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 - Jernej

Pages: [1] 2
1
General / Re: GCPs and accurate reconstruction accuracy estimate
« on: January 09, 2023, 12:36:18 PM »
Thanks for the answer. I've been doing some testing and comparisons with Metashape and RealityCapture. What I've seen is that RealityCaputre usually performs much faster but lacks any reliable and fast way to get good "control" over the alignment process. It has a really poor marker detection algorithm and if you want to mark all of the GCP-s it's a slog. Also, the component workflow on larger projects seems very clumsy. In this way, Metashape, at least to me, has a vastly superior workflow compared to RealityCaputre

2
General / GCPs and accurate reconstruction accuracy estimate
« on: January 03, 2023, 11:35:39 AM »
Hello

On how many pictures in the dataset must you mark the control points to get the most accurate estimate of the reconstruction accuracy of the model?

I have a model that I need to reconstruct with a mean error of less than 1.8 cm. To get the most accurate report of the error should I mark the control points in all of the images where they are visible or should I leave only those that were automatically detected with the Detect Markers function?

This question has been probably been asked a million times already but for the life of me, I cannot find a satisfactory answer.

Thanks for your help and insight.

3
General / Hardware and specs for an ideal processing server
« on: September 06, 2022, 01:08:39 PM »
Hello

I'm currently working on a project (purely theoretical) to build the ideal work server or HPC for the photogrammic workflow. From image processing to photogrammetry itself to rendering and creating assets and such.
I'm interested in the ideal hardware specifications for a machine like this. I've read some articles but none talk about an HPC configuration for such a use case and with Metashape workflow in mind.
My question is, what should I focus more when building such a machine? GPU or CPU processing power, RAM and storage sizes and so on. The projects running on the machine would be large, 20 000 images and more, running simultaneously.

4
General / Depth maps generation explanation
« on: August 30, 2021, 12:46:48 AM »
Hello

I'm currently puzzled as to how depth maps are generated and filtered. This is from the manual:

"Depth maps are calculated for the overlapping image pairs considering their relative exterior and interior orientation
parameters estimated with bundle adjustment. Multiple pairwise depth maps generated for each camera
are merged together into a combined depth map, using excessive information in the overlapping regions to
filter wrong depth measurements."

How do the levels in the GUI come into play here as in what do they mean?
And how is the depth map filtered? What is this excessive information that is used?

I would welcome an explanation.


5
Hello

I'm currently researching the advantages and disadvantages of generating meshes from Depth Maps and Dense Clouds. My findings from trial and error are as follows.
Overall, the Depth map is a superior method of generating meshes when a large enough set of photos is available.  It creates a more detailed mesh with less noise and processing time and skips the dense cloud generating step.
The Dense Cloud method is better when a smaller sample of images is available but creates a noisier, less detailed mesh.

I was wondering if there is a detailed documentation or use case for when to use one method or the other. I couldn't find anything useful on my own.


6
General / Re: Error when building UV for object
« on: July 19, 2021, 09:53:06 AM »
Sent to support because of file size restraints. Hope it helps to solve this issue.

Also, I have tried to texture the object in version 1.7.1 and the problem does not appear there

7
General / Re: Error when building UV for object
« on: July 16, 2021, 04:40:35 PM »
Sent to support. There also appears to be another problem after generating the textures. The UV map generated on some large surfaces is very small and cannot hold sufficient detail.

Example in the attachment

8
General / Re: Error when building UV for object
« on: July 15, 2021, 11:43:37 AM »
The version of Metashape is 1.7.4.
For the mesh generation process I used depth maps, on High quality and 14 milion polygons as the target count.

Output on the node:
2021-07-15 10:21:18 Gutter changed ( 4 -> 0)
2021-07-15 10:21:18 Pack iteration 0 ratio 0.525
2021-07-15 10:22:54 Error: invalid vector<T> subscript
2021-07-15 10:22:54 processing failed in 218.077 sec

Output on the server:
[xx.xx.xx.xx:56507] failed #61 BuildUV: invalid vector<T> subscript

9
General / Error when building UV for object
« on: July 15, 2021, 09:20:54 AM »
When building UV map for an object, I get an error saying "invalid vector<T> subscript". I'm building 4 8k textures for this object. If I lower the count of those textures, the problem dissapears. Any way to avoid or solve this issues without lowering the quanitity of textures? UV mapping mode is generic.

Thanks for your help

10
General / Re: Low quality (part of a) texture
« on: July 06, 2021, 05:00:49 PM »
It seems lika a UV problem where the part that is blurry is really small in the UV space compared to its "real" size. At least that is what I assume based on my examination of the mesh UVs

11
General / Low quality (part of a) texture
« on: July 06, 2021, 04:01:21 PM »
When texturing an object, a blurry low quality patch of texture appears in the middle of an otherwise high quality texture.

The images used are not blurry and the camears are perfectly aligned.
The object itself has about 12 milion poligons. The preset for texturing is in the attachments as well as a picture of the described problem.

This is not the only object i'm currently working on that has this problem.

Any help would be appriciated.

12
General / Looking for papers about the process of photogrammetry
« on: June 23, 2021, 10:15:53 PM »
Hello

I'm currently in the process of writing my thesis which is about the use of photogrammetry in cultural heritage.
I'm looking for any papers, websites, and publications that explain the process of photogrammetry with the use of 2D images (not laser scans) in detail.

Any help would be appreciated  :)

Thanks

13
General / Re: Bad model from laser scan dense cloud
« on: April 29, 2021, 12:09:46 PM »
I have sent part of the cloud via WeTransfer to support@agisoft.com.

14
General / Re: Bad model from laser scan dense cloud
« on: April 28, 2021, 10:01:33 AM »
Hello Alexey

I don't have any cameras in the project. I used 3DReshaper to import (.e57) the cloud and trim it. I then exported the cloud in PLY format.
If the normals aren't oriented the way I want them it shows when the model is generated as darker faces on the model (back facing). I also double-checked the normals in CloudCompare and MeshLab to confirm the orientation.

15
General / Bad model from laser scan dense cloud
« on: April 26, 2021, 10:53:41 AM »
Hello

I'm trying to create a model from a dense cloud that was made with a laser scanner. I exported a part of the dense cloud into Cloud Compare to calculate the normals since Metashape doesn't calculate them facing inside as I want. The cloud has about 55.000.000 points which should be sufficient. I then set the bounding box to cover only part of the cloud from which I want to generate my mesh. The generated model comes out full of holes and noise.
I use the default settings in the mesh generation process and set the face count to high (which is about 11.000.000). 

Any idea as to why this happens and how could I fix this?


Pages: [1] 2