anybody dare to attempt this??? :)
http://www.carlheilman.com/wallpaper.screensavers/heilman.glacier.national.park.360.jpg
What a beautiful shot :). I think it will be awhile before I get the time/skill to do that in Terragen though ;).
I can provide a 10m res TER if there are any takers ;) Just give me the coordinates. I'm pretty sure I've got most of the data to cover all of the park.
lol....do you have any TERs or DEMs of Canadian national parks???
That is sure beautiful but I don't think theres anything there that should challenge TG. The hardest part would be to get the lighting right.
Richard
Damn, I was thinking Glacier National Park.... nearly Canada ;)
My downloading "campaign" is continuing... I finally got all of the SRTM3 files and am currently working my way through global Landsat images and 10m NEDs from the US (59 x 1 degree squares so far). I have some additional data for Canada but not 10m though...
Had a quick look at some data I got to pad the northern edge of Glacier National Park and it's around 25m resolution after reprojecting to UTM which isn't too bad. If I don't have the specific regions I can get it. The Canadian server is easier to download from than the USGS and they also have some interesting vector layers which could be used for general masks (once I figure out which layer is which).
Hey Ben,
The Canadian Glacier National Park (North West of the one in the USA) has 10m (10m Horizontal, 5m Vertical accuracy) data available. If you are downloading from the Canadian Server the sheets you want for basic coverage are (82N03, 82N04, 82N05 and 82N06) obviously if you wanted a larger area around the park, you would need additional sheets.
Just curious, but what is your process for converting the data for use as a TER? Do you use the default TER export options in Global Mapper, or do you customize the terrain spacing?
Dhavalmistry, which Canadian parks were you interested in? They all have varying qualities of data, but it is all available.
I am specially looking for PEI National park...but I wouldnt mind others too.....
Wow. Wonderful.
Quote from: RArcher on October 18, 2007, 10:11:32 AM
Hey Ben,
The Canadian Glacier National Park (North West of the one in the USA) has 10m (10m Horizontal, 5m Vertical accuracy) data available. If you are downloading from the Canadian Server the sheets you want for basic coverage are (82N03, 82N04, 82N05 and 82N06) obviously if you wanted a larger area around the park, you would need additional sheets.
Just curious, but what is your process for converting the data for use as a TER? Do you use the default TER export options in Global Mapper, or do you customize the terrain spacing?
Cool :) I can only get a good connection to the USGS for a couple of hours a day wihch is getting very frustrating.
Back at work now.
Checked my landsat collection and I already have up to row 55, columns 10-13 <https://zulu.ssc.nasa.gov/mrsid/mrsid.pl (https://zulu.ssc.nasa.gov/mrsid/mrsid.pl)>
USGS is not talking to me at all today so Canada here I come. Starting with the 82 grid and have queued 2 either side and 2 more rows above. So the offer still stands.... Give me the coordinates for the centre of a 40km square of your choice and I'll make a terrains set over the weekend :) The set will include some padding with lower res terrain to extend coverage, so the 40km square is just a guide. Had I realised there was 10m data earlier I would have done this sooner.... should have paid more attention the last time you directed me to the site. ;)
how do I get co-ordinates???
Google Maps
Centre the area and then click on the "link to this page" on the top right.
Here's the Canadian Glacier National Park to start you off.
http://maps.google.com/?ie=UTF8&ll=51.296276,-117.57019&spn=0.623413,1.387024&t=k&z=9&om=1 (http://maps.google.com/?ie=UTF8&ll=51.296276,-117.57019&spn=0.623413,1.387024&t=k&z=9&om=1)
I have all of grid 82 already... looks like a good day on the net
Here are a couple you might find fascinating:
http://maps.google.com/maps/mm?ie=UTF8&t=k&hl=en&ll=53.179704,-121.014404&spn=1.157164,2.570801&z=9&om=1
Bowron Lakes Provincial Park is in the centre - Great canoe route
http://maps.google.com/maps/mm?ie=UTF8&t=h&hl=en&ll=52.403257,-120.231628&spn=0.589001,1.2854&z=10&om=1
Wells Gray Provincial Park - Great mountains
One other thing, you need to be careful along the Alberta / British Columbia border. The data in BC side is very good 10m data, the data in Alberta is pretty terrible 25m data. Very unfortunate as some of the best mountains are along the border.
