Messages

31

Camera Calibration / Re: Adaptive camera model or calibration groups

« on: August 07, 2022, 10:02:17 AM »

Is there a way to select or sort photos by speed and direction of drone flight so that each group can get different rolling shutter corrections? I can't even find a way to view the metadata that shows speed

I don't think so, except maybe through python scripting, but what might help is that you can select photos using the 3D/Model tab or normal photos pane, and then open the camera calibration window, and any photos that were selected before will now be selected in the list at the bottom of the calibration window, though you may have to scroll down to find them.

Still doesn't help you filter by speed or direction, but it may be easier to identify and select these photos intuitively in the 3D view before you go into camera calibration.

THANK YOU, that helps a lot because the gallery view makes it easier to go through and find the ones where they're in a sequence so you can tell it was with the drone moving

32

Camera Calibration / Re: Interior camera calibration based on single image and many GCPs

« on: August 01, 2022, 04:23:07 AM »

Yes, I would mark the control points in every photo they appear in, then optimize or run a new camera alignment. Then save the camera calibration to CSV file and you can load it in later projects. When you run the photo align, the Agisoft uses that calibration data but will also adjust it for the new new project

33

Camera Calibration / Re: Adaptive camera model or calibration groups

« on: July 31, 2022, 03:19:46 AM »

In the lower right of the 'Camera Calibration' window is a list of all the photos in each group. You can select the photos you want to move to a new group, right click and select 'Create Group'.

Thank you, I never would have found that. My workaround was to split the group on the left window which would make each photo its own group, then select on that side and group them.

I have another question. I have an Air 2S and I fly manually and often have different speeds because of wind. Today I even flew backward for many photos because the drone kept losing signal when I turned it. Is there a way to select or sort photos by speed and direction of drone flight so that each group can get different rolling shutter corrections? I can't even find a way to view the metadata that shows speed

34

Camera Calibration / Re: Adaptive camera model or calibration groups

« on: July 20, 2022, 02:59:36 AM »

How do you calibrate groups separately?

I have some photos flown at 15 meters per second speed and some at 5 meters per second so I want different rolling shutter correction.

I have them in two different groups. But there seems to be no method to set different calibration. In the Camera Calibration dialog I can only split the group so each camera is its own group

35

General / Re: Loosen up requirement for matching with manually places markers

« on: July 01, 2022, 08:37:22 AM »

Place your markers and see what happens. There's no requirement, I've had ground control points on only single photos

36

General / Re: How to rotate mesh or point cloud?

« on: June 10, 2022, 11:10:22 AM »

I learned from this thread that there's a Rotate Object tool under Model > Transform Object > Rotate Object

37

General / Re: Rolling shutter compensation in 1.8 does not correct all parameters

« on: June 10, 2022, 11:01:36 AM »

I recently tried loading camera calibration data exported from another project where I know there was no rolling shutter effect. Then I checked the rolling shutter compensation box. I figured this way the program would have the full calibration data, but the adjust the rolling shutter parameters only. The results seemed good

38

General / Re: Waving a white flag

« on: June 10, 2022, 11:00:04 AM »

GCPs do turn blue if the program is confident about their locations. You may have to place quite a few before some start turning blue

39

General / Re: Perfect settings to process data having fine geometry

« on: June 07, 2022, 10:29:47 AM »

If dense cloud is not good(without necessary details and with unwanted noise), then mesh/geometry can not be fine. Did you try ultra high quality dense cloud?

Did you try build mesh from depth maps only?...you should get better mesh result.

Are you sure about depth maps being better? I've only gotten much worse results from depth maps

40

General / Re: 2-D image stitching

« on: June 07, 2022, 10:04:30 AM »

Pretty sure that's not possible but what you have to do is export an orthomosaic. You have to somehow get the program to understand that the XY axes are the surface of the wall and then your orthomosaic will be a flat projection of the surface

41

General / Re: Directional vector from yaw, pitch, and roll?

« on: June 06, 2022, 08:20:07 AM »

I'm pretty sure the pitch, roll, and law only help the photo alignment process. If they're aligned properly then they don't matter anyway.

If you have blurry spots on your orthos and textures, it's probably because the program is choosing further photos or edges of photos to fill those spots. You have to draw polygons, right click, and then assign a different image. I'm hoping they update Agisoft to let you prioritize photos that are closer to the area in question

42

Camera Calibration / Re: Placing markers increase cameras error after optimization

« on: June 05, 2022, 08:09:35 AM »

Now I can safely say that what I said is correct. The model, once it has accurate GCPs, then can figure out how inaccurate your camera GPS locations are. If you give it only camera GPS, it doesn't know how bad they are

43

Camera Calibration / Re: Quick Question - RTK/PPK precision, drone image position accuracy

« on: June 05, 2022, 07:38:18 AM »

Naaa, I'm taking that back. watched the linked yt-webcast to the end now. So ground control points CAN help to reveal possible z-value errors in the rtk-data, if you messed up something I did not fully understand yet.
Takeaway: doublecheck your data... 

The cameras are positioned in 3D space using the ground control points as an imput. It's not just for revealing Z errors. RTK GPS error isn't the only thing you have to worry about, also the angle of the cameras. If a camera is 1 degree off from nadir, that alone adds a huge amount of error.

44

Feature Requests / Re: Prioritise images taken from closest distance during texture generation

« on: June 04, 2022, 08:47:19 PM »

Another way to improve this would be to allow selection of a seamline polygon and change the assigned photo of that, rather than requiring drawing a new polygon

45

Feature Requests / Re: Standard Plus (i.e. Medium) version please!

« on: June 04, 2022, 09:21:10 AM »

I was able to get an educational license by doing free modeling for the university! But I still think a cheaper "plus" version would be good