Hi All
I've had a couple of questions about the methods I used for these images: http://www.planetside.co.uk/forums/index.php/topic,18550.0.html (http://www.planetside.co.uk/forums/index.php/topic,18550.0.html) so I thought I'd make some notes here for reference.
First off, I'm preparing my data in GlobalMapper which, among other things, lets me mix data of various resolutions and from various sources. I've found this to be a better way to mix data rather than loading DEMs from different sources in TG. I used to export TIFFs but I'm now trying out goetiff. For example, my global terrain of the earth includes ASTER DEM with bathymetry on top of which I add SRTM4 or USGS 10m NED. For texture maps, I load in an image from the Blue Marble series.
TIPS:
* Load ALL of the data you're going to combine before exporting files. GIS applications will use the highest resolution data available to fill holes, which TG doesn't do. Using geotiffs from different sources in TG may produce noticeable effects at the edges.
* Centre the lat/long projection at 0,0. It's not critical but it makes working with full globe data/images easier.
Once I've got all of the data and imagery loaded it's pretty easy to export different areas at different resolutions. High resolution for the area I'm looking to render and then a wider area at lower resolution for distant terrain and upwards to a complete global terrain/texture if required. Then it's just a case of loading the terrain/images into TG from lowest to highest resolution.
For planetary renders this makes for a relatively simple TG project
[attach=1]
For Mars, it's possible to load the entire 128px/° DEM at once. (I've split them into 2 geotiffs) and a 0.025°/px texture for just under 6Gb RAM usage. This DEM has a 2° hole at the poles so I've filled these with the 64px/° DEM.
[attach=2]
For renders near the surface it pays to change the coordinates for the apex of the planet to the area you want to render.
Thanks, Ben. I copied this to my tutorial section, hoping to use it some day.
Making some progress with data for Mars after sorting a few technical issues out (images inconsistently loading with the wrong projection in GlobalMapper). The gaps in the range of resolution is proving challenging but it's also forced me to rethink how I export the data which should be quite useful for mapping out terrains for animations. Here are some samples of the terrain data with fractal detail disabled.
MOLA128. the original :)
[attach=1]
HRSC plus one strip of HIRISE
[attach=2]
And yes, we're gonna need a mask ;D This one's just a quick geotiff edited in Photoshop but as geotiff needs to be RGB for this to work the extra effort of positioning a SSS will be worth it.
[attach=3]
The HIRISE data here was exported at 38m/px, a long way from its full resolution of 1.9m for this one, but good enough from this distance. Given the noise from the photogrammetry processing I probably wouldn't try higher than 5m, but we shall see.
Man that HIRISE data is amazing-looking. I really hope this ends up working out, I've been looking forward to a higher resolution real Mars animation for a long time!
- Oshyan
It certainly is. This is 2m resolution at a camera height of 50m. No fractal detail, render detail = 1. It won't be without its challenges but I'm quite hopeful. Lots of data to explore.
Stupendous, you can even see the erosion channels :o
Very nice indeed. The banding is a bit of a concern, but the detail is undeniable still. Check out this video, also using HIRISE data, but apparently with even more detailed data (perhaps the data just in Candor Chasma is better?): https://www.youtube.com/watch?v=0WsjeJiAR4E (https://www.youtube.com/watch?v=0WsjeJiAR4E)
No major banding either...
Edit: this one is pretty incredible as well: https://www.youtube.com/watch?v=QutMf79sFhk (https://www.youtube.com/watch?v=QutMf79sFhk)
Or check out the crazy detail in the rocky talus slopes in this crater! https://www.youtube.com/watch?v=ftS4rujxJWo
- Oshyan
WOW..that mars explorer does pretty nice animation on all 3 of those...are the over hangs real do y think, or just an optional delusion....?
Don't forget I'm rendering without fractal noise to see what the data is really like. This one is at a camera height of 1.5m. Foreground is 5m resolution, 2m res starts about 1/2 way from the bottom. Need to change flow factor to something really low or remove it altogether as it does odd stuff with the small mounds resulting from noise
Quote from: bobbystahr on July 07, 2014, 11:19:48 AM
WOW..that mars explorer does pretty nice animation on all 3 of those...are the over hangs real do y think, or just an optional delusion....?