I would like this for now....
Wood Buffalo National Park of Canada
http://maps.google.ca/maps?f=q&hl=en&geocode=&q=Wood+Buffalo+National+Park+of+Canada&sll=51.3497,-116.490784&sspn=0.152886,0.32135&ie=UTF8&ll=59.326359,-113.320541&spn=0.124878,0.32135&t=k&z=12&iwloc=addr&om=1
oh...these are nice RArcher....I wouldnt mind these either.....any ones will do....
Looks like Wells Gray and Wood Buffalo.
I'll centre the larger padding terrains between the two. Seems like an area I'll be exploring now that I'm getting the data. The connection to the server has just gone down the toilet, but I'm still getting stuff through on a single connection.
I will pray that you get good connection....LOL....
That helped... or was it that I stopped the landsat download as well ;) I should have all grids adjacent to 82 by the end of the day which will be more than enough for now. I'll upload the terrains on Monday as a preliminary set including blue marble and landsat references. How much you use is up to you.
thanx ben....thats BIG help!
A little something to get you through the weekend ;)
Forgot the names of the locations so the names are a bit odd. Here's a 513x513 TER of the location I chose to make a set of this weekend, after looking at 30m res terrains of each location. TGD and lake mask to follow.... then on Monday you can just replace the terrain for the 4097x4097 version.
4097 lake mask and tgd to follow.
Lake mask and TGD enclosed in the zip. Quick render using the 4097 terrain
Looks pretty good Ben, Just trying to orient myself and figure out which lake it is. Here is a map i put together of the park a couple years ago with the lake names listed. By looking at the map can you tell which one you covered in the .ter?
http://www.maptown.com/temp/bowron-lakes-provincial-park-compressed.pdf (http://www.maptown.com/temp/bowron-lakes-provincial-park-compressed.pdf)
thanx again ben!
hey ben!.....I have a question about the mask.....did you extract the mask from the data or you just made it in photoshop?
Both. I extract a slope map from the data. This makes all lake surface white, and their banks not white.
I load this into Photoshop and use a magic wand with a tolerance of 1 (contiguous, no anti-aliasing) to select the lakes.... invert the section and fill with black. I usually use a satellite overlay to verify that the flat area I'm selecting actually is a lake, but I haven't done this one yet so some of the small lakes may be a bit odd.
This is a really good method because it uses the same data as the terrain. I add a slope restriction just to keep things tidy. This should also make it work reasonably well with the lower res terrain where smoothing will raise some of the areas covered by the mask.
Quote from: RArcher on October 19, 2007, 10:54:24 AM
Looks pretty good Ben, Just trying to orient myself and figure out which lake it is. Here is a map i put together of the park a couple years ago with the lake names listed. By looking at the map can you tell which one you covered in the .ter?
http://www.maptown.com/temp/bowron-lakes-provincial-park-compressed.pdf (http://www.maptown.com/temp/bowron-lakes-provincial-park-compressed.pdf)
The centre is just to the left of campsite 24, Isaac Lake. I may make 2 x 10m res terrains to get more of the length of the lake and adjacent mountains at high res and enclose it in a 30m terrain for padding (and then 60m for views from the peaks).... Looking at the map now, I seem to recall you wanted Spectacle Lakes?
How about a U-shaped set covering the lakes in the map?
I'm thinking that putting together an animation of the full 7 day canoe route (starts and ends in the top left corner of the map) might turn out rather well. Of course that depends if I can manage to surface things to look nice enough. The great thing is that there are plenty of reference photographs on the web to try and get the trees and lake sides correct.
I'm just loading all the data I have in GM now to see if I can figure out your export process.
Can you get UK data, I've always found it next to impossible without paying an arm and a leg.
