Topics

16

Bug Reports / Classification looks broken

« on: December 20, 2018, 06:20:41 PM »

Ground point classification gives really bad results in Metashape (but also in PhotoScan 1.4), whereas it was really better before.
Is this normal ?

Here is a screen shot from a standard survey where we can see really random result.

Regards

17

Python and Java API / Merge markers after detection

« on: January 23, 2018, 05:57:25 PM »

Hello !

Would it be possible to merge markers after they've been detected ?

i.e. Does it exist something to create a marker that gather the projections of two (or more) markers that have been automatically detected ?

Indeed, after non-coded target detection (circle), I got several markers really close to each other (less than a centimeter) coming from the detection of an unique object. And at this point I don't know how to gather all the projections into a single marker.

Regards

18

General / Hotkey form custom tool

« on: January 15, 2018, 04:46:38 PM »

Hello !

For some times we are developing several homemade tools in order to make our job easier.
Moreover, must of them are now shared on all colleagues computer and automatically read at PhotoScan start.

Nevetheless, it looks there is impossible to affect (and keep) a hotkey for custom tools. Indeed, the hotkey isn't save after software restart.

Somebody may help ?

Regards

19

Feature Requests / "Raster transform : Palette" manually configured

« on: October 26, 2017, 04:04:39 PM »

Hi !

Would it be possible to allowed a manual configuration for the palette color of DEM ?
At least it would be nice to be able to set the minimum (blue) and the maximum (red) value. Afterward other kind of settings could be considered, such as those which are available in 3DReshaper.

Indeed we would like to make the most of this visualization for flat platform accurate leveling, but for the moment we have to spend a lot of time to accurately manually classify points in a small range of altitude so that the color scale in the smallest possible.

Regards

20

Feature Requests / Alignment step add-on

« on: October 25, 2017, 10:25:07 AM »

Hi !

With the 1.4 pre-release, I've noticed it's now possible to easily add pictures to a georeferenced chunk and align them without reset the current alignment, which is a great tool.

But, it would be nice if those new tie points would be added to the previous tie point cloud instead of being substitute to it.
Indeed, it would allowed to optimize the model after adding those pictures, which is not possible for now.

Regards

21

General / Number of tie points between image groups

« on: October 04, 2017, 01:24:41 PM »

Hi everyone !

Would it be possible, with python scripting or somehow, to get the number of tie points between two predefined groups of pictures ?

Indeed, I would like to assess the robustness of a global alignment of pictures coming from different surveys (typically one drone flight and one terrestrial survey). I can get picture by picture the number of matches (Total, Valid and Filtered), but cannot get the global number between two groups.

I would have the same needs to assess the robustness of an "Align chunk / point based", which could give me back the number of points (and, ideally, there locations) between the two chunks (Total, Valid and Filtered).

Regards

22

Feature Requests / Chunk partial adjustment

« on: September 26, 2017, 03:00:14 PM »

Dear PhotoScan Team !

With 1.4 release, I discovered that you implemented, in the Align Photos tool, the option to not reset the current alignment, which is really useful when added more pictures on a georeferenced model.
That's something I just did with python scripting in 1.3 

But I had a limit that still remain with your option : After added pictures and aligned them on my build model, I would like to adjust them with the magic wand as usual. But, as there is no tie points on most of my main model anymore, I cannot adjust it without losing most of my main model.

Would it be possible to add an option to adjust only a selection of pictures, or maybe only a camera ?
(The purpose of this process would be indeed satisfied with an adjustment by camera)

Regards

23

Feature Requests / Release (Z) from (XY) GCPs coordinates

« on: September 15, 2017, 06:52:55 PM »

Hi !

I would like to know if it could be possible to add this feature at some point.

Indeed, sometimes GCPs have Z-coordinate inaccurate or even missing.
Then, a point could be use as (X,Y) GCP only.

As it's not an daily usage, it could be hide somewhere.
For example after a right clic on a couple of points, we could set them as "none 3D point" and then check or uncheck each coordinate.

Regards

24

Bug Reports / Camera type configuration

« on: September 14, 2017, 12:58:31 PM »

Hi !

Since a while I'm notice a problem in the Camera Calibration panel:

Indeed, when working with more than one camera in a single project, and especially when they are different in type (Frame, Fisheye), the type of a camera is sometimes changing when looking at the second camera.

