Messages

76

Python and Java API / Re: orthomosaic functions projection plane, max dimension, estimate region, export

« on: November 02, 2019, 05:20:16 AM »

Could someone please give an example?
Maybe for "max dimension"?

Thank you!

77

Python and Java API / Re: Python Script for Rotating the Region

« on: December 04, 2018, 12:12:45 AM »

Hello tttjy3,

I think we have the same question...
Could you manage it? 

78

Python and Java API / Re: Rotate chunk by the orientation of markers

« on: December 01, 2018, 06:27:55 PM »

Hello Alexey 

Is it possible to do the same with a set of markers on the ground plane? (no vertical markers)

Let's say we have markers on the ground, placed like an "L".
How can I change the script above so that the ground plane is rotated well? (all markers on the gnd)


Thank you very much! 

79

Python and Java API / orthomosaic functions projection plane, max dimension, estimate region, export

« on: July 22, 2018, 12:34:15 PM »

Hello Alexey! 

For a python script I am searching the following functions for "build" and "export" orthomosaic:
I really tried to find them out but without success. So please tell me how I can (screenshots below):

build orthomosaic...
- projection plane: marker (e.g. hor points 2 - 3; vert points 3 - 4)
- projection plane: Top XY
- estimate the region size
- set the max dimension (pix)

export ortho...
- with max dimension (e.g. 8000)
- split by 4000 x 6662
- jpg 90%

rotate model by points
- It also would be very helpful to rotate the model by the given points, so that the orientation (projection plane top XY) is the same for ortho export and the mesh. Can you help with that too?

Thank you very much for your replies in the past! 
Best regards

80

General / Re: Two different GPUs : does Photoscan allow it / performances expectations

« on: February 27, 2018, 10:19:10 AM »

Hello everybody,

I totally believe that PhotoScan uses both GPUs. But what about the speed of the calculation with both GPUs?
Does the slower GPU, in this case, brake the system down?

Any experience?

Has someone tested such a hardware setup and switched one of those two GPUs off/on?

Thank you 

81

Python and Java API / Re: Add known Pitch/Roll/Yaw to Exif and Auto-Rotate Model

« on: January 01, 2018, 01:08:32 PM »

Hello Alexex,

thank you for that! I can now manipulate pitch in that way. 
But it's still unclear if roll is seen in PhotoScan as a value of 0-360 or -180 + 180 but I try to figure it out.
Is YAW a Value of +-180 or 0-360?

What about question 4? Is it possible to use rotation values only to autorotate a model or does PhotoScan need Location for this? (Photos inside a building...no GPS but Orientation)

THANK YOU! 

82

Python and Java API / Re: Add known Pitch/Roll/Yaw to Exif and Auto-Rotate Model

« on: December 30, 2017, 09:22:29 PM »

Hello Alexey,

1) Do I understand correctly...?
- pos. pitch = nose up
- pos. roll = left wing up
- pos. yaw = turn from north to east (to right)

2) I guess the min/max values for pitch, roll, yaw are different...
- Pitch: Level=0°; -90 (nose down) to +90 (nose up)
- Roll: Level=0°; +-180° ... Roll left wing down = -90°
- Yaw: North = 0;   East=+90;   South= +-180;   West=-90?
Or is there a different convention?

3) What is wrong with this example?
- All the photos of this chunk are shot at yaw= -55°, roll=0° and pitch= -90° (nose down)
- I've added these information. After aligning the models rotation is not how expected..."Top View ... NW -55"
What is wrong?

4) Clear all / clear rotation
I just noticed that I can clear all or just rotation in reference menu - but there is no function to clear location only.
Is it possible to use rotation without location to auto-rotate a model?

THANK YOU!!!! 

... just saw that one picture is named incorrectly. The Rotation wasn't cleared in picture "Top View cleared NW -55.jpg"

83

General / Re: GROUND ALTITUDE option meaning in the reference setting panel

« on: December 30, 2017, 06:31:35 PM »

Hello Alexey,

this is from the user manual:
" For oblique imagery, it is necessary to set Ground altitude
value (average ground height in the same coordinate system which is set for camera coordinates data) "

Let's say my drone is flying at 20 m above a flat surface and the (average) GPSAltitude of my drone is 300 m MSL.
Should I set the ground altitude to 280 m if the coordinate system is WGS 84?

Thank you!

84

Python and Java API / Re: Add known Pitch/Roll/Yaw to Exif and Auto-Rotate Model

« on: December 29, 2017, 12:23:38 PM »

That was quick  Thanks for that!

Some questions:
1) Are these orientation values used only when aligning cameras with reference preselection?

