1
General / Re: Display problem with RTX 4000 Ada Generation
« on: March 23, 2026, 06:21:26 AM »
Clean install of driver 582.41 fixed the problem.
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.
Pages: [1]
2
General / Re: Display problem with RTX 4000 Ada Generation
« on: March 23, 2026, 05:48:01 AM »
I have the same problem with RTX4000 Ada Generation, driver version 595.79, CUDA toolkit version 13.2.51
What driver version is working for you?
What driver version is working for you?
3
Bug Reports / Re: 2.3.0 - No cameras with depth maps in working volume
« on: October 11, 2025, 07:21:11 AM »
I have the same experience, using a 96 image fisheye set that was processed fine by v.2.2. With either of the new fisheye models, alignment succeeds and gives a good sparse cloud, but create dense point cloud fails with error "Zero resolution", after apparently building and saving depths maps successfully. The log shows that it was unable to locate any usable cameras. Build model fails for the same reason, but issues a different error message.
I did verify that there is no such problem with the 'frame' camera model.
The statistics of the sparse alignment are as good as with the old fisheye model, however there is a some structure in the residuals, indicating that the actual lens curve was not fit perfectly. That may be because my lens has neither equisolid nor equiangular projection, but a nearly stereographic one -- it is the Samyang 8mm f/2.8 series 2, widely used by panographers. You might want to consider adopting a more flexible fisheye model like the one used in PTGui -- the gold standard pano stitcher -- which is based on the universal lens curve that Donald Gennery developed for NASA's moon missions. It is just a simple one-parameter trigonometric formula, but it can fit all realizable lens curves remarkably well.
I did verify that there is no such problem with the 'frame' camera model.
The statistics of the sparse alignment are as good as with the old fisheye model, however there is a some structure in the residuals, indicating that the actual lens curve was not fit perfectly. That may be because my lens has neither equisolid nor equiangular projection, but a nearly stereographic one -- it is the Samyang 8mm f/2.8 series 2, widely used by panographers. You might want to consider adopting a more flexible fisheye model like the one used in PTGui -- the gold standard pano stitcher -- which is based on the universal lens curve that Donald Gennery developed for NASA's moon missions. It is just a simple one-parameter trigonometric formula, but it can fit all realizable lens curves remarkably well.
4
Camera Calibration / Can I place tie points by hand?
« on: August 13, 2025, 04:48:36 PM »
I'm using Metashape standard. For aligning some difficult image stets it would help if I could place tie points manually by clicking on images. Is that possible?
5
Camera Calibration / Re: Using exported camera parameters
« on: February 19, 2024, 02:18:47 AM »
Thanks, Alexey
I am testing those formulas from the appendix of the manual, using tie point data exported in BlocksExchange format and camera calibration data exported in Agisoft XML format. I am getting reasonable results with the frame camera model but not with the fisheye model. It appears that the focal length in pixels exported for the fisheye model is too small -- only about 60% of the value I get from other calibration methods. Am I misunderstanding something?
-- Tom
I am testing those formulas from the appendix of the manual, using tie point data exported in BlocksExchange format and camera calibration data exported in Agisoft XML format. I am getting reasonable results with the frame camera model but not with the fisheye model. It appears that the focal length in pixels exported for the fisheye model is too small -- only about 60% of the value I get from other calibration methods. Am I misunderstanding something?
-- Tom
6
Camera Calibration / Re: A better way to calibrate cameras
« on: February 05, 2024, 11:13:31 PM »
My suggestion is not about finding camera poses, but rather establishing the internal camera parameters that determine the mapping between incoming rays and image positions; specifically the principal point, focal length and distortion parameters. These are critical to accurate modeling, and best practice is to estimate them by a calibration procedure that is independent of the model data set .
The traditional calibration method relies on photos of charts having known structure, usually a planar chess board pattern. This is the method supported by MetaShape software. It works well for rectilinear lenses of long focal length, but not well for wide-angle and fish-eye lenses, which are actually more useful for large scale modeling. The main problem is that a chart of practical size, presented at a realistic working distance, covers only a small part of the field of view of such a lens. Placing the chart close to the lens is not a good solution, because at that distance the distortion function can be quite different from the one for a subject at normal working distance.