From a data perspective I'd say they're an optical illusion. The typical data processing for those images to get the terrain would create a heightfield rather than an arbitrary mesh so there would be no overhangs. The imagery has a much higher resolution than the terrain which helps make it look so good. I'm still getting used to the scale of things on Mars, but some of those "ground level" animations I would guess are at a camera height of at least 50m and probably more like 150-300m. The crater wall in my image for example is 3km away, but even noisy 2m data at that distance looks pretty great
Quote from: bigben on July 07, 2014, 05:17:09 PM
Don't forget I'm rendering without fractal noise to see what the data is really like. This one is at a camera height of 1.5m. Foreground is 5m resolution, 2m res starts about 1/2 way from the bottom. Need to change flow factor to something really low or remove it altogether as it does odd stuff with the small mounds resulting from noise
that's pretty amazing...having a great time exploring your Grande Canyon btw
Quote from: bobbystahr on July 07, 2014, 05:31:58 PM
that's pretty amazing...having a great time exploring your Grande Canyon btw
Pretty easy spend many hours exploring that. Enjoy :) The other good thing about using geotiffs is that you can chop and change your elevation data to be more efficient for the area you're working on without having to adjust too much. Those geotiffs are pretty big but you don't need 10m data for everything in a scene and it's possible to create multiple TIFFs at increasing dimensions and lower resolution to get a very expansive view for 40-80mb of LZW compressed geotiffs, leaving more RAM for plants etc...
Quote from: bigben on July 07, 2014, 08:43:27 PM
) The other good thing about using geotiffs is that you can chop and change your elevation data to be more efficient for the area you're working on without having to adjust too much. Those geotiffs are pretty big but you don't need 10m data for everything in a scene and it's possible to create multiple TIFFs at increasing dimensions and lower resolution to get a very expansive view for 40-80mb of LZW compressed geotiffs, leaving more RAM for plants etc...
You should really publish a complete method of how you do what you do as a single volume .pdf so even a raw noobie could take advantage of this powerful feature TG has that none of the others use. It's tho only spherical world which is why it works so well I'm guessing.
I guess one could collect all the posts and order them and I would have when I was 40 but at 65.5 yrs old I'm more content with making the resulting pics.
Quote from: bobbystahr on July 07, 2014, 08:54:28 PM
You should really publish a complete method of how you do what you do as a single volume .pdf so even a raw noobie could take advantage of this powerful feature TG has that none of the others use. It's tho only spherical world which is why it works so well I'm guessing.
I guess one could collect all the posts and order them and I would have when I was 40 but at 65.5 yrs old I'm more content with making the resulting pics.
I'll try and make up a tutorial once I figure out what works and what doesn't but that requires others to have (access to) the data and a GIS application.
Of more interest to you will be that I'm going to be making a more RAM-friendly version of the Grand Canyon datafiles. This was just a quick release so that people can play with it now. Producing the data is a lot easier than the previous methods I had for creating tiled terrains so I may make a number of TG-friendly datasets available for people to use
Quote
Of more interest to you will be that I'm going to be making a more RAM-friendly version of the Grand Canyon datafiles. This was just a quick release so that people can play with it now. Producing the data is a lot easier than the previous methods I had for creating tiled terrains so I may make a number of TG-friendly datasets available for people to use
have I said woo hoo lately.....should be awesome Ben
Even on Mars they do donuts ;)
Google Earth is looking like a useful too for tracking down/previewing data for Mars. The HRSC and Hirise DTMs and imagery are used in the model.
I've set wget loose on the HRSC elevation data :D
Quote from: bigben on July 08, 2014, 10:46:12 AM
Even on Mars they do donuts ;)
Google Earth is looking like a useful too for tracking down/previewing data for Mars. The HRSC and Hirise DTMs and imagery are used in the model.
I've set wget loose on the HRSC elevation data :D
Heh heh heh, reorienting the rover no doubt but it has a very
red neck human signature about it...And
wow again...never thought of taking a GoogleShip to Mars...
Hey Ben, sorry, but I'm confused by the threads. Which part of which thread are you posting as a share? the stuff here or file sharing? I thought mars, then I thought the grand canyon. But now I'm not sure which thread I'm typing in! OK, going back to sleep now :-\
This thread's about the tests I'm doing on preparing real elevation for data optimised for TG, so that includes both earth and mars. The terrains I posted recently in File sharing are a global earth set and the Grand Canyon. I'll try and post a terrain set soon that illustrates the findings I've come up with so far (using mulitple geotiffs to provide a variable LoD terrain for a large area with "very low" RAM usage). At the moment though I'm working on sorting out a few issues with the raw data for Mars.
And speaking of which, another milestone was reached on the red planet with this World Cup-inspired post: http://hirise.lpl.arizona.edu/ESP_037117_1755 (http://hirise.lpl.arizona.edu/ESP_037117_1755). This really illustrates the insane resolution of the imagery. A bit disappointing that they put a yellow dot on top of the rover but you can still see it (and its tracks). http://static.uahirise.org/images/2014/details/cut/ESP_037117_1755-4.jpg (http://static.uahirise.org/images/2014/details/cut/ESP_037117_1755-4.jpg) 8)
I have the data for this crater now, pity it's in the middle of nowhere but it'll be interesting to compare.
Finally figured out how to get Global Mapper to merge data nicely which is proving to be a critical step, both in terms of dealing with the large differences in resolution between datasets and small variations between individual data files.
