+- Mooncoder Forum (http://www.mooncoder.com/forum)
+-- Forum: Software (/forumdisplay.php?fid=1)
+--- Forum: PrettyMap (/forumdisplay.php?fid=4)
+--- Thread: Problem with raster converter (/showthread.php?tid=20)
Problem with raster converter - AntonM - 11-28-2007 10:24 AM
I'm having a problem with the raster converter -- it seems to get the size wrong in one direction.
I did a screendump from Ordnance Survey's MemoryMap (a highly authoritative UK cartographer) and carefully entered into the converter the co-ordinates MemoryMap gave for the four edges of the map.
By inspection, the map looks like it has a W:H aspect ratio of about 8:5 (i.e., ±1.6), and looking at the co-ordinates provided by MemoryMap, that seems about right.
The converter agrees that the image has an aspect ratio of 1.578... but says the co-ordinates suggest a ration of 2.66...
The co-ords are -2.163 & -2.087 and 50.082 & 53.11. Just looking at the fractional parts of those numbers also suggests a ratio of ±1.6.
The converted map is, in fact, squashed in the Y direction, and thus unusable.
What am I doing wrong?
I attach a screendump of the converter window.
RE: Problem witrh raster converter - Mooncoder - 11-28-2007 10:39 AM
Hello Anton,
Considering coordinates from your screenshot:
-2.087+2.163 = 0.076000
53.1103-53.0818 = 0.028500
so the aspect ratio
0.076000/0.028500 = 2.666667
Please verify...
Maciej
RE: Problem witrh raster converter - AntonM - 11-28-2007 11:36 PM
Mooncoder Wrote:so the aspect ratioHmm -- yes, your calculations are quite right. My visual inspection was totally wrong!
0.076000/0.028500 = 2.666667
Please verify...
I'll go back and have another look at MM's co-ordinates.
Thank you!
Anton
RE: Problem witrh raster converter - Mooncoder - 11-29-2007 02:35 AM
The map may look squeezed/stretched depending on the projection, but the coordinates of map corners should match and should be accurate.
Best,
Maciej
RE: Problem witrh raster converter - AntonM - 11-29-2007 06:44 AM
OK, I've tested co-ords now with MM and with Google Earth -- and I'm now really confused.
Both MM and GE give me the same Y:X 1:2 aspect ratio to what looks to me like a more or less square piece of earth.
I'm attaching a screenshot of GE, centred approximately around 53.0104 and -2.227, at the level shown, so it should be easy to reproduce the situation. The same area in MM gave more or less the same co-ordinates.
The co-ords at the edges (in GE) are:
top: 53.025690
left: -2.260963
right: -2.192941
bottom: 52.992256
According to my calculator, that gives a North/South difference of 0.033434 and an East/West difference of 0.076689.
It's been pointed out to me that the curvature of the earth is what's causing this, as at these latitudes, lines of longitude are getting a lot closer together. That sounds sensible, but then any flat representation of that piece of earth should be portrayed as it actually is (in this case, a square).
I've also attached the converter window of the GE display. as you will see, it also gives the aspect ratio warning: "Warning: Width/Height ratio should match, image=1.2190889, coordinates=2.0345142". I agree that there is a disparity between the image's aspect ratio and the co-ordinates' aspect ratio -- but is that not to be expected when, at these latitudes, co-ordinate "squares" are actually oblongs?
Somewhere I'm missing something in the whole process. I really want to get this worked out, because it will be an extremely useful feature.
Anton
RE: Problem with raster converter - Mooncoder - 11-29-2007 09:08 AM
AntonM Wrote:It's been pointed out to me that the curvature of the earth is what's causing this, as at these latitudes, lines of longitude are getting a lot closer together. That sounds sensible, but then any flat representation of that piece of earth should be portrayed as it actually is (in this case, a square).
Welcome to the world of map projections
http://en.wikipedia.org/wiki/Map_projection
You are right about the distortions of longitude/latitude.
Regarding "flat representation" that should be a square... depends on what you mean by flat.
Anyway: what you see in PrettyMap "standard view" (flat or 3D) uses equirectangular projection. Read more here:
http://en.wikipedia.org/wiki/Equirectangular_projection
So yes, your square area may look rectangular (stretched), but as long as you confirm that coordinates of your map corners are accurate, it is fine (i.e. the GPS mark will be properly placed on the map).
If you want a more realistic (and computationally intensive) view, use the spherical projection mode in PrettyMap.
Maciej
RE: Problem with raster converter - AntonM - 11-29-2007 09:52 AM
Okay -- I can understand how that all works, and I'll read the references you gave. And I'll try the spherical projection as well.
Out of curiosity -- why does the converter give a warning about the aspect ratios of image versus co-ordinates, when the only place they will really be "right" is peri-equatorial?
Just as background, for what it's worth: I purchased PrettyMap for this feature alone, i.e., that I can use any raster map as a GPS map on my Palm TX. I have TomTom which is fine for driving, but what I also need is something to guide us on our walks on footpaths, where TomTom is of no use, because our precise location is not always clear from a visual study of the footpath map, and a map is no use if you don't know where you are on it. PrettyMap seemed exactly right -- but it's going to be less easy to use of it gives a distorted view of the map. I'm really hoping that the spherical projection option will solve that! I'll try it this weekend.
Anton
RE: Problem with raster converter - Mooncoder - 11-29-2007 11:23 AM
Oh yes, perhaps "warning" and "should match" are not the best terms. On the other hand, the warning is issued only when a significant difference occurs, which is often caused by people entering incorrect coordinates. It might be better to say "Note: width/height ratios differ".
Maciej
RE: Problem with raster converter - Boisclair - 04-11-2011 06:36 PM
Mooncoder Wrote:Oh yes, perhaps "warning" and "should match" are not the best terms. On the other hand, the warning is issued only when a significant difference occurs, which is often caused by people entering incorrect coordinates. It might be better to say "Note: width/height ratios differ".
Maciej
I did a screendump from Ordnance Survey's MemoryMap (a highly authoritative UK cartographer) and carefully entered into the converter the co-ordinates MemoryMap gave for the four edges of the map.