The robot panohead method uses photos taken at normal working distance, and substitutes known camera pointing angles for known subject structure. I know from experience that it finds very accurate and repeatable lens parameters. That includes the principal point, which is fundamental to accurate distortion correction but is so difficult to estimate from model data that the default in MetaShape is to not even try. This is unfortunate, because the principal point depends on mechanical factors that can change from one shoot to the next, namely play in the lens mount and temperature. Best practice with a removable lens camera is to shoot a calibration series at the same time as the model images, with the lens held firmly in position for example by an elastic cord.
The traditional calibration method relies on photos of charts having known structure, usually a planar chess board pattern. This is the method supported by MetaShape software. It works well for rectilinear lenses of long focal length, but not well for wide-angle and fish-eye lenses, which are actually more useful for large scale modeling. The main problem is that a chart of practical size, presented at a realistic working distance, covers only a small part of the field of view of such a lens. Placing the chart close to the lens is not a good solution, because at that distance the distortion function can be quite different from the one for a subject at normal working distance.
The robot panohead method uses photos taken at normal working distance, and substitutes known camera pointing angles for known subject structure. I know from experience that it finds very accurate and repeatable lens parameters. That includes the principal point, which is fundamental to accurate distortion correction but is so difficult to estimate from model data that the default in MetaShape is to not even try. This is unfortunate, because the principal point depends on mechanical factors that can change from one shoot to the next, namely play in the lens mount and temperature. Best practice with a removable lens camera is to shoot a calibration series at the same time as the model images, with the lens held firmly in position for example by an elastic cord.
7
Camera Calibration / A better way to calibrate cameras
« on: February 02, 2024, 08:52:50 PM »
Many panoramic photographers now use motorized robot rotators that can position a camera with milli-degree accuracy. These devices turn the camera around a fixed point, so are not suitable for acquiring images for photogrammetry. However they do make possible a new way of calibrating cameras and lenses, that is both more convenient and more accurate than traditional methods based on special targets. In brief, one takes a series of photos of any suitably detailed large subject, turning the camera by a small angle between shots, and processes them with a panorama stitcher set up to estimate just the camera parameters, the pointing angles being given.
Suitable tools for this are the Fanotec Nodal Ninja Mecha robots and PTGui, the professional stitching software. Mechas cost little and are accurate enough. PTGui works similarly to a sfm tool, identifying matching keypoints and aligning them by a kind of bundle adjustment. The estimation problem is simpler because the photos all have the same viewpoint; consequently the camera parameters are usually very accurate -- especially so when accurate view directions are provided a priori.
Unfortunately PTGui's lens model differs considerably from any of the ones supported by MetaShape, so there is no possibility of a direct transfer of parameters. However, I think it would be worthwhile to explore possible ways to overcome this barrier, as the robot calibration method really is superior.
Suitable tools for this are the Fanotec Nodal Ninja Mecha robots and PTGui, the professional stitching software. Mechas cost little and are accurate enough. PTGui works similarly to a sfm tool, identifying matching keypoints and aligning them by a kind of bundle adjustment. The estimation problem is simpler because the photos all have the same viewpoint; consequently the camera parameters are usually very accurate -- especially so when accurate view directions are provided a priori.
Unfortunately PTGui's lens model differs considerably from any of the ones supported by MetaShape, so there is no possibility of a direct transfer of parameters. However, I think it would be worthwhile to explore possible ways to overcome this barrier, as the robot calibration method really is superior.
8
Camera Calibration / Using exported camera parameters
« on: February 02, 2024, 08:40:42 PM »
In my own software, I need to map 3D points to image coordinates using points and camera calibrations exported by MetaShape. To be sure I am doing this right, I would like to verify the mapping formulas that you use. Is that information available? I am specifically interested in the frame and fish-eye camera models.
9
Feature Requests / Manual photo alignment in Standard edition
« on: June 22, 2014, 12:58:53 AM »
I have no budget for the Pro level Photoscan, nor do I need its many georeferencing features. But I do think I deserve a chance to manually correct the situation when the automatic photo alignment fails. This is not uncommon in my experience on photos whose overlap, and visible features, I would judge as more than adequate for panorama stitching (which is my primary occupation and interest). Even a very primitive way to guide the s/w to a valid alignment -- such as pointing out which edges of 2 images overlap -- would probably be enough; in other words the solution need not involve your "marker" system, use of which seems to be limited to Pro version. Autodesk 'Photo on Recap 360' allows such manual "stitching" at the low-cost level, and I think Agisoft should also.
Pages: [1]
