Messages

31

General / Re: Agisoft PhotoScan 1.2.0 pre-release

« on: August 07, 2015, 11:21:05 PM »

Paul, yes, i know,


I do also see these white features, but always where there are 'problems', e.g. 1. a tree appears in the image, but is not described in/by the model, 2. features in water, which lead to very noise point cloud/model 3. dark regions in the images (shadow, for instance), etc.  At it seems, the regions of images which cannot be assigned to a region in the model - be it because that region in the image is bad, or because the model is bad or missing for the respective region, PS creates a hole in the image.

This seems to be consistent with my observation.

@Paul, in your model the dark zones correspond exactly to the problem region.

If you want, i can give you examples of one map I did.

Cheers

32

General / Re: Agisoft PhotoScan 1.2.0 pre-release

« on: August 07, 2015, 10:22:35 PM »

I suppose that the holes in the images available for the editing of the seam line come from missing 3D information when orthoretificating of the original photos. Or am I just writing nonsense?

Well, nevertheless, with this new feature, PS will really became unbeatable!!!!!

33

Feature Requests / Re: Point Cloud classification by color

« on: August 07, 2015, 09:58:03 PM »

the average blending had the advantage that the orthoimage looked much more plastic... it was always a bit more blury, but as an advntage, you could really see the stiff borders of objects (well, in our cases trees, since we are working in forests).

What I mean can be seen in a webmap we produced some time ago http://elsilencio.forestvisions.it/

I prepared a small video which shows how to use it. https://youtu.be/V4pADuIL0ao

So, as you can see, through the average process you get border lines on the objects (probably form shadow information, I never really understood how they originate) which are an additional information for the viewer (user) to interpret the maps.

I hope you understand what I mean.

The other advantage we had with averaging was that you got rid of different exposures in the image (shadow areas vs sunny areas, for instance). This made the map more homogeneous. I still did not check how this aspect works with the new version of PS. But at least it was an aspect which was important for us with the averaging of the color values.

Cheers

34

Feature Requests / Re: Point Cloud classification by color

« on: August 07, 2015, 09:15:36 PM »

the only thing i do not find and which is very important for me is the edition of the seam line. Where do I find it?

35

Feature Requests / Re: Point Cloud classification by color

« on: August 07, 2015, 09:07:42 PM »

Well I musst say: you're really the best...

I did not only check the selection of points by color, which is a great and very important tool, but also the new mosaic generation (very quickly).

I was almost irritated, that the 'average' option is gone from the possible export types, but when i opened the new ortomosaic, the image is perfect. And I'm talking about orthomosaics of forests, which are really difficult to look clean.

Nevertheless the question, why is the average option gone?

36

Feature Requests / Re: Point Cloud classification by color

« on: August 04, 2015, 08:49:52 PM »

Uauu, that's fantastic.

Just pitty it didn't came earlier, since we lost the possibility to implement a procedure based on Agisoft on a large scale due to this missing feature.

But there is still another client who would represent a good chance to use Agisoft with this feature included for a process which would could be used also on a  larger scale (i.e. several teams would need a Agisoft copy for it),

But first I'll try the feature as soon as I find a possibility.

I'm really curious how effectively it works.

Congratulations for this new implementation.

37

Feature Requests / Re: Limiting mosaic sizes when generating texture from a large number of images.

« on: May 20, 2015, 04:56:21 PM »

Ok, is there no comment on this one?

Pitty 

38

Feature Requests / Re: Point Cloud classification by color

« on: May 13, 2015, 09:45:46 PM »

OK, so I give up on this issue. 

I hope this will be integrated some day, be it through the UI, be it through phyton interface.

39

Feature Requests / Reset Mask for more than one Photo: How?

« on: May 13, 2015, 05:55:02 PM »

I have all my photos with a mask, which I want to reset.

In the context menu of the Photos panel I find a  IMPORT...., EXPORT... and INVERT MASKS  but no RESET MASKS.

The command RESET MASK can only be found in the menu PHOTO > RESET MASK.

The problem is, it applies only to one (1) photo...  But I have several hundreds of photos where I want to reset the mask.

First: how can I reset the mask of more than one or even all photos
and Second: Wouldn't it be logical to have  a RESET MASKS in the context menu of the photo panel?

Cheers

40

General / Re: Slight rotation in referenced model

« on: May 11, 2015, 11:11:26 PM »

Hello,

did you also use ground control points? Which drone do you use?

I have had similar issues. In some cases I was aware about the reason, in some other cases not really.

One reason I found out was related to the geo-referencing of the images. Due to missing geo-tags (already one is enough) the tagging of the photos using the software of my drone company interpolated the geo value for all other photos, so there was a slight shift, which however led to shifts or slight rotations of the model.  But this was more than 1 year ago and the geo-tagging process was improved.

Nevertheless, in my last work, I see also a shift and slight rotation of the final model, although the error should be smaller according to the results of PS. I'm not able to understand this and I will write to my drone company to ask about possible connections to the GPS values of the drone.

I'll let you know if I find something out.

41

Feature Requests / Re: Point Cloud classification by color

« on: May 11, 2015, 11:02:24 PM »

Can someone help me? 

42

Feature Requests / Re: Point Cloud classification by color

« on: May 09, 2015, 02:48:39 PM »

Hi Alexey,

somehow I was expecting that answer ;-)

That's why I started to go through the Python Reference Manual.

I would like to have access to the color value of each point of the dense point cloud, but I do not find anything in the manual which corresponds to this.

Could you give me a hint?

I would try to write a script myself to select points by color to mark and delete them from the cloud.

Cheers

43

Feature Requests / Limiting mosaic sizes when generating texture from a large number of images.

« on: May 08, 2015, 07:44:22 PM »

Hello,

I have been having a problem for long time now which I would finally like present here.

Situation: I do mapping of forests using UAVs. In many cases of high resolution, which means low flight altitude. This in turn means that I have to increase the amount of image per area in order Photoscan to be able to reconstruct the forest structure (which is generally quite complex) generating points from all possible canopy and ground surfaces.

The problem: The reconstructed structures (DPC, MESH, etc) has a high level of detail, which gets better the more images I have. But the more images there are, the smaller the mosaics which are generated during orthomap generation. This is so extreme, that the size of the mosaics end being only of a few pixel. In some cases this might lead to 'nicer' images, but generally this is not the case. Or you have situations, where the character of the map changes from one spot to the other due to different image densities. The screen-shots attached to this post show what I mean.

I could manually reduce the number of available images for the texture generation, but when you have projects with several thousand images, this gets a bit impracticable. 

Possible solution: It would be very helpful to be able to determine the smallest size of the mosaics which are generated during the orthomap export. This would help to control and avoid bad results.

Is it possible to integrate such a functionality in PS?

Best regards
Patribus

44

Feature Requests / Re: Point Cloud classification by color

« on: May 08, 2015, 06:22:02 PM »

Please, dear Agisoft Team, give me just a hint, if I can expect some development in this direction,

because if not, I'll really have to search and find some other software alternative for the methods I'm developing.

Cheers

45

Bug Reports / Re: PS does not merge dense point clouds

« on: May 06, 2015, 02:23:01 PM »

perfect, it worked!