SRTM Contour Issues
choppinlt
Traveler
I have bound 4 different 3' SRTM's together. I bring the file in to FT3 and I create a CC3 contour map. Experimenting around I have a 10m and a 20m map version. For some reason my 20m version has gaps missing in the contour lines, but the 10m contours are fine. Why would this be?
Question 2: The contour lines appear to differ significantly/change on my CC3 map when created when the full SRTM is in the FT3 window (i.e. zoomed out) compared to when I zoom in at a particular area on the in FT3. Is this normal? How can I account for differing contours lines based on the same data with only different zooms?
Best Answers
-
jslayton
Moderator, ProFantasy Mapmaker
When you export contours from FT3, you get contours based on two major things: a sampling area (corresponding to the current zoom and projection) and a sampling density (corresponding to the Detail Level slider value). During contour export, FT3 makes a temporary image of the surface at the resolution specified by the Detail Level and sampled from your world map according to your current data set. The samples that it uses to fill out the data only take into account the nearest few samples of the data set: if your data is much more detailed than can be accommodated by the temporary image's resolution, then detail will be ignored. If the export resolution is very much less than the input data, you may see radically different contours at different zoom levels.
A 3' SRTM 1 degree DEM is 3600x3600 samples, if I recall correctly. Putting four of those together gives you a data detail of 7200x7200. FT3 will only generate contours based on a maximum of 1024x1024 samples based on the Detail Level slider, so it does indeed look like you're undersampling the data, even at the highest resolution for contour output. Why is the maximum output resolution from FT3 so low? Several reasons, but mostly because FT wasn't originally intended to be working from fixed-resolution DEMs, but rather from dynamically-generated fractals. It was also intended to be used to generate multiple maps in a set rather than a few giant maps.
A feature that's important to understand (and one that is related to the maximum allowed calculation size and may be related to some of the skips in data that you're seeing) is the maximum length of contours that can be generated in FT3 and used in CC3+. It's around 23000 points due to limitations in older versions of Windows. At high resolutions, FT3 can easily generated far more than 23000 samples on a contour. That many samples in the FCW file will cause some (mostly early) graphics drivers for Windows to choke and die, sometimes taking down the machine with them. CC3+ also has problems when editing that many points on a single path. It's less important since the SIMPLIFY command appeared in CC3+, but there are still hard limitation in the code for both programs relating to the lengths of paths.
Two things that can make certain lines look strange are using a dashed contour line during export from FT3 and using multipolies from FT3. I recommend avoiding using those features in an FT3 export when you can.
-
jslayton
Moderator, ProFantasy Mapmaker
Ideally, you'd have your data at (or at slightly lower than ) the resolution of the export. If you're exporting a 1024-detail CC3 map, set a zoom level that's about 512 to 1024 input samples wide. You'd need to do the math on that to get things close. It doesn't need to be exact, just not be majorly different.
Another option is to blur your input data (or resample it to lower resolution) to reduce the high-frequency components of the data, but you would need to do that in an external program. FT3 can't do processing on binary data sets. You could use Tools>>Actions>>Burn In To Surface after setting an appropriate editing resolution, but you then quickly get into questions about whole-world editing resolution and other topics that may not be helpful.
Answers
It would be helpful if you could show an example of your contour maps.
When you export maps to CC3 you usually get more details in your CC3 maps when you (in FT3) zoom in to a smaller area before you export the map to CC3 - but I maybe wrong. @jslayton is the master of FT3 here who can dive deeper into the problem I think...
When you export contours from FT3, you get contours based on two major things: a sampling area (corresponding to the current zoom and projection) and a sampling density (corresponding to the Detail Level slider value). During contour export, FT3 makes a temporary image of the surface at the resolution specified by the Detail Level and sampled from your world map according to your current data set. The samples that it uses to fill out the data only take into account the nearest few samples of the data set: if your data is much more detailed than can be accommodated by the temporary image's resolution, then detail will be ignored. If the export resolution is very much less than the input data, you may see radically different contours at different zoom levels.
