Welcome to GPSFileDepot!

Main Menu

Limits of the .mp and .img formats

Started by Boyd, June 28, 2010, 11:12:48 AM

Previous topic - Next topic


I have been working on a personal project to map the little trails on my land, and learned something in the process which might be of interest. Many of you have read the tutorial here, specifically the comments on page 23 about map resolution:

Quote'24' refers to the precision of the latitude and longitude coordinates provided for the object. When declaring the location of objects in the world, the highest precision available uses 24 bits of resolution, which resolves to an accuracy of about 2.5 meters. Consumer GPS units have a best-case accuracy of about 8 meters.

In another thread here, somebody questioned the math that lead to the 2.5 meter figure. I am not sure how the calculation was performed, but the following screenshot will graphically demonstrate it. Using GlobalMapper, 1 foot per pixel aerial imagery, and track points accumlated from many walks on my little trails, I created my map. The points on the polylines are quite close together in some cases - one or two feet apart. I exported the map as a .mp file and then opened it in GlobalMapper again, overlaid on the original. In the image below, the pink lines are the original and the blue lines are the .mp file.

So you can see that the actual coordinates have been rounded off to 24 bit numbers, creating stair steps. Right away it became obvious that I wouldn't be able to make the kind of map I wanted using Garmin's vector format. Now granted, this detail is beyond the accuracy I can expect from my Oregon anyway... but I still wanted to make my map!

The solution was simple; I created a "custom map" (raster image) as a .kmz file. I had overlaid my vector data on 1 foot per pixel aerial imagery, so I just exported everything as a .kmz file in the proper format for the Oregon. But the results still weren't quite what I had hoped for. At close zoom settings, the lines looked too jaggy. So I re-sampled everything at 4 inches per pixel and exported again. This produced a very nice looking map with smooth lines even at extreme zoom levels.

You may not ever want to create a map at this resolution, but if you do then it's nice to know that the 2.5m/24 bit limit does not apply to "custom maps". Of course, you will need an Oregon, Dakota, Colorado, GPSMap 78 or 62 to use this kind of map.


Great ideal; thanks for sharing. 
Keep in mind that with Garmin's custom 'raster' maps being limited to <= 100 cells of 1024 by 1024 pixels, a .kmz file will be limited to about 3000'+ by 3000'+.


Yes, that is true. However the area I needed to map was pretty small and only consists of 25 JPG's at 1024x1024. If I had used 1 foot per pixel it still would have produced much better resolution than Garmin's format, it just wouldn't have been as pretty.  :)  That would have let me map an area almost 2 miles x 2 miles.

Now I am also curious as to whether the position as indicated by the GPS pointer "snaps" to 24 bit numbers. It doesn't really appear to do that.


Very interesting.  I wonder if Garmin will have to update their format at some point to account for all these "issues".

I need to get better at these kmz maps.  It worked great overseas just not so well up here in Alaska.
Dan Blomberg
Administrator - GPSFileDepot
GPS Units: Garmin Dakota 20, Garmin GPSMap 60csx, Nuvi 255W, Nuvi 250W, ForeRunner 110, Fenix 2, Tactix Bravo, Foretrex 401
See/Download My Maps!


Perhaps Stan limited the coordinates to 24-bits while Garmin uses 32-bit?


Yes, I wondered about that myself. It may really just be a cgpsmapper limit.


Another thought.

I asked and was told about 1 1/2 years ago that Garmin normally uses bit-levels 23,21,19 in the products.  If that is true/still true; then there might be some code that tells the GPSr to use rounding between points versus angular. 
Might also be the number/closeness of the points.
Anyone try comparing file sizes for a tile in a Garmin product versus that same area made from with cgpsmapper?


QuoteEver see this issue when zoomed in to CN? Data goes from shapefile to .img with curves intact with much higher resolution than 2.5m

You got me wondering about this, and Basecamp can be very instructive with BirdsEye loaded. Here are a couple examples from City Navigator which seem very typical. Basecamp only allows zooming to 80ft (can zoom to 20 feet on the Oregon). But even at the 80 ft level you can tell that City Navigator is a relatively crude approximation of curved roads. The points are pretty accurately positioned however.