Richard
Quote from: RArcher on October 19, 2007, 11:55:08 AM
I'm thinking that putting together an animation of the full 7 day canoe route (starts and ends in the top left corner of the map) might turn out rather well. Of course that depends if I can manage to surface things to look nice enough. The great thing is that there are plenty of reference photographs on the web to try and get the trees and lake sides correct.
I'm just loading all the data I have in GM now to see if I can figure out your export process.
For the lake mask all I do is set the shader to a slope shader (I think it ignores any lighting, but turn it off just in case) and then export a 24-bit RGB tiff with the same settings as your terrain.
For the terrain there's a bit more planning of coordinates to give terrains of the right size and overlap. Technically it doesn't really matter, the only crucial piece of information is the coordinates of the centre or one corner of the terrain so that you can calculate the offset. Working with UTM projection (primarily to remove a lot of distortion) throws everything into metres by default which makes it very easy to get the units you need for TG. I use a database to do all the brain work for me. If you have the data I can attach a text file with the coordinates I use for the terrain set as an example. Then you don't have to download the TER files.
The matching landsat images are really helpful for distributing trees as well as lot's of other stuff. Some exmaples I've done include the NZ orbital renders I'm working on at the moment and the Tetons images. The full resolution images are roughly the same res as this terrain data.
Quote from: cyphyr on October 19, 2007, 01:08:51 PM
Can you get UK data, I've always found it next to impossible without paying an arm and a leg.
Richard
I have the same problem here in Australia. Maximum free resolution is SRTM3. There's an election coming up in a few weeks.... maybe I could ask my local candidates?? :D
Damn.... left the files at home (forgot to transfer them to my USB drive).. will have to be tomorrow...
I've posted a provisional set of files. The canoe theme stuck with the filenames ;)
http://www.path.unimelb.edu.au/~bernardk/terragen/terrains/canoe/files/ (http://www.path.unimelb.edu.au/~bernardk/terragen/terrains/canoe/files/)
While I say they're provisional, that's only referring to the supporting images which are ridiculously large at the moment as they were just a quick export. The base terrain will remain unchanged. It's a 4097x4097 terrain, the total width is included is the number in the filename.
Full resolution for this data set is set to 15mx15m. The landsat data is marginally higher, and the DEMs are around 14x23m metres with this projection. The slope map is at this resolution. I'll also release smaller versions for the other terrains but you can always use this to crop a region and calculate it's offset. There's also a small version of the landsat image scaled back to 4097x4097. Don't try inserting the images into TG at full resolution.. just to save you a bit of time.
File sizes are a bit big, but using other compression methods either don't gain that much, or reduce the quality which makes working with them much harder. My own files are all LZW compressed TIFF, but the max. quality JPEGs are much smaller and still work OK since the original data has JPEG artefacts in it anyway.
I would have added the other terrains but I made a booboo in the numbers and they don't overlap... duh.... There will be two side by side terrains providing roughly 100x60km at full resolution in the centre of this terrain. Keep an eye on the URL for new files... or you'll see an announcement in the file sharing section when the set has been completed.
PS. I was also going to play with georeferencing the map... handy for reference and could be interesting blending from map to TG surfaces.
what is the use of slope map and how do I use it???...its pretty large for a map....
I use the slope map for 2 things.
1: Creating a mask for lake surfaces. Magic wand in PSD, tolerance 1, no antialiasing. Lakes surfaces are white, anything not flat is not white.
2: Guide for drawing/selecting rivers. Landsat images are not exactly aligned on the terrain and if you blindly trace over a river for creating a mask you can end you with rivers running up along the sides of gorges.
I'll probably throw a lake mask in with the high res panoramas anyway as these do compress a lot..
See... I told you they compressed a lot. The zip contains a very quick lake selection and basic lake setup. 90x90km. Once again very quick selections used and not verified as actually being lakes, but in 10 minutes that's a lot of lakes in a 90x90km area.... (and part of a river).
The screen grab is a good sample that shows how it is usfeul for drawing rivers. As this is based on your actual terrain it's more accurate than using the landsat images.
Enjoy.
