Forum

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 - Seb_B

Pages: [1] 2 3
1
General / Re: Any tips to use images as GCP
« on: May 02, 2020, 03:18:05 PM »
A bit later, but thank you very much!

2
General / Any tips to use images as GCP
« on: April 22, 2020, 02:56:45 PM »
Dears,

I have some images of my ROI with a high precision in XYZ, ypr. I would like to add them to my chunk with all other images of the ROI, but with its own accuracy (not the overall accuracy in the setting menu for all the images). Any idea how to do this?

Best regard,

3
General / Re: INS/GPS offset
« on: March 13, 2020, 03:45:27 PM »
OK, I found the information for adaptative modeling:
"Adaptive camera model fitting
This option enables automatic selection of camera parameters to be included into adjustment based on their reliability estimates. For data sets with strong camera geometry, like images of a building taken from all the sides around, including different levels, it helps to adjust more parameters during initial camera alignment. For data sets with weak camera geometry , like a typical aerial data set, it helps to prevent divergence of some parameters. For example, estimation of radial distortion parameters for data sets with only small central parts covered by the object is very unreliable. When the option is unchecked, Metashape will refine only the fixed set of parameters: focal length, principal point position, three radial distortion coefficients (K1, K2, K3) and two tangential distortion coefficients (P1, P2)."

Nevertheless, this does not explain the issue. I just tried with adaptative camera modeling and I still don't have any adjusted values.

I tried with enabling also the "enable reference", beside on the left of "Adjust GPS/INS offset" and I have now adjsued values, but the spare point cloud is worse.... I'm digging into the complex behavior of sfm...

4
General / Re: INS/GPS offset
« on: March 13, 2020, 02:59:15 PM »
Hello Alexey,

I did not use adaptative camera model fitting to have these results.

Actually, what does it mean "adaptative camera modeling"? Does it change the lens model to each image? Not clear to me.
If I use it, will I have a lens model for the group of images with the offset enabled?

Regards,
Seb

5
General / INS/GPS offset
« on: March 13, 2020, 02:04:56 PM »
Hello,

I have images coming from a "PPK" drone and other from a "normal accuracy" drone.

In the camera calibration tool. I enable "Adjust GPS/INS offset" for the PPK images. After the alignment and optimize step, I checked the camera calibration window and I see that the group of images with "Adjust GPS/INS offset" enabled do not have any lens adjustement (cx, cy, k, ...) whereas there are camera lens adjustments for the other groups of images (wihtout PPK and "adjsut GPS/INS offset enabled)!

What I need to do for having both "Adjust GPS/INS offset" AND camera lens calibration?

Regards,

Seb

6
Hello everybody !

I'm working on a multi-chunk project. Here's what I've already done.

By using the batch process, I've align, built dense cloud and built the mesh for my four chunks. Then I've merged all the chunks.
I've build the mesh for the merged chunks.
Now I'm trying to export an orthophoto out of it and always receive this message : "Not enough memory".

I've try to export it at different resolution. I've also try to change boundaries to decresae the size of the orthophoto but I always get the same message "Not enough memory" and in photoscan I'm getting a "bad allocation" message.

My project is compose like this :
Cameras : 4828,4730 aligned
Tie Points : 466,024 points
Dense cloud : 537,705,828 points
3D models : 35,631,307 faces

My computer has 64G RAM (intel Core i7). Graphical card : NVIDIA Quadro K600
Photoscan v1.1.5

Could someone help me ?

Thanks


 

7
Hello,

How can I write a function which returns the projection numbers camera and the projection error of each cameras?

I tryied to build this function based on the code from:
http://www.agisoft.com/forum/index.php?topic=3189.msg16800#msg16800

But It seems to not compute the right projection number and the projection error!

Code: [Select]
def calc_reprojection(chunk):
point_cloud = chunk.point_cloud
points = point_cloud.points
npoints = len(points)
projections = chunk.point_cloud.projections
err_sum = 0
num = 0
photo_avg = {}

photo_info=[[float('nan') for j in range(3)] for i in range(len(chunk.cameras))]#Ajout seb
i=-1
#print("total camera number: ",len(chunk.cameras))
for camera in chunk.cameras:
i+=1
if not camera.transform:
photo_info[i]=[camera,float('nan'),float('nan')]
continue
T = camera.transform.inv()
calib = camera.sensor.calibration
point_index = 0
photo_num = 0
photo_err = 0
for proj in projections[camera]:
track_id = proj.track_id
while point_index < npoints and points[point_index].track_id < track_id:
point_index += 1
if point_index < npoints and points[point_index].track_id == track_id:
if not points[point_index].valid:
continue
dist = calib.error(T.mulp(points[point_index].coord), proj.coord).norm() ** 2
err_sum += dist
num += 1
photo_num += 1
photo_err += dist
#Ajout seb
#print("i:",i)
if photo_num==0:
photo_info[i]=[camera,photo_num,float('nan')]
else:
photo_info[i]=[camera,photo_num,math.sqrt(photo_err / photo_num)]
return photo_info

8
Python and Java API / Re: Align photos with markers only
« on: June 10, 2015, 01:28:51 PM »
Hello,

I am also interested to understand if Photoscan uses markers for the alignment step?

As I understand your answer Alexey, reducing the number of tie points will increase the wieght of the markers?

Sebastien

9
Python and Java API / Export Markers
« on: May 16, 2015, 01:45:59 PM »
Hello,

How can I export the markers with python as what is done with Tools/Export/markers in the GUI? Same question for loading its.

chunk.saveReference() only export the coordinates of the markers in the project but not the location of markers in each images. This is doing the same as Saving in the Reference window!


10
Sorry my T=chunk.transform.matrix

11
Hello Alexey,

I am confused! I do project with makers that I give "local coordonates" (eg: x=1m, y=0.5m, z=1m). These coordinates are geographic coordinates or geocentric coordinate?
I want a region.size of V_s= (5m,5m,20m)
How I convert my vector V_s in internal coordinate to resize my region?

Following your last comment in this post I should do this:
V_s= T.inv().mulp(PhotoScan.Vector([5, 5,22])
new_reg = chunk.region
new_reg.size = V_s
chunk.region = new_reg
Does it right? If yes, I don't get the desired size of my bounding box!

If I do this:
V_s=chunk.crs.unproject(PhotoScan.Vector([5,5,20]))
V_s= T.inv().mulp(V_s)
I have an error message: " AttributeError: 'NoneType' object has no attribute 'unproject'  "

If I do this:
V_s= PhotoScan.Vector([5, 5,22])
new_reg = chunk.region
new_reg.size = V_s
chunk.region = new_reg
The bounding box is alright in some project but not in all! What I am doing wrong???



12
Python and Java API / Re: Script to match Z axis
« on: April 07, 2015, 02:42:17 PM »
Hello,

the script to rotate the bounding box is written in the wiki:
http://wiki.agisoft.com/wiki/Python


13
Python and Java API / Remove 3D points
« on: April 07, 2015, 11:45:59 AM »
Hello,

How can I remove a 3d point?

What does the ".valid" in the class  PhotoScan.PointCloudPoint?

I use it to remove points, but I am not sure it is all right! The points wiht a False value for ".valid" disappear from my GUI but the total number of points is still the same than before to use the ".valid".

Does optimize will use the points with the valid = False?

Thanks,

Sebastien

14
On the other hand, chunk.size is in the chunk coodinate system and not in "internal coordinate system"! This confusing!  ;)

15
Hello,

all is in the subject! When would it be interesting to use this method? What are the editings which might require a refineMatches? Does Keypoints descriptors can be edited?

Sebastien

Pages: [1] 2 3