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 - bigben

Pages: 1 2 [3] 4 5 ... 28
31
I ran a very small pilot project looking at a couple of scanning options for large volumes of negatives.  I only had 4 negatives, but I'd also received a set of 9 images that had been previously scanned, as well as individually georeferenced versions of these. Some early notes here: http://www.agisoft.com/forum/index.php?topic=3790.msg19777#msg19777 and a model created from these https://skfb.ly/F9FE

In the end I used an image template of the fiducial marks to manually position the scans, and a Photoshop action to add an alpha mask and apply radial lightening.
16bit scans worked better than 8 bit scans. Best results were obtained when max,min pixel values were ~90%,10% respectively (much less noise)

The 9 prescanned images I received were also scanned on an Epson and I ran them through the equivalent workflow even though they were only 8bit and exhibited obvious curling on some images.  As they supplied geotiff versions of these as well I was able to redo the georeferencing of the raw scans.  The results I got from Photoscan were better than the supplied georeferenced images.

The biggest problem with the Epson 10000XL (or V700) for large negatives is that you need to find something to flatten the negatives with. Curling negatives can create significant distortion.

32
General / Re: Different scale for different chunks, is that normal?
« on: January 03, 2016, 12:44:58 AM »
Different scale is normal as it's not clalculated unless you provide specific scale information.  Don't know why they didn't align, but unless you have huge numbers of images it's best to use just one chunk

33
No. Lens correction profiles don't include centre offset.

Distortion correction calculates a new distance of a point from the optical centre of the image.  If you have one process that performs this fron the centre of the image file and then a second process performs this from a different position, then the first process has screwed things up for the second one.

The calculation of lens parameters by Photoscan (and all of the image stitching applications I've used) is an optimisation routine, ie. The end result depends on the data you supply it.  Distorted models are caused by the lens parameters optimising to values that are not accurate for the lens. This in turn can be caused by a whole range of factors including inadequate number of images... At different angles... Poor image quality, too low GSD for the expected resolution etc...

You can either use Agisoft lens to calibrate your lens parameters beforehand, or save the lens parameters form a project that has worked well and then load these into subsequent projects.

34
I don't know about unstable, but I have noticed a few odd failures that I've attributed to various aspects of the image sequence.  Taking these into consideration on subsequent shoots has provided much more reliable alignment.

  • I have experienced total failure of the intial alignment when the first few images did not contain any overlap between them but did have substantial overlap with many images later in the sequence. Theoretically this shouldn't be a problem, but ensuring that the first few images contain a lot of overlap avoids this.
  • Normally I shoot in such a way that each image overlaps the previous one, but occasionally I'll move to another part of the scene and start shooting again, creating a gap in "continuity".  If the second sequence does not overlap any of the previous image sequence (with images later in the sequence providing overlap) this may cause the alignment to fail either for this second sequence or from this point onwards through the entire project...  Again, theoretically this shouldn't be a problem, but making sure that first few images of the new image sequence overlaps the previous image set seems to avoid this scenario.
  • Significant jumps in GSD can result in the failure of individual images to align

In your case you mention regions of the images being out of focus. This may result in inadequate overlap between images locally within the sequence which may flow on to the rest of the sequence.  Manually aligning them afterwards often works but in some cases the alignment may be off so it pays to check the pointcloud where the new images have been aligned. 

For boulders, I'd only mask out the sky if visible.  Extra background detail may provide the tie points you need to align the images with out of focus sections.

This point cloud (http://files.digitisation.unimelb.edu.au/potree/pointclouds/ricketts-test2.html) was processed with a single alignment step with only one image failing to align (turned out to be motion blur)...  but then I also use a full frame fisheye which provides ample overlap ;)

35
General / Re: using symmetry to render a solid object
« on: December 30, 2015, 03:31:43 AM »
You'd need to do that in another application.

36
General / Re: Aligning single flight line, i.e. Pipe or road
« on: December 30, 2015, 03:29:56 AM »
The biggest problem will be from drifting lens distortion parameters during optimisation.  If you have calibrated lens settings (or reuse lens settings from another project that aligned very well) then fixing the lens parameters will help avoid the usual problems with a single camera path

37
No. Lens correction profiles don't include centre offset.

38
General / Re: How to fill holes and build accurate textures
« on: December 29, 2015, 01:08:50 AM »
 A lot will depend on the detail in the images as far as creating a mesh is concerned.  For hole filling I usually export the mesh, fill the holes in another program, e.g. Meshmixer or ZBrush and then import back into Photoscan to create the texture.

39
General / Re: Memory Problem
« on: December 24, 2015, 11:37:55 AM »
There are alternate methods of creating a mesh from the pointcloud depending on the level of detail/polygon count you need. eg.
http://www.agisoft.com/forum/index.php?topic=4007.msg20728#msg20728

40
General / Re: Drone Custom Camera Specs
« on: December 05, 2015, 01:41:11 AM »
What 3 lenses would you want it to ship with or have available?

Depends on what it's being used for... Working distance, desired GSD

41
General / Re: Photo Align problem
« on: December 05, 2015, 01:37:35 AM »
After the alignment stage you can select the unaligned images, right click and select align. This may work, but if there are problems with the mages it may not either.

42
General / Re: Stereo Pairs
« on: December 04, 2015, 01:44:09 PM »
Mainly saves you time as you're capturing images faster. Doesn't matter whether the cameras are synced or not.

43
And shoot time lapse, not video. 0.5 sec per frame and walk slowly.  then you have 12mp images.  Samsung nx3000 and 10mm lens would be even better for the money

44
General / Re: Upload to Sketchfab - no texture?
« on: December 02, 2015, 01:08:59 PM »
Export an OBJ model with texture.  If there's no texture then the problem is not with Sketchfab. 

45
General / Re: Stereo Pairs
« on: November 30, 2015, 03:44:48 AM »
It depends on what you're shooting and how you use them.  A stereo setup isn't going to get you much benefit in its normal configuration, but using 2 synced cameras can cut down the time it takes to shoot a scene.  2 cameras mounted on a vertical pole 1-3m apart would be useful for outdoor scenes, although I'd use wider lenses.

Pages: 1 2 [3] 4 5 ... 28