Messages

31

General / Re: Clarification of Reconstruction uncertainty

« on: June 21, 2014, 06:40:12 AM »

Thanks marcel, there is some useful info in that window, although it only shows reprojection error (mean and max). I'm not really concerned with the averages of the errors in the point cloud after camera alignment, I'm looking at how many bad points I can get rid of so I'm looking for the distribution of the error values.  e.g. for one of my GoPro image sets I end up removing 90% of the sparse point cloud.

Reprojection errors seem to be more evenly distributed throughout the point cloud whereas reconstruction uncertainty seems to be more localised.

32

General / Re: Clarification of Reconstruction uncertainty

« on: June 21, 2014, 06:00:45 AM »

I've been experimenting some more and have a few ideas for some more tests to try and find optimum values for producing the best quality sparse point cloud.  Is there a way of accessing the error values of points in the sparse point cloud?  If it's not easy I can always use gradual selection to get the numbers manually.

I still have to run some definitive tests but on appearance it looks like the best option would be to align photos with a point limit of 0 to get everything, and then use gradual selection to reduce the sparse point cloud. I've been using reprojection error = 0.5 and reconstruction uncertainty = 50 as mentioned earlier.  Using high values for the point limit in the align photos sometimes resulted in a much smaller sparse point cloud after gradual selection and I'd chicken out and use values of 0.7 and 70 respectively to increase the number of points left. Starting off with all available points let me go back to my preferred values for gradual selection.

I'd still like to understand how the numerical value of reconstruction uncertainty relate to the quality of the point. It certainly appears to be an important component of cleaning up the point cloud. Points removed (following a reduction using reprojection error) seem to be mainly distant points, errors near the edge of images and/or points appearing in fewer cameras.

33

General / Re: 360° camera - PhotoScan compatibility?

« on: June 21, 2014, 05:24:37 AM »

I'd say no.  Even if you could save individual photos from the 3 cameras, the usable image portion would only be 2/3 at best and the resolution would be very low.  4096x2048 for a 360x180° isn't really "HD" even for panoramas. You would require 4 Gopro cameras for a similar setup and that camera is borderline for usability for photogrammetry.

34

General / Re: How far can a GoPro go...

« on: June 20, 2014, 09:00:30 AM »

1. That seems to be the case for the + model from everything I've read, which is why I bought mine (although a little less than radical ) It is pretty good on closeups for a fixed lens.

2. AFAIK there is no way to set a slowest exposure time or ISO. It's a problem for stills from a bike as well, but less of a problem for video because the higher frame rate requires a faster shutter speed.

I wouldn't buy one specifically for photogrammetry.

35

General / Re: Newbe question: difference between Standard and Pro edition

« on: June 19, 2014, 11:54:38 AM »

As marcel suggested, if you can share some images here then others may be able to help you identify the problem and/or suggest settings which could improve the mesh/point cloud quality.

I'm using the standard version at the moment and am looking at a workflow for scanning entire objects.  For a high resolution mesh I take 3 sets of images with the object in different orientations, generate the point clouds for each set in Photoscan, crop the point clouds and then align, merge and generate a mesh in Meshlab and edit the resulting mesh in Meshmixer if required.

e.g. https://sketchfab.com/models/4bb8f7abd0a84a36be37a81feb9732f2

With a DSLR I can give our NextEngine laser scanner a run for its money.

36

General / Re: Newbe question: difference between Standard and Pro edition

« on: June 18, 2014, 10:36:57 AM »

The pro edition has extra features for aligning cameras using control points in the images to help align things. These are also used to provide information for setting the correct scale. Inserting a scale reference may be useful for visual interpretation but the model should also be scaled accurately so that measurements can be made. This is much simpler with the Pro version.

There are some basic mesh repairs functions but you can also export the mesh for editing in other applications and then import the new mesh back into Photoscan to continue the workflow.  I use Meshlab and Meshmixer for editing meshes (and sometimes creating the mesh).

The quality of the mesh will depend on some of the setting you use in the workflow, particularly removing bad points from the sparse point cloud before generating the dense point cloud. You can also export the dense point cloud and create a mesh in another application such as Meshlab.

For the PDF you could add a scale reference to the mesh in another application and then create the PDF.  the PDF model is just an embedded U3D object file which can be added with Adobe Acrobat.

37

General / Re: Aerial Photogrammetry Phantom 2

« on: June 18, 2014, 02:01:50 AM »

At the end of the day the critical thing is the camera, the details of which are a bit sketchy as there are no specs on the lens.  The intro video mentions 3 fov settings so I'd suspect that they're just crops of the full ultra-wide image. If that is the case you could be down to 6mp by the time you get to the "narrow" fov. It would be nice if they provided some samples of raw images.

