As of January 2, 2016 all georeferenced Maps Made Easy jobs will have an overlap report image that is generated at the time of processing. The overlap report is available at the bottom of the Map Detail page.
Overlap Reports can be used to diagnose 95% of issues that users experience when they are unhappy with their outcomes.
_______________________________________________________________
An Overlap Report is in image that shows the user how many images contain a view of any location within the processed survey area. Generally, the more "looks" at an object the system has to work with, the better the outputs will be (within reason). The Overlap Report color codes these numbers to let users know when the system considers to be good or bad overlap coverage and helps them visualize the things that affect the collection of consistent overlap.
1, 2, 3, 4 or 5 views (reds) of an area is considered to be totally inadequate for the purposes of reconstructing a 3D scene. 6, 7 or 8 views (greens) is considered adequate for the reconstruction of fairly simple objects like fields, basic structures or stockpiles. For the greatest ability to reconstruct a complex scene, getting 9 or more views (blues) is required to have the detail necessary to reconstruct complex objects like buildings, trees or other vegetation.
A red dot at an imaging location denotes that there was an image that was processed but it did not have sufficient overlap and/or matchable features so it was not included in the output.
The following examples are provided to help users see how various scenes will affect the achieved overlap. As always, our Map Pilot application for iOS will help to manage the basics but knowing about your survey area's terrain and contents will go a long way towards collection proper data.
Good Overlap
This is an example of good and consistent overlap. There will usually be some red areas around the edges that may not turn out perfectly but everything in the interior of the survey area will turn out very well as it is shaded mostly dark blue and blue.
Overlap Settings
The move overlap you have the less gaps there will be in the dark blue sections of the Overlap Report. A fully dark blue overlap report is a good goal. We generally recommend using 80% overlap as a starting point. 80% is the default value in Map Pilot Pro.
60% Overlap - 5-8 looks at everything - No blue, might work on the most simple scenes. Maybe.
70% Overlap - 8 to 15 looks at everything - Areas of blue but inconsistent.
80% Overlap - 18 to 20 looks at everything - Consistent coverage that makes successful maps.
Bad Overlap
50% overlap is barely enough to work. Consider this the bare minimum. This was taken over a flat and feature rich area with no trees. Any complex features in this survey area got mangled.
These are example or insufficient and inconsistent overlap that was collected using free flight and the camera intervalometer. Overlap like this will result in bad edges, large gaps, badly distorted imagery and a flat 3D model reconstruction except for in the green and blue areas.
Gaps in Imaging
This shows that the overlap is mostly good but the lack of black dots, which represent the imaging locations, in various areas long the passes will result in poor reconstruction in those areas. Gaps in the imaging interval are caused by loss of line of site when using the Map Pilot app.
Elevation Difference Overlap Reduction in Structures
The structure shown in the colorized DEM above are roughly 17 meters tall. When flying relatively low at 40 m, as was done for this survey, that will correspond to a significant reduction in the overlap achieved on the tops of the structure. Being that the structures were likely the target of the survey, this is not good... This is one of the main causes of the imagery being "swirly" on the tops of structures. This particular example worked out because it was feature rich. Flat roofs will not likely be so lucky.
Elevation Difference Overlap Reduction in Terrain
The terrain on the right was roughly 50 feet higher than the terrain on the left. For the particular example, the cruising altitude was 130 feet above the starting point, which was about 30 feet below the high point. This resulted in significant overlap loss as the higher terrain crept up closer to the cruising altitude of the aircraft.
Too Much Overlap is Inefficient
While too little overlap will result in sub-par results, having too much overlap is inefficient with our system resources and the user's processing points. There is no evidence that having this much more overlap will help produce higher quality outputs.
Overlap Reports are a great tool for users to learn about the 3D reconstruction process and debug their own data collection processes.
Very helpful, thanks.
Any chance of making the overlap reports a downloadable geotiff (raw, with data value = #overlaps at location, or NODATA)? Or at least a companion world (georeferencing) file for the rendered overlap.png that shows up on the website? Would be helpful for those of us doing our own postprocessing.
This isn't something that we will likely add since it is there as a guide to show the user what overlaps were achieved using our system specifically. You could just download the PNG file and manually reference it using the instructions here:
https://support.dronesmadeeasy.com/hc/en-us/articles/115003655066-Georeferencing-Aerial-Maps
It should be pretty easy to pick out usable corner features.
Thanks for the reply, Zane. Indeed, it's not hard to manually georeference, and that is what I am doing. If your concern for implementing is just value-for-time-spent, I fully understand. If it's more a philosophical concern (your comment that's it's just a "guide.using our system specifically"), I'd think of it more as effectively a quality/mask band for the resultant ortho and DEM output, and so worth sharing in autogeoref form, just like mask bands on satellite image downloads. In any case, no biggie if not a priority.