
-Quality: Ultra High
-Depth filtering: Mild


Build Mesh:
-Polygon count: High
-Interpolation: Enabled
-Point classes: All

Run those two after the sparse cloud. It yields very nice results for me regarding trees.

Important to mention is the computer configuration you're using. I'm on 128GB and have processed models with over 3 billion points.

General / Re: Algorithms used in Photoscan
« on: July 12, 2014, 03:52:24 AM »
Have you really just bumped up a topic from page 29?! :o

General / Re: Problem with georeference system in ArcGIS
« on: June 26, 2014, 05:44:21 PM »

I believe he or she did. If that wasn't the case, the DEM wouldn't even show up near the GCP locations in a GIS at all.
This also sounds very similar to the issue I had. I also worked in Amersfoort RD New with the same ESPG code.

General / Re: Problem with georeference system in ArcGIS
« on: June 26, 2014, 05:08:42 PM »

Did you export the DEM as a TIFF or ARC ASCII?

I had serious issues with the GeoTIFF header in the exported TIFF file from Photoscan. This caused a significant shift in the ortho and DEM from the actual GCP locations.

If you have access to the software Global Mapper, could you try and and import the DEM there and export it again as a TIFF or ASCII but with the option "Don't write GeoTIFF header" enabled. This should remove the shift. It has for me at least.
Instead you can also import the Orthophoto in Adobe photoshop and saving it again as a TIFF. This should also remove the GeoTIFF header. Make sure to make a copy of the original .tfw and/or .prj so that the TIFF will have a CRS. Just rename them to the same file name as the newly saved TIFF from Photoshop.

It's an ArcGIS thing though. As we don't use ArcGIS, we never noticed the shift. A client of ours notified us of the issue.

Pages: 1 ... 24 25 [26] 27 28