38

General / Re: Can agisoft generate something like this?

« on: June 17, 2014, 05:43:16 AM »

For the delivery side of things they're using a pretty good javascript library to stream the model in a similar way to streaming high resolution images...
The model is broken up into square regions so only the 3D data required to fill the viewer is downloaded and there are 2 (at least 2) copies of the model at different resolutions so that you don't have to download a lot of high res sections when zoomed out.  It's not so much a question of whether Photoscan or Acute 3D can do this... they can both generate the source models for this, but what subsequent processing and scriptiing is required to deliver a model in this way.

Maybe something like this: http://www.pheelicks.com/2014/03/rendering-large-terrains/

39

General / Re: Bracketed exposure models

« on: June 14, 2014, 05:21:04 PM »

That one is just the normal exposure as a reference. Didn't make as much difference as I expected, although there are differences, but i haven't uploaded anything yet. Still a few other things I need to understand. I've specifically avoided HDR for now because I wanted a handheld shooting workflow, mainly because i know the people I'll be passing this on to won't have the patience (or possibly time) to be extra careful.

but so far:

Trial 1: Separate chunks for each exposure, dense point cloud, aligned, merged and meshed. Some minor differences but nothing really significant.

Trial 2: All images in one chunk. Aligned photos resulted in fewer points, but sparse point cloud was quite dense in some areas, picked up more detail in trees off to the right of this model (in shade) but lost a lot of the side wall on the left. Not sure if it just ran out of points. I'll have another look at that later.

Trial 3: Similar to trial 1 but building dense point cloud after merging chunks (currently running)

I was shooting with aperture priority but some of the normal frames were a little dark for my liking.  I'll have to take a bit more time for the next shoot and pay more attention to the exposure.  The building that's inspired the test is also my test subject for HDR stuff because the exposure drops significantly in the corners (6-8 stops within 8m) https://www.flickr.com/photos/ben-kreunen/4053467313/ so shooting on manual and just bracketing a single 2 stop increment isn't going to be practical.

Masking the over-exposed image set is possibly quicker than HDR merging as the batch action is pretty quick... but I'll wait until I get a result worth sharing before I call it. The masking did remove the extra white noise along the edge of the roof/chimneys

If this doesn't work the way I expected that will still be worth knowing.

40

General / Bracketed exposure models

« on: June 14, 2014, 03:37:01 AM »

As I mentioned in an earlier post I was thinking of experimenting with bracketed exposures for shooting buildings, with the view of filling in holes in the darker recesses of the building with points derived from images with higher exposures.  I'm not entirely sure whether this is sensible or not, but I shot a quick test yesterday and the result was quite promising.

For now I'm processing the different exposure sets separately and then aligning and merging the point clouds. I'm only shooting one extra exposure (+2 EV). The first test added some extra noise along edges but I only did a quick mask and I can see why this happened. I've set up a separate Photoshop action to mask these images, excluding everything at the top end of the pixel value range (starting with a max of 245 but that's just a number I pulled out of the air)

The initial model from the normal exposures is here: https://sketchfab.com/models/d24123350ae2440086856d1c17193667

41

General / Re: GeoTIFF file, is it possible to edit it after agisoft?

« on: June 13, 2014, 06:38:10 PM »

OSGEO4W, Globalmapper. Basically any application that can georeference images.

42

General / Re: How far can a GoPro go...

« on: June 13, 2014, 02:47:25 AM »

But today I generated a mask in Photoshop and applied it to all images in order to get rid off all the blurry areas around the edge of the images (see attached png-file, works for GoPro Hero 3 black images with 12 MP, WIDE setting).

If your mask gets too close to the one attached here, just switch to 7MP, medium. I ran a test and confirmed that 7MP medium is just a 12MP wide cropped.

43

General / Re: Worse sparse cloud with higher aligment setting.

« on: June 13, 2014, 02:43:00 AM »

I'd be interested to know the range of Reconstruction uncertainty values that people are getting for 20k and 40k, where the 40k sparse cloud looks worse

44

General / Re: Worse sparse cloud with higher aligment setting.

« on: June 12, 2014, 12:16:58 PM »

Did you try gradual selection on the sparse point cloud?  I've had better results from using more points and weeding out the bad ones before generating the dense cloud.

45

General / Re: RAW, JPG and different types of noise

« on: June 11, 2014, 07:12:06 AM »

On the exposure side of things, has anyone done any experiments making multiple point clouds with different exposures and then merging them? I'm toying with a couple of ideas for a building and don't necessarily want to go down the full HDR/tone mapping path which would require precise image alignment of exposure groups during shooting.