Example:
A project with a fisheye (GoPro) and a frame (EOS 5D) camera:
- The GoPro is configured as fisheye lens and precalibrated.
- But when I clic on the second camera, the GoPro type change to "frame"...
- If I'm not taking care about that, it lost the precalibration and also use a frame calibration type for it...

May you fix this ?

Regards

25

Feature Requests / Smooth contours

« on: August 29, 2017, 09:32:01 AM »

Hey !

I transfer a topic from General section because nothing can be done about that in PhotoScan and I think it could be a nice add on !

Hello !

I would like to get smooth contours (+- 5cm) on a model which has high resolution data (Area = 17.5ha with GSD = 1.6cm)...
There is indeed no tools to smooth those contours on a hand, and no possibility to get low resolution DEM on the other one...

Any help would be appreciate again !

Regards

Regards

26

General / Smooth contours...

« on: August 28, 2017, 06:41:24 PM »

Hello !

I would like to get smooth contours (+- 5cm) on a model which has high resolution data (Area = 17.5ha with GSD = 1.6cm)...
There is indeed no tools to smooth those contours on a hand, and no possibility to get low resolution DEM on the other one...

Any help would be appreciate again !

Regards

27

General / Shapes projection on DEM

« on: August 28, 2017, 06:34:14 PM »

Hey !

I would like to export some 3D vectoriel data from PhotoScan, such as profiles, but I didn't find any possibility to project my shapes onto the DEM.
I mean, when I "measure ..." a polyline shape, I can visualize the profil with high resolution of data but there is no other possibility to export it than only get polyline vertices 3D infos...

Any help would be appreciated !

Regards.

28

Feature Requests / Markers reconstruction uncertainty

« on: August 28, 2017, 06:21:25 PM »

Hi !

Would it be possible to add a [ Reconstruction Uncertainty ] column in the markers panel ?

I mean, the current last column is [ Error (pix) ] which is kind of [ Reprojection Error ] of markers.
Adding the reconstruction uncertainty would give us the possibility to highlight Ground Control Points which have been projected on pictures which are close (high value of reconstruction uncertainty), facing GCPs which have been projected on well distributed pictures (low value of reconstruction uncertainty).

Regards

29

Feature Requests / Frame extraction from video file

« on: July 25, 2017, 12:10:55 PM »

Dear PhotoScan Team,

Thanks to my work in PhotoScan and around photogrammetry for the 6 last month, I was able to established some automatic or semi-automatic procedures for all kind of projects my firm handles.
Since, we have many different projects to go further in those developments.

For some specific applications, we would like at some point to delegate the data acquisition step. Given that photogrammetry surveys require well defined way of taking pictures, we would like to go deeper with video usage. Indeed, for those specific applications, FHD or even 4K video would give us enough resolution and accuracy, while removing the fear of "not enough covering set of pictures".

Some topics and many works have discussed and showed it's satisfactory to work with video, and even some photogrammetry softwares (but Pix4D for example) have implemented tools to extract desired frames from a video, directly in their environment (no need to add another software in the procedure).

Are some similar tools in the development pipes in Agisoft ? Would be a nice add on and help to retain customers

Regards

30

Bug Reports / Coded targets detection failed

« on: June 15, 2017, 02:36:23 PM »

Hi everyone !
(I would say "Hi Alexey" but maybe somebody else has faced to this problem)

I've handle a couple of aerial measurements using UAV (DJI Phantom 4 Pro) and coded target (circular 12 bits) as GCPs.
As I presented in other topics, I have had lots of test to determine the minimal size of target regarding the GSD, so my following problem does not concern target size.

My surveys were composed of around 1500 pictures for an area of 15ha. 18 targets were set on the floor and measured using GNSS RTK.

The problem is that the target detection struggle to get all the visible targets on pictures. I would say for a 1/3 of targets, they are detected on all pictures, for a 1/3, detected on only a couple of pictures, and for the last 1/3 of targets, nothing is detected.

I attached a set of 4 pictures that show there is no link with distance / size / angle or whatsoever. Indeed on pictures 1 and 4 the target 9 is detected, but not in the pictures 2 and 3.
I got the same result for tolerance 30 and 100...

If somebody can help... Everything was OK before.
Regards