2) How does PhotoScan expect Pitch & Roll ...
      -90° pitch is down?
      -90° roll is rolling left?
      -90° yaw is west?

3) Where can I store the accuracy for yaw, pitch, roll?

4) Where can I store my own better GPS coordinates and the better accuracy for them?

5) Do I need to update something after adding my values or just align my cams?

THANK YOU SO MUCH!

85

Python and Java API / Add known Pitch/Roll/Yaw to Exif and Auto-Rotate Model

« on: December 29, 2017, 11:35:33 AM »

Hello Alexey,

I've been programming with python in the last 2 weeks and created some nice personal tools for Photoscan.
Now I want to write some code which should do the following:

I know some "correct" values of my cams orientation from my selfmade IMU. These are:
  - yaw+pitch+roll [degrees]
  - altitude [m]
  - gps coordinates
  - accuracy for each value

Of course I want to let PhotoScan process these values to have (pre-)scaled but well rotated model.
How can I import these values to my cams? 
I don't know where to store Pitch,Roll,Yaw ...

To give you an example:
Please see the EXIF of a MavicPro-IMU attached. I guess the orientation PhotoScan reads from EXIF are not the ones it should read - am I right?
I can access Pitch_Value = chunk.cameras[camnum].photo.meta["DJI/GimbalPitchDegree"]) but I don't know where to store these values and how to tell PhotoScan to use them for alignment,

Please let me know if you can help me with that! Thank you!
All the best for the new year! 

86

Python and Java API / Re: Fisheye sensor - setting type

« on: December 20, 2017, 06:50:55 PM »

Hello Alexey!

Sorry, but I don't understand how to fix this problem eventhough you wrote

Code: [Select]

"sensor.type = PhotoScan.Sensor.Type.Fisheye"If I add this code in my skript I get:

Code: [Select]

[i] ...line 9, in <module>
2017-12-20 16:44:07     sensor.type = PhotoScan.Sensor.Type.Fisheye
2017-12-20 16:44:07 NameError: name 'sensor' is not defined[/i]
Can you give me a full example code for changing the sensor type to fisheye (if the camera name is GoPro)?
Thank you!!!

87

Bug Reports / Re: Frozen Project while Align Photos

« on: October 24, 2016, 09:37:44 AM »

This took a while ... 0 errors.

CPU activity is different.
Sometimes all cores do nothing (alignmet was completed but project frozen)
But I think that I have seen an example where each core seemed to be active (project at work but frozen).

When I reduce the amount of cameras (disable every 2nd. camera) it sometimes "solves" the problem.
Also reducing alignment parameters (medium oder low , generic, 4000, 1000) can help.

What can I do? : )

88

Bug Reports / Re: Frozen Project while Align Photos

« on: October 21, 2016, 12:22:43 PM »

Thanks for the quick reply!

This happend for the first time some weeks ago with a different project.
It feels like the problem occurs more frequently - and on any projects.

I reinstalled Photoscan and updated graphics driver...

89

Bug Reports / Frozen Project while Align Photos

« on: October 21, 2016, 11:55:34 AM »

Hello everybody!

My problem is a frozen project when aligning photos.
Sometimes this occurs while processing, sometimes when finishing. (See Log below text)

I've already had projects with 2000+ images (same cameras, same computer),
that were aligned perfectly - even on setups like:
Accuracy: Highest
Pair pres.: Disabled
Key limit: 50,000
Tie limit: 0
Constrain features by mask: yes
Adaptive camera model fitting: yes

Computer:
Win7 64bit SP1
i7-5820K
32GB RAM
GeForce GTX 980 Version 372.90 (2.11.4.0 GeForce Experience)
SSD Samsung 850 EVO 250G
Energy management: highest
Free Disk Space: 99GB

Cameras:
GoPro 3+ black
GoPro 4 black
(12MP Wide)

Setup:
Photoscan 1.2.6 build 2834 (64bit)
Camera Calibration: Fisheye
Default View: Point Cloud
Active CPU cores: 11/12
OpenCL devices: GTX 980
------------------
Accuracy: Medium
Pair pres.: Generic
Key limit: 8,000
Tie limit: 0
Constrain features by mask: yes
Adaptive camera model fitting: yes
Cameras: 408
-------------------------------------
I hope someone can help on this...
THANKS ALOT

90

Python and Java API / Re: how do you limit reprojection error and reconstruction uncertainty

« on: September 02, 2016, 11:57:14 PM »

This seemed to be the most relevant search result I found...  I'm beginning to look at scripting workflows and this is one that I'd really be interested in as well (gradual selection of sparse point cloud and deleting selected points).

That's exactly what I need. Is there a solution?