Topics

1

General / Changing filters used to find points of interest for alignment

« on: February 26, 2019, 04:53:38 PM »

     I want to experiment with Sobel, Prewett, and Canny filters used to find points of interest during alignment.  Is there a way to do this in MetaShape?

2

General / Intel GPU Maxed Out, i7 CPU and Nvidia GPU less that 1%

« on: December 27, 2018, 08:13:16 AM »

Hi.
I am using an old HP Envy 17 gamming laptop for photogrammetry.  I noticed that during the alignment phase that the i7-4700MQ quad core CPU is running at 1%, the Nvidia GT 740 GPU is running Cuda at 0%, and the Intel HD Graphics 4600 running OpenCL is maxed out at 90%.   Have I set my GPU preferences incorrectly?

3

General / Importing GPS Position and IMU Orientation data into photoscan

« on: July 14, 2018, 06:07:16 PM »

What is the best approach for importing GPS Location and IMU Orientation data into Photoscan?

4

Feature Requests / Implement Lens Distortion model for Zoom Lens

« on: December 31, 2017, 05:02:55 PM »

Everything I read on this forum implies that the implementation of "Brown's" Lens Distortion model used by Photoscan can not handle zoom lenses.  Have you tried to implement the Z-D calibration modification described in "Zoom-Dependent Camera Calibration in Digital Close-Range Photogrammetry" by C.S. Fraser and S. Al-Ajlouni out of the University of Melbourne in Australia?

5

General / Manually placed Markers not working

« on: September 15, 2017, 07:02:45 AM »

I am frustrated by the lack of effectiveness of manually placed markers to help in image alignment!  I have over 300 photos of a car that I have masked out the background and placed almost 70 manually placed markers at well defined locations.  Frequently there are 5 or more markers on each image.  I tried setting the max number of key points to 1, 5, 10, and 20.  Alignment was awful.  I broke the images into 5 chunks ensuring common markers across chunks and could not align the chunks. 

I have to be doing something wrong!  Where is a good tutorial about aligning images manually

6

Feature Requests / Import camera positions and orientatons to initalize camera alignment

« on: June 29, 2017, 02:37:44 AM »

Please import AGIsoft.xml or use image Exif to initalize camera position and orientation for alignment.  I would like to have the option to fix the orientation and position during the alignment phase...just like the lens distortion coefficients.  I do not know why I cant find this feature request...there are many posts asking for the ability to import orientation into Photoscan in addition to position GPS or static camera rigs...but I can't see Agisoft actually uses orientation in their alignment.

7

General / How to align cameras only using markers?

« on: April 23, 2017, 06:07:11 PM »

Is there a way to ignore the sparse cloud points and align cameras using only markers?

8

General / Recommended Gradual Filter Settings

« on: April 15, 2017, 07:56:56 PM »

I don't know about the rest of you, but the documentation of Gradual Filter Settings does not clearly describe the meaning of the parameters, units, or recommended values...at least a place for a newbe to start would be nice!  There are some hints elsewhere in this forum, but nothing direct and to the point.  What do you think about including a table like this:

Parameter                                  Units    Value                             Description
Reprojection error                    pixels     0.2       Statistical error in point placement
Reconstruction uncertainty     pixels      10       Possible variation in point placement
Image count                                  -            4         Maximum number of images having cross matching points
Projection Accuracy                 pixels      10       Accuracy of point placement from local neighbor points


Maybe as a community we can establish a place to start for the newbies (like me) out there.  Please review and correct as you see fit.

9

General / Should I Buy DxO Optics Pro

« on: April 02, 2017, 07:27:33 PM »

I am seeing conflicting opinions in my search about preprocessing RAW images in DxO Optics Pro before running in AGISOFT PhotoScan. Does it improve or hurt results?  What settings should be set in DxO Optics Pro to get best results in PhotoScan? What is AGISoft's official judgement on this?

10

Feature Requests / Support for DNG format in AGIsoft Lens

« on: March 27, 2017, 04:08:34 AM »

I tried to use dng files in AGISoft Lens but they would not show up in the photo selection dialog because the extension isn't listed.  Does file format affect the lens calibration? Do I have v1.3?

11

Feature Requests / Exporting Photo pane Details as text file

« on: August 14, 2016, 06:40:05 PM »

I have been doing some experiments trying to figure what the optimal camera settings are for maximizing photoscans image quality index.  I am spending too much time translating the Image quality data into excel.  I would really like to have the ability to export the Photos pane detailed data into a csv file for Excel to process.

12

Feature Requests / Paring photos for alignment like dominoes

« on: August 14, 2016, 06:29:36 PM »

I experienced the same ability to manually improve the photo alignment produced by Photoscan as RalfH did in his post "sequence of photo alignment".  In fact, I even tried to reset alignment on all but a few photos with Photoscan and then manually re-aligned the rest of the photos I thought should  be aligned for better results.  As a consequence of this activity, I found myself wishing Photoscan's Photos pane would display thumbnails of the photos with green lines between strongly matching photos and red lines between weakly aligned photos.  I would like to be able to re-arrange the photo thumbnails drawing black lines between pairs I think Photoscan should pair and breaking links that I think are bad paring....kind of like playing dominoes.

13

Feature Requests / Recommending a RAW to TIF Conversion Utility, Image Preprocessing

« on: January 10, 2016, 06:55:05 PM »

I tested three CR2 (Cannon RAW) to TIF converters and got very mixed results.  My results as measured by the AGISoft Image Quality Index varied by a factor of 3 at the high end and didn't vary much at the low end.  I ended up staying with Cannon's own "Digital Photo Professional" because your Image Quality Index rated it the best.  Seems to me, you should at least recommended a RAW conversion utility and settings for your users that you think gets the best results in Photoscan or support reading RAW formats directly from within Photoscan. (The later is preferred)

Based on other posts on this forum I get the impression AGISoft prefers the TIF format....What about using image filtering to improve contrast (Wallis filter) and do edge detection (Canny, Kirsh, or Sobel) to help find features?

14

Feature Requests / Visually Validation of Camera Alignment

« on: January 05, 2016, 09:47:54 PM »

I have too many cameras looking at an object from random directions.  I want to be able to step through each camera in the Workspace displaying the model overlaid on top of the camera image as the background from the same view point.  It would then be visually obvious when that camera is not aligned properly and would enable me to easily realign that individual camera (or at least try to....)

15

Feature Requests / Use Space Carving to Initalize Mesh

« on: December 31, 2015, 09:40:41 PM »

Ok....we already did all the work to place markers on our photos for alignment.... and masked out the "uninteresting" parts of the images.... Why not use those mask boundaries to build an initial mesh of the surface using a space carving algorithm? Then you could refine the initial mesh using the methods already implemented in PhotoScan?  (see "A Theory of Shape by Space Carving" by KIRIAKOS N. KUTULAKOS and STEVEN M. SEITZ)

Can you point me to a reference describing the algorithms details or Computer vision libraries used in Photoscan?  Do you have any guidance as to how I should go about implementing this Space Carving idea using your python script interface in PhotoScan?  I am especially concerned about accessing the mask data and providing the initial mesh to the rest of the PhotoScan workflow.