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.

Topics - outsider

Pages: [1]
General / Selection based on bounding box
« on: May 24, 2020, 07:31:25 PM »
Is there any way to select the elements inside a bounding box?

I did a quick search and came up with this post which deletes all points outside bounding box.

I would like to select everything inside the bounding box(mainly mesh elements, but could be useful for all other objects as well).

Can anyone point me in the right direction or suggest how to modify that script?

General / "Unexpected Channel Count" error
« on: January 13, 2019, 09:06:08 PM »
I'm getting a "Unexpected Channel Count" error in Metashape 1.5.0 build 7492 when I run the "Build Dense Cloud" step.
The equirectangular images align just fine.

This only happens when I have Spherical (equirectangular) images in my dataset. The same spherical images worked with no problems in Photoscan 1.4.4 in the Build Dense Cloud step.

Anyone else getting this error related to spherical images?

General / Increase processing performance by a factor of 2
« on: November 06, 2018, 12:52:12 AM »
I've found that I can increase the speed which models are processed if I run multiple simultaneous instances of PhotoScan on the same machine, and achieve over twice the performance increase!

On a TR1950X workstation:
- Ran a dataset, and I got a processing time of 2670 sec for processing the "Build Mesh" step.
- Used the '' python script, and divided up the model into 4 chunks, and saved each chunk as a separate file.
- Opened up 4 instances of Photoscan on the same machine, and loaded one chunck into each instance.
- Started the "Build Mesh" step in all 4 instances at nearly the same time, and got the following processing time for each instance: 577 sec, 1131 sec, 1132 sec, 1273 sec.

Of the 4 chucks running simultaneously , the longest one was 1273 sec. This means that I was able to process the model faster (by a factor of 2.1, which is quite significant) if I split the work between multiple instances.

On a different workstation, running a i7-6700k:
- Ran the "Building" dataset, and got a procesing time of 512 sec for the "Build Mesh" step.
- Split the model into two chunks, and processed the two chunks simultaneously.
- For the "Build Mesh" step, I got 417 sec and 415 sec.

This time the two chunks finished in almost the same time, but still provided a processing time increase of 1.22 times by running multiple simultaneous instances vs one instance of PhotoScan.

Has anyone else noticed this kind of behavior or done any similar testing?

The relationship between increasing number of cores and processing time is far from linerar.
I'm guessing this type of processing makes the best use of the cores/threads available.

I'd encourage others to try this and post their results on their multi-core systems.

General / Optimum texture size
« on: October 30, 2018, 01:20:38 AM »
An image set taken with a 50MPix camera will have more texture detail then the same set of 10MPix images. How do I calculate what my resulting texture should be (texture resolution and number of textures) so that it takes full advantage of the source image resolution.
Is there a feature I missed which can calculate the optimal texture?

I know that in building a tiled mesh, a choice is given to the amount of detail that the user wants in the texture, but nothing similar exists for generating a regular texture.

Are there any tricks or advice?

General / Exporting Cesium Tiled Models lacks GPS location
« on: September 28, 2018, 03:53:24 PM »
Should the export of a tiled model preserve the model's geo-location?

I have an area scanned with a drone (all images are geo-referenced). When I export the tiled model as a Cesium Tiled Model, the resulting tileset.json file doesn't include the GPS location of the model.

Is this by design?

General / Meshing settings
« on: June 23, 2018, 11:36:16 PM »
Is the Low/Medium/High/Custom setting in the meshing settings just a setting for the decimation step? Or does the High setting actually generate more polys?

General / My observations with a Phantom 4 Pro ; DNG vs JPG
« on: June 15, 2018, 08:27:37 PM »
I know this is a debated subject (RAW vs JPG for photogrammetry), but I just wanted to share some objective evidence I came across. I'm working on a project to figure out how much detail my Phantom 4 Pro can resolve (will post a full write-up on that soon), I came across an interesting observation.
This only applies to images captured with the Phantom 4 Pro as it's the only sensor I've tested so far.

Took DNG+JPG images with the P4P camera, and ran them through PS. (1/200 shutter, f6.5, ISO 400)

The object that is being reconstructed is a calibration chart I made to do the detail resolution test. The object is not very big in the frame (about 480 pixels across) and the image width is 5464 pixels across (so less then 1/10th of the image)

My observations on the source images: (see attached image)
The JPG image (out of camera) looks cleaner. Less image noise. (likely smoothed and sharpened by the jpg compressor)
In contrast, the DNG image looks full or grain and noise.

My observations on the reconstructed mesh:
The mesh created from the DNG files is cleaner while the mesh created from the JPG files is noticeably rougher looking/noisier.

Just to note that DJI doesn't provide ability to change settings for JPG creation.

If you're using a P4P, I'd love to hear from you and what your experience has been with JPG vs DNG.

General / Weird message when aligning
« on: June 15, 2018, 06:45:33 PM »
A weird message gets displayed when I align a bunch of DNG images (shot on the Phantom 4 Pro)
The message is:
libtiff error: input: Null count for "Tag 51008" (type 7, writecount -3, passcount 1)

The images seem to align just fine, but I was wondering what that message meant?
I could not find any reference of it on google.

Anyone get this before?

General / Measuring photogrammetry resolution
« on: June 06, 2018, 10:57:02 PM »
How does one go about measuring the resolution that has been captured with a bunch of photos and the photogrammetric process? By resolution I mean the detail frequency (?is that even a term?) that the point cloud and/or mesh is able to reconstruct.

And I'm not talking about what the theoretical cm/pixel resolution that a certain sensor should be able to capture. I'm talking more about how to measure the highest frequency of detail a certain set of photos can capture, which will of course be influenced by the sensor, lens, distance to subject, f/stop...( and likely other factors).
I know I can set the reconstruction settings to High, or Ultra, or Max, but once the model is complete, what tools are available to tell me how much detail was captured in the point cloud or the mesh, if I compare with a real-life object.

Are there any proven workflows?
Any tips?
Any tools you've seen/used?

Pages: [1]