Normal Map Rendering

Started by JDex, February 02, 2007, 06:04:00 PM

Previous topic - Next topic

JDex

Basically Terragen cannot render the format I require (dome shader - cubic projections are unacceptable) so I am rendering in another application... what I have done is exported a .ter file from a GIS app and have imported it into bothe TG2TP and XSI.  I'd like to render a normal map representing the difference between the base .ter mesh and the added fractal detail in camera space... I could then project this normal map with a Camera Projection in XSI and get the increased details that I need.

Can this be done in the current release?  Can anyone describe or illustrate how to approach it?  I've tried using the various shaders related to normals (Compute Normals, Get Normal, Get Normal in Geometry... etc) alone and in cobination with each other with various setups... but can't get a vector to render.

Thanks In Advance.

Will

Im sorry to say this but I don't think is posable but I've been wrong in the past. Though i'll keep an eye out for you.

Regards,

Will
The world is round... so you have to use spherical projection.

Oshyan

I don't know of any way to get a normal map out of TG2 at present. You can export geometry using the LWO Micro-exporter (techniques described elsewhere on the forums), but for decent detail you'd end up with a pretty dense mesh. And you'll never be able to duplicate TG's surfacing ability or (particularly) atmospherics in XSI, I don't think.

- Oshyan

JimB

Try setting up lights from different angles and then use Photoshop trickery. There are tutorials and small apps on the web showing how to do that for real objects.
http://zarria.net/heightmap/heightmap.html

Messy, and not ideal, but it might help depending on what you can do with lights, or preferably a sun that doesn't cast shadows, rendering multiple exposures.
Some bits and bobs
The Galileo Fallacy, 'Argumentum ad Galileus':
"They laughed at Galileo. They're laughing at me. Therefore I am the next Galileo."

Nope. Galileo was right for the simpler reason that he was right.

MooseDog

chk this out.  converts greyscale bump-maps into normal maps:

http://www.crazybump.com/beta/download.html

discussion of it's use and evolution here:

http://boards.polycount.net/showflat.php?Cat=0&Number=149005&Main=148897

seriously cool stuff, if applied to an overhead shot of a tg terrain.

MooseDog

#5
proof of concept.

from the b/w orthographic shot, i extracted the normalmap.

if you've got your terrain uv-mapped in xsi (and i'll assume as much ;) ), you're off to the races!

JDex

Thanks everyone for your replies...  I've been experimenting with alternate ways of getting the detail I need, including Moosedog's approach bypassing T2 altogether, I'll quickly see if I can figure out how to do JimB's method in T2 as well...

Unfortunately we'd really like to integrate T2 into our project, but keep finding it wanting in the basic functionality that the project requires.  If there was a dome camera shader today (which we would have happily paid to have dev'd), or a way to quickly get these normal maps, we would be justified in continuing to persue the app (which means a couple artists bug hunting and ideas based on production), but we're finding that its -for now- a dead end for our needs.

The results we're getting in XSI on the atmospherics and volumetric clouds coupled with rendertimes, by-far surpass the results we can get for T2 at this time... the land has proven somewhat less forgiving, but we'll get it there.

I'll see if this last weekend of testing can net a reason to keep T2 on this project, and if not I'll see if I can play with it from time to time on personal projects.  Hopefully when this project is done, I can post some shots done in XSI|MR reflecting what we are trying to do.

JimB

JDex, I'm also an XSI animator and also a matte painter. It should be no problem to create a .tgc that could accomplish most of what you want within TG2 for creating the images for the normal maps at the click of a few buttons.

Alternatively, why not use something like Johnnyboy's excellent XSI to Chan script to get the XSI cam into TG2, render the landscape in TG2, and just comp it in among the XSI elements using exported per-frame LWO's as a "3D mask" (black constant landscape) for the atmospherics within XSI? Come to think of it, the TG2 landscape render only needs to be reprojected onto the XSI imported terrain's polygons as a texture camera projection, which essentially combines the two apps in a roundabout way.
Some bits and bobs
The Galileo Fallacy, 'Argumentum ad Galileus':
"They laughed at Galileo. They're laughing at me. Therefore I am the next Galileo."

Nope. Galileo was right for the simpler reason that he was right.

JDex

Hey JimB,

The reason that that won't work in this case is because we need 360deg field of view for all the landscapes... the project is a planetarium dome that involves landscape with character/vehicle/particle interaction...  I can do some trickery if we can render the up-detailed normal maps directly from T2, or conversely if we had the luxury of rendering with the dome distortion already there, I could do more on the compositing side... but I'm already far behind on R&D'ing T2 with this project, and will have to have an appreciable use for it on Monday, or go back to the plan that was in place before T2 was released.

T2 is in beta essentially, and normally we would not have even tried to use a beta app on a project, but T2 has so much potential that we decided to see if we could find a way to make it work with this project because we'd like to help contribute to where it's going... unfortunately this project comes with a handful of unordinary variables that make T2 less useful for us.

Gonna spend a few more hours trying to setup a lighing based normal maps solution, and then get started on the old way. :(

JimB

Ahhhh right. Now I see what the big problem is. Good luck with it.
Some bits and bobs
The Galileo Fallacy, 'Argumentum ad Galileus':
"They laughed at Galileo. They're laughing at me. Therefore I am the next Galileo."

Nope. Galileo was right for the simpler reason that he was right.