First image is a render of merged mola128 and HRSC data from an altitude of 50km. The HRSC tiles vary between 25 and 75m per pixel. Second image includes hirise data that has been overlayed without feathering the edges. Camera altitude is 1km and you can easily see the discrepancies in altitude at the edge of the tiles. Foreground data was exported at 4m per pixel. Screengrabs are from Global Mapper showing mola128 and feathered HRSC data. You can see the distribution of the different resolutions
Turned out to be a lot more pre-processing than I expected but it's starting to come together.
Well y got me watching for sure.....nice seeing the various stages, shows how much has gone into the final comp.
Quote from: Oshyan on July 07, 2014, 04:39:15 AM
Very nice indeed. The banding is a bit of a concern, but the detail is undeniable still. Check out this video, also using HIRISE data, but apparently with even more detailed data (perhaps the data just in Candor Chasma is better?): https://www.youtube.com/watch?v=0WsjeJiAR4E (https://www.youtube.com/watch?v=0WsjeJiAR4E)
No major banding either...
Edit: this one is pretty incredible as well: https://www.youtube.com/watch?v=QutMf79sFhk (https://www.youtube.com/watch?v=QutMf79sFhk)
Or check out the crazy detail in the rocky talus slopes in this crater! https://www.youtube.com/watch?v=ftS4rujxJWo
- Oshyan
Those are very very cool!
@ben
Im looking forward to when you get the maps put together the way you want, and also start talking about adding color information and fractal detail.
I am guessing that you will add NASA provided color info, since using just color shaders and nodes to limit color by slope and stuff would be very difficult to get right. And the earlier images you posted are full color. But I can't guess at what your close in details will look like with the color info. Just as good?.. The last three images above are very inspiring.
Also very interested to see what you and others will do with adding fractal displacements for the fine detail. Lots of room for creativity beyond loading maps and coloring!
As for me, I always wanted to build a very detailed satellite model and have it over a TG world in low orbit. Or even a spaceship entering orbit as in the beginning of the Prometheus film. What your doing here will really make that look good!
I think the handling of data and images will probably keep me busy enough. I'm interested in repeating some of those videos in TG3. They're based on a single DTM and texture so they've wimped out a little ;), and the terrain resolution looks a little low but it's very hard to judge with a texture overlay. But image textures, even at this resolution, can be limiting when it comes to ground level, and consume a lot of RAM which then limits the available resources for TG to add finer detail.
[Edit] watched th videos at work with a faster connection and now I can see the full res data... pretty sweet [/edit]
But there are others with more TG skills than me (http://www.planetside.co.uk/forums/index.php/topic,15070.0.html (http://www.planetside.co.uk/forums/index.php/topic,15070.0.html)) who can benefit from this. I'm more of a technical tinkerer.
OK, a quick test to illustrate the potential of TG for rendering Mars data. The youtube animations linked in this thread are all of single hi res DTMs. In some instances there is a considerable loss of context from the surrounding terrain. Here's a quick render of Candor Chasma (3m/px) with and without surrounding MOLA data (900x600km chunk at 128px/°). So for an extra 11mb of (LZW-compressed) geotiff you get to see that it actually is the floor of a deep chasm.
Wow, now *that* is a great (dramatic) illustration! If ever there was a use case for multi-resolution data merging...
- Oshyan
Yes, I got inspired to dress this up a little, add HRSC data into the mix and render a spherical panorama from the middle of this DTM. Looking good so far, although I still have to try the image texture map.
Agree, WOW...great leap forwards. Can't wait to see it tex mapped.
Quote from: bobbystahr on July 12, 2014, 07:03:30 AM
Agree, WOW...great leap forwards. Can't wait to see it tex mapped.
Got an error that someone had replied since I started writing.... now where was I? ;)
Got a little impatient to see my georeferencing attempt and cancelled the spherical render. This illustrates what I was saying about the texture disguising the resolution of the terrain.
First image is the terrain with a plain grey surface. Second image has 2 greyscale textures from the hirise image: 4x3km of 47cm/px and 12x8km of 3m/px. The sphere is 2m in diameter, terrain resolution at the camera is 1.9m/px.
Guess that was me. I reiterate...WOW
Testing some comparative scenes from http://www.youtube.com/watch?v=0WsjeJiAR4E. The Youtube video show just a single terrain and a false-colour image map. I'vel loaded up TG with the full res terrain and image and some surrounding terrain (still one image set to go)
This one is around 1:07
Well done and thanks for that video...shared it on FB for my MarsNutz friends....
Really pretty cool ben
Bugger, just discovered they used a second DTM for that video that I haven't found yet. I ran out of data following their camera around :-\
[edit] Gonna wget me some more data ;)
Hi all!
I'm trying to produce the same results myself but finding it hard to follow the steps on this thread.
I basically want to create the Ares 3 landing site in as high detail as possible using MOLA 128 and HiRISE data to then be exported to Terragen and then into a game engine. Can anyone help? I have a thread here: http://www.planetside.co.uk/forums/index.php/topic,20346.msg201486.html
Thanks!