A 3' SRTM 1 degree DEM is 3600x3600 samples, if I recall correctly. Putting four of those together gives you a data detail of 7200x7200. FT3 will only generate contours based on a maximum of 1024x1024 samples based on the Detail Level slider, so it does indeed look like you're undersampling the data, even at the highest resolution for contour output. Why is the maximum output resolution from FT3 so low? Several reasons, but mostly because FT wasn't originally intended to be working from fixed-resolution DEMs, but rather from dynamically-generated fractals. It was also intended to be used to generate multiple maps in a set rather than a few giant maps.
A feature that's important to understand (and one that is related to the maximum allowed calculation size and may be related to some of the skips in data that you're seeing) is the maximum length of contours that can be generated in FT3 and used in CC3+. It's around 23000 points due to limitations in older versions of Windows. At high resolutions, FT3 can easily generated far more than 23000 samples on a contour. That many samples in the FCW file will cause some (mostly early) graphics drivers for Windows to choke and die, sometimes taking down the machine with them. CC3+ also has problems when editing that many points on a single path. It's less important since the SIMPLIFY command appeared in CC3+, but there are still hard limitation in the code for both programs relating to the lengths of paths.
Two things that can make certain lines look strange are using a dashed contour line during export from FT3 and using multipolies from FT3. I recommend avoiding using those features in an FT3 export when you can.
Wow thanks for the answers; lots to digest! OK first here are the pics. The first pic was a 10m contour map with 4 SRTM together. I zoomed in to better show the contours AFTER import to CC3... As you can see, it turned out very well (except for the whole under sampling issue)
The second pic is 20m contour map of all 4 SRTM maps in FT3, HOWEVER I zoomed in before exporting to CC3...not sure what is happening there. I did multiple iterations of different settings and I never figured out why it was doing this.
As for under sampling the data, that make sense to me and answers my Question 2 above. I am familiar with the detail slider, but excuse my next question due to my newness...How can I make sure that I am staying within the 1024 limits for my exports to CC3?
My main question above remains about making sure I can create a 1024 map, HOWEVER I have an update on the missing lines in pic 2 above. I saw @jslayton tell me about the dashed line, but I thought to myself "well it's not that because I NEVER use the dashed line". Upon further review, it appears the cause of the problem was indeed the dashed line... oops! ? Thanks for bringing it up so that I would expressly check it. So back to my remaining question? How can I make sure I stay within 1024 on my map conversions?
The Detail Level slider won't go above 1024 at its maximum value.
Understood, but how do I make sure that I don't convert an undersampled 3' SRTM map into CC3? Or am I misunderstanding something? My understanding is that I need to zoom in more in FT3 prior to conversion...but how do I determine the boundary?
Ideally, you'd have your data at (or at slightly lower than ) the resolution of the export. If you're exporting a 1024-detail CC3 map, set a zoom level that's about 512 to 1024 input samples wide. You'd need to do the math on that to get things close. It doesn't need to be exact, just not be majorly different.
Another option is to blur your input data (or resample it to lower resolution) to reduce the high-frequency components of the data, but you would need to do that in an external program. FT3 can't do processing on binary data sets. You could use Tools>>Actions>>Burn In To Surface after setting an appropriate editing resolution, but you then quickly get into questions about whole-world editing resolution and other topics that may not be helpful.
Great thank you! Here is my process: I imported the SRTM to FT3 and noted the sample size of the width (2400 in this case). I measured the width of the SRTM at 93 miles, I converted this to 2400/93 samples per mile, and determined the number of miles in 1000 samples to be approximately 38.75 miles. I zoomed in to the FT3 map and made sure that the width was no greater than 38.75 miles, then exported the map to CC3 making sure the detail slider was at 1024.
Can we see the FT end result? I found this thread very useful, as I want to do something similar
Great, I'm glad someone else is getting some good out of all of my fumbling through this! Well here is the final FT3 product before exporting to CC3. The measuring tool indicates 34.83 miles across the bottom. Not sure you can really see it in this pic, but I added a coordinate grid overlay to help better orient me.
Thanks. Looks great. Love to see what it looks like imported into CC3+
Here you go!
Hi all, @WeathermanSweden , @Loopysue , @Quenten, @jslayton, @mike robel I wanted to thank you guys for your guidance thus far. I gave a sample of my work here: https://forum.profantasy.com/discussion/11264/historical-topo-map#latest
It's not complete yet, but it shows the primary details I am looking for. I need to play with it some to make some of the details show better at zoom out levels (in particular roads). Anyway I just wanted to share and thank you for the help.
Cheers