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 - Bruno Andrieu

Pages: [1] 2 3 4
Feature Requests / mesh interpolation
« on: April 30, 2015, 11:56:22 PM »

I am working with  2D objects (leaves of plants). Often I would like to use some interpolation when constructing the mesh, however the default interpolation results in a large border around each leaf, so that the area is much more than the real one.

I would find very helpful to have a parameter that defines the amount of interpolation. So that we can have  intermediates between "no interpolation" and "defaut interpolation".

I would be interested also to know about alternative methods to solve this?


Bug Reports / issues with align chunk
« on: April 07, 2015, 09:23:50 AM »
An issue has appeared with "align chunk"

When I align several chunks now each of the aligned chunk contains the point cloud of the whole dataset.
So it seems as align does as align+merge

General / select photo by point(s) or by selection
« on: March 10, 2015, 11:28:01 PM »

If I select a number of points in the point cloud, I get different results if I ask "select photos by points" or "select photos by selection".  however I do not understand what is the difference. 

Also : I do not undertsand how to  use the "select photo by point" (without s) feature.

Can you explain ?



Bug Reports / Re: thin tie points let much more points than asked
« on: March 07, 2015, 06:08:32 PM »

Thanks, Alexey for the explanation !

(but I would  suggest to change the dialog in the menu to make this fully clear)

Bug Reports / Re: Detect Markers - disable parity
« on: March 07, 2015, 06:00:04 PM »

It would be very useful that we can keep as an option to have old target detection.

In my case data acquisition is costly and labor intensive, and it is important that I can re-use photographs taken previously, they represent information of high value. At the moment re-using them is much complicated because of target detection issues (one data set represents typically 200 photographs with 50 markers on each of them...)


Bug Reports / thin tie points let much more points than asked
« on: March 07, 2015, 05:43:11 PM »

When I use the "thin tie point" function, I notice that it results in approximately 10 time more tie point left  than asked.
For instance, asking to thin a 1000 000  point sparse cloud  to 50 000 would let 450 000 points in the sparse cloud .


Feature Requests / infos on markers
« on: January 21, 2015, 12:12:02 PM »
Thanks for the the useful improvments found in 1.1 version.
Amongst other new features, It is useful to have the "info" panel  on markers, that gives the positionning error of a marker for each image. I use it quite a lot.

I have two related suggestions :
-In the panel that gives the list of cameras and associated errors  for a marker, it would be nice if right-click on a camera name would open the corresponding image.  This would allow to go faster when checking a marker on a number of images.
- It would be nice if we would find in the info panel of the cameras a similar information, that is the list of markers present on the images , with the associated error.  So when a camera appears badly positionned, the user can easily see which markers should be checked.


Since there  was no answer from Agisoft for this point, I would like to stress that a 3D coordonnate picker  would be a very useful feature.
Is there any hope to have this ?


when optiimizing the cameras parameters, I usually adjust parameter k4.
I think it would be good that once one asked this for one chunk, the option for this fitting would stay to "on" for the chunk . As it is, it returns to "off" so that we have to put it on again at each new fit.


General / Re: decrasing the number of tie points : how is it done ?
« on: November 04, 2014, 04:25:16 PM »
Thanks  for the explanation

Does photoscan compute some value of likelihood when  associating  points from two different images ?

It would be nice to have  a  global value for the sparse point cloud that would reflect the " average" likelihood of the tie points, or even to give some statistical  distribution.

I try to have the "best" sparse points clouds to optimise alignment. So  I proceed with removing outliers, selecting points identified on more than 2 cameras, and now there is this new thining option.  I would find useful to hace a measure of the achieved quality

It would be great to have a measure of  accuracy of the sparse point cloud  and  even (great +++)  of the information brought for alignment ( to choose between less but highly accurate points or more points with lower accuracy ? )


Bug Reports / Re: model shown with a null size
« on: November 04, 2014, 02:31:00 PM »

They were checked cameras, and unchecking them  worked ,  thanks !!

However I have no idea of what happened.
Can you explain what is the use of checking/unchecking cameras  and what was the problem here ?


Bug Reports / Re: model shown with a null size
« on: November 04, 2014, 02:05:02 PM »
Hello this is he screen shot.

Note that  I reduced to 1000000 the size of the point cloud, by gradual selection
So the point cloud does exist !

Bug Reports / Re: model shown with a null size
« on: November 04, 2014, 01:45:11 PM »

I work with local coordonnate so I did not give any absolute position for the markers.

General / Re: decrasing the number of tie points : how is it done ?
« on: November 04, 2014, 01:37:04 PM »
Thanks Alexey

And does it make a difference wether we limit the number of tie points during the alignment process ( the new option in v1.1)  or we thin the point cloud a posteriori  ?

Bug Reports / model shown with a null size
« on: November 04, 2014, 12:40:59 AM »

This is with version 1.04:
I got approx 200 cameras aligned and a sparse dense cloud of 2000000 points.
But I cannot see the sparse point cloud on the screen.

I can see the flags for the markers. They all superpose on the screen.
 It looks as the visualisation is calculated for an observer at an infinite distance  and so the  model is represented with a size null.


Pages: [1] 2 3 4