Fixed the other two terrains. I'll post them when I get back to work. Apart from getting some interesting terrains and a *lot* of extra data to play with, setting up the TGD for this tiled terrains showed up a flaw in my function mask. It worked as long as the terrains didn't overlap more than the blending margin. In this case they did, which resulted in a sharp differentce in altitude along one seam.
It took me a little while to figure out what went wrong, but in the end is was simply a negative value in the blending mask as the result of subtracting one mask from another. This was easily fixed by including a clamp 01 node after the subtract node.
So here are two sample renders. The first shows the extent of the full resolution terrain (and why this is such a popular canoe route.... that's one very big lap of the mountains ;)
The second shows some alpine lakes near the northern edge of the lake mask I posted... Just testing out the resolution of the terrain and the mask accuracy. I'm going back to my New Zealand masks for a while. This should be enough to keep you busy for a while. I'll post some more Canadian terrains later now that I have the full set of elevation data (some 12,000 files).
Enjoy
thank you Ben.........
just a question....can you not use a slope mask to distribute population.....more accurately???....
Quote from: dhavalmistry on October 23, 2007, 11:33:43 AM
thank you Ben.........
just a question....can you not use a slope mask to distribute population.....more accurately???....
Yes..... and No.
My tree population masks can get relatively complex. It comes down to a process of combining a number of different components that make up the final distribution of trees. From the landsat images I usually just create a boudary for the main tree population. This is then merged with other shaders to provide the variation in density within that region.
It's worth including this component as it provides a reasonably realistic distribution of trees based on the actual distribution taking other factors into account that cannot be derived from the terrain alone (geology, roads and man made features, agriculture, forestry etc...). The USGS also provide canopy density data which, while low resolution, provides a useful starting point.
Other factors for masking include slope, altitude, direction of slope/sun, distance from camera, subtraction of other masks (e.g. lakes/rivers) and so on. This is one reason why I normally end up with parallel node networks, 1 for surface shaders, 1 for mask components and 1 for mask combinations. You might look at a render and think "Why bother?" but when you look at a render of a region you know very well it's really amazing.
Now stop reading this and get the rest of the files I posted ;). (included a landsat image with increased compression to drop the file size, the 2 new terrains and a TGD linking all 3 terrains and the sample lake mask)
Enjoy ;D
hey Ben.....I was wondering if you could get me the data for Banff National Park of Canada in TER format
http://maps.google.ca/maps?f=q&hl=en&geocode=&time=&date=&ttype=&q=Banff+National+Park+of+Canada&sll=49.891235,-97.15369&sspn=25.783423,89.296875&ie=UTF8&ll=51.493355,-116.238098&spn=0.388179,1.395264&t=k&z=10&iwloc=addr&om=1
thanx in advance!
OK.. may be a weekend job. I think that side is only 25-30m, but that's still a lot better than SRTM3
ya sure....no hurry...take your time....
Melbourne Cup this week.... long weekend :P
Posting the terrains when I get to work in the morning. I'm making a 15 and 30m version, although the 15m data stops along the ridge to the west of the valley :(
Here's the coverage of the 30m terrain and a sample render with 15m data to the right of the ridge and 30m data to the left
Quick URL for now... tidying up later.
http://www.path.unimelb.edu.au/~bernardk/terragen/terrains/banff/files/ (http://www.path.unimelb.edu.au/~bernardk/terragen/terrains/banff/files/)
2 terrains, full res landsat and slope map, tgd merging terrains.
Enjoy :)
sorry Ben....I had problem like this before too....how do I use the 7z files???
and 79MB jpg file???
http://www.7-zip.org/download.html (http://www.7-zip.org/download.html)
Or I use ZipGenius instead of Winzip (just as good, if not better and completely free)
http://www.zipgenius.it/ (http://www.zipgenius.it/)
Only get the landsat JPEG if you intend to use it for creating masks from it. Otherwise it's not much use. Google Maps or Google Earth are better references as they're "real" colour and usually better resolution. I usually work with larger images for generating masks and then downsample them... although a 1.5Gb PSB file takes a while to load.