Mars Explorer with HiRISE DEM/Texture Options

[djellison]

Think local news chopper on a car chase


And I can see what Nirgal means - but that particular Gusev DEM is just awesome - it really is that lumpy!

[tanjent]

Wow - no vertical exaggeration. It looks like rough terrain even for (or perhaps especially for) a three billion dollar nuclear powered rover. Could one of these images be modified to indicate a ground track for this flight, especially with respect to the anticipated Curiosity landing ellipse? Maybe there are even better context photos available - I just recalled seeing this one on the planetary society blog.

http://www.planetary.org/blog/article/00000589/

It's a wonderful flying sensation. I have tried the interactive flight simulator included in Google Mars, but this has much more detail. Just for fun, how about adding a sound track with some engine hum and the whistle of a very attenuated Martian wind?

[elakdawalla]

One of the main selling points for MSL has always been that it should be able to land in more exciting areas (topographically speaking) than an airbag-equipped lander. (MSL used to stand for "Mars Smart Lander," but the meaning of the acronym was changed partway through development.) But you're right that some of this topography is too extreme for the landing ellipse. The other thing that MSL is supposed to be able to do is to land in a less exciting area and drive a very long distance to a more exciting area. Ken Edgett has an informative guest blog post on this -- Gale is a so called "Go-To" proposed landing site for MSL, where you arrive at the interesting stuff after a long drive.

(It's funny how many of my Google searches on topics lead me either to posts I've forgotten about on my own blog, or to past discussions I've forgotten about on UMSF!)

[djellison]

That dem area is reachable by Curiosity if it goes to Gale - but isn't inside the landing ellipse

[tanjent]

In my earlier post, I did not really mean to focus so much on whether Gale was the best target for MSL/Curiosity or not. I have a lot of confidence that that decision will be made by people who are very good at balancing the risks against the rewards, and I also recognize that there will still be a lot of risk no matter what site is chosen.

What inspired me to write the post was the more general problem regarding the forest and the trees. Viewing the Pathfinder clip, it was easy to stay oriented because I was on familiar terrain with a small number of clearly recognizable landmarks. By contrast, the Gale clip took me down to a level of detail at which I was completely lost. Despite the wild, fascinating terrain and the giddy sense of flight, I couldn't relate what I was seeing to any map or 2D representation of the larger context. I wasn't even sure I could identify the consistent circularity of a crater wall. Even flying on airplanes here on Earth I have noticed that when the sun is high enough in the sky not to provide a reliable guide, it is difficult to maintain my sense of direction by simply counting the degrees in a turn. Maybe pilots have the experience to do this easily, but in my case, I'd really like to see the ground track from a higher altitude, so I could say to myself, "Oh, that's what that feature looks like from up close!"

[Mars3D]

Can you tell us at approximately what ground speed and altitude the camera is flying across the landscape?

Roughly it is 100 mph at 100 meters.


Here's a couple more animations with an increased texture resolution of 0.37m.

Crater in Athabasca valles
http://www.youtube.com/watch?v=kyEqv9aIy8E
Mojave again.
http://www.youtube.com/watch?v=GAPnhNFykis

[charborob]

Great stuff, Mars3D! We really get the impression that we're seeing the real thing!

[Nirgal]

Here's a couple more animations with an increased texture resolution of 0.37m.

Wow ... breathtaking, as always !

short question:
According to the HiRISE docs, the post spacing of most of the DTMs is 1m ... How did you achive the 0.37m resolution ?

[djellison]

DEM is 1m

Texture is 0.37m

It takes two elements to do these things - the elevation model, and the texture map you put on it!

[Mars3D]

I've upgdraded my RAM and tweaked the software and now I'm able to use the full HiRISE texture resolution (~25cm). Here's my first animation at the full resolution, this one shows Candor Chasma.

http://www.youtube.com/watch?v=0WsjeJiAR4E

[Stu]

Stunning!!

Seriously, no vertical exaggeration? Wow... that's Mars' Monument Valley right there...

[Nirgal]

Simply amazing ! your best one yet ...

I've upgdraded my RAM and tweaked the software

obviously you must be using a true 64 bit operating system with > 4GB RAM

EDIT: I just read your video info in youtube:

The animation was created using Mars Explorer and was captured in realtime.

in realtime !! Wow, that is very impressive: rendering such high quality, wide horizon views with what must be several millions of polygons at 30 fps ... I thought this nearly impossible without exploiting hardware/GPU to the limit.
For example, the relatively simple render engine I wrote for my still DEMs (written directly in plain C, optimized code but without GPU support) takes several seconds to render 10 million polygons.

[Mars3D]

Simply amazing ! your best one yet ...

obviously you must be using a true 64 bit operating system with > 4GB RAM

I upgraded from 2GB to 4GB on 32bit Vista (only 3GB is probably availiable because its 32bit).

I have to crop the datasets to 8192x10240 for the DTM and 32768x40960 for the texture. In the case of Candor I only had to remove about 5% of the dataset because it is only just over that size, but for the Mojave dataset I can only use about half of it.

I need about 6GB and a 64bit OS to not have to crop any of the datasets.

[Mars3D]

in realtime !! Wow, that is very impressive: rendering such high quality, wide horizon views with what must be several millions of polygons at 30 fps ... I thought this nearly impossible without exploiting hardware/GPU to the limit.
For example, the relatively simple render engine I wrote for my still DEMs (written directly in plain C, optimized code but without GPU support) takes several seconds to render 10 million polygons.

I have a 8800GTS which can give me up to 250 Million polys/second. I render less than 3 Million polygons per frame as I am using an LOD alogirthm. I get mostly above 60 frames per second. Now I am using the full texture resolution there is some slowdown if I turn quickly because of the amount of texture data that is being loaded from system memory to the graphics card. More RAM on the graphics card should cure this though.

[djellison]

Seriously, no vertical exaggeration?

SEE... SEE... told you I wasn't exagerating my HiRISE DEM movies

(Although Adrian is producing far prettier results, in real time. I hate him..just a tiny bit )