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 - Marcel

Pages: [1] 2 3 ... 21
Feature Requests / Re: Weld vertices
« on: October 29, 2021, 12:09:28 PM »
+1 This would be super useful.

Bug Reports / Re: Filter Photos by Point possibly not working
« on: April 22, 2020, 10:30:40 AM »
For other people who are running into this problem:

The Filter by Tie points does work, but the photos are selected in the Photos Tab (which is not visible by default).

It would be more clear if they would be also selected Workspace->Chunk->Cameras.

Bug Reports / Re: UV packing in 1.6.2 is broken
« on: March 03, 2020, 03:03:35 PM »
I want to voice my support for a parameter or version setting so we can control the UV packing, that would be very useful.

I figured this was similar to the Estimate feature, but am still scratching my head as to it's considering all my images blurry. Maybe I need to get my eyes checked. :)

Nope, Pixelpeeper is completely manual, you have to select blurry images yourself (but with one hand on the pageDown key and the other on your mouse, you can go through the images quite quickly).


The Estimate Image Quality function cannot accurately detect motion blur from camera shake. With this type of directional blur you still have sharp pixels, so the algorithm gives the image a good rating.

Even for moderate blur vs perfectly sharp, the rating the is almost the same (making it almost impossible to cull bad images).

More information in this thread:

General / Re: generation of dense cloud point
« on: August 13, 2015, 10:30:45 AM »
We have had this problem when the alignment was not correct. What happens is that a cluster of cameras (for example 5 cameras) has a slightly incorrect alignment compared to the other cameras in the project (but a correct alignment compared to the cameras in the cluster). Because the alignment within this cluster of 5 cameras is correct, it still contributes points to the Dence Cloud, but they are offset compared to the rest. So parts of the dense cloud have an incorrect position, this leads to "stepping" errors in the Dense Cloud.

I am not sure how to fix this. You could try to run the alignment phase with higher settings (but when we had this issue we were already aligning at the highest possible settings. In our case it was because of slightly blurred images in the photo set).

You can also do a quick inspection of the photos using the free utility we release a while ago:

General / Re: Symmetrical Object Problem
« on: August 05, 2015, 11:33:05 AM »
The link to your images is not working:

404 - File or directory not found.
The resource you are looking for might have been removed, had its name changed, or is temporarily unavailable.

General / Re: Camera station questions
« on: August 05, 2015, 11:32:05 AM »
My experiments with a dual camera setup was before the Camera Stations was introduced, so I can't tell.

Is the object on a turntable, or do you move the object to photograph the other side?

If that is the case, you need to mask out the background. The background confuses Photoscan.

General / Re: PhotoScan swapping to disk with plenty of RAM available
« on: July 27, 2015, 12:32:48 PM »
I've tried several times to generate a mesh from a dense point cloud derived from 172 6MP photos -- each attempt, even with 'low' quality specified, halts midway with an 'out of memory' error.

The Mesh quality setting has no influence on the memory use. For arbitrary Mesh construction, Photoscan always builds the mesh at the maximum number of faces. Only after the mesh has completely been build, will it decimate to the number of faces of your target setting (Low, Medium,etc). That means that if you have a Dense Cloud with a lot of points, it will always take a lot of memory to mesh because it first build the mesh at maximum resolution. (I am not sure if it works the same for a Heighfield mesh, I think so).

It's currently not possible to reduce the number of points in the Dense Cloud, so if your Dense Cloud has too many points you are stuck (only solution is to rebuild the DC at a lower setting). I've suggested Dense Cloud filtering as a solution:

I have no clue why you are running out of memory. Could it be that Linux restricts the amount of memory for a process? On a Windows computer Photoscan has no trouble allocating all memory in the system.

Are you sure that it is Photoscan is using the swap space and not some other program or the OS itself?

General / Re: $600 upgrade for Photoscan Pro machine.
« on: July 22, 2015, 02:21:04 PM »
It's unfortunate that the machine has such an expensive but yet not so strong Xeon processor. Only four phyiscal cores..
Photoscan really benefits from having six phyiscal cores. But upgrading the CPU would mean a new motherboard etc etc.. Not worth the costs.

A 4 core CPU with a high clock speed can be faster than a 6 core CPU with a lower clock speed in some cases. This is because some parts of the processing (for example meshing, texture generation) are not multi threaded very well, so your extra cores are not used. For those processes that run on a single thread, the CPU with a high clock speed will have the benefit.

General / Re: Blurry Patches in Orthophotos
« on: July 22, 2015, 10:14:09 AM »
Did you check some source images that are pointing directly at a blurry patch? The blurry patches seem to have some ISO noise still, so I think the photos itself are blurry.

We just released a free Windows utility to check visually sharpness on a large number of photos (although 4000 photos would be a stretch):

For the best results in Photoscan you need perfectly sharp photos. Checking hundreds or even thousands of photos for pixel level sharpness is no fun at all, so we made a utility called Pixelpeeper that turns this task from "mind numbingly boring" to "slightly annoying".

The program displays magnified crops for each image. You can quickly inspect, select and move images that are not up to quality (for example images with camera shake or misfocus). Multiple cropping modes are available, to show crops of both the center and corners of the image.

Pixelpeeper loads images in both 8-bit and 16-bit, and supports the following file formats:

  • JPEG
  • TIFF
  • PNG
  • PSD (files saved with compatibility mode only)
We found it to be a huge timesaver, so we decided to have it rewritten and release it as a free utility for the (always helpful) Photoscan community.

Pixelpeeper Download (0.99MB)

Windows 7 64-bit or higher required, more than 4GB of memory recommended. No installer needed, the program is a single executable that can be placed anywhere. Use at your own risk, please do not redistribute or re-upload.

Questions or ideas can be posted in this thread or emailed to

General / Re: Projecting 3D point cloud to 2D images
« on: July 14, 2015, 09:12:56 PM »
Hi all,

I would like to be able to project the dense 3D point cloud points onto each of the 2D images used to create the reconstruction.  Is this possible with Agisoft's outputs?


Wouldn't that give exactly the same image as the 2D image used for the source?

I'm curious, what do you want to achieve?

General / Re: Raw file batch calibration/sync of image exposure?
« on: July 10, 2015, 09:54:59 PM »
Which is why I try and shoot with just 1 exposure  ;)
Expose as high as possible ignoring the sky blowing out

Yes, but sometimes the lighting condition changes (like thin with overcast clouds, or just before sunset) and the exposure of your photos will vary even with the camera in M mode.

@Mala: I'll look into RAWTherapee, sounds interesting!

Pages: [1] 2 3 ... 21