[Peregrine]

So I've been trying to get some starship models rendered, and I'm having some major problems with this. Doing physically accurate modeling of the light is impossible (for the obvious problem of making a scene with millions of kms between the sun and ship). But I can't figure out a way to approximate it in Maxwell, something that was pretty easy in other renderers (direct light, disable falloff). There are two problems I'm having:

1) Getting the emitter energy working properly. What size and energy value should I be using?

2) Getting Maxwell to ignore falloff with distance. Using a scaled down "sun" at a close distance should work, but my weak and close light visibly loses intensity over the length of the ship. Or, if I have the entire ship lit, I have to put the intensity too high and it's way too bright.

A HDRI environment seems like an option, but the ones I've tried don't really give a sharp enough contrast between the light and dark areas.


So what methods do you people use?

[def4d]

nice question, i just made a few test to see

I first tried with the 3dsmax Daylight, and the Physical Sun but it doesn't work with night hours for example

It seems that a tiny, far and strong 1x1 emitter is the best way! And it produces hard shadows

An huge hdr for the environment works too

[Peregrine]

The problem is just how big "far" ends up being. For a ship that might be 1km long, "far" would go way beyond the limits of the software. At least I think so... what kind of ratio of object size to emitter distance do you need to kill the falloff?

[WillMartin]

I just use the Maxwell sun with turbidity turned down to 1 (which you can only do via the plug-in at the moment as far as I know -- if you send the scene to Studio then you get the unwanted "atmosphere" back). Enable sun, Turbidity 1.0, Ozone 0.0, Water 0.0. Would something like that work for you?

Edit (seeing Mihai's post): Obviously you'd need to make the b.g. black in addition to what I said (using what he said ) -- I concentrated on just the lighting aspect in my post.

[Mihai]

You could use Env lighting with just a black or star hdr in the channels, and still use the sunlight from Physical Sky. With 1.6 you can also use Skydome together with the sun.

[Peregrine]

I just use the Maxwell sun with turbidity turned down to 1 (which you can only do via the plug-in at the moment as far as I know -- if you send the scene to Studio then you get the unwanted "atmosphere" back). Enable sun, Turbidity 1.0, Ozone 0.0, Water 0.0. Would something like that work for you?

How do you get it down to 1.0? I'm using 3dsmax 8 with the latest plugin, and the value immediately resets to 1.8 if I try to enter anything lower. And at that setting, I get a really ugly yellow light over everything, background included.

You could use Env lighting with just a black or star hdr in the channels, and still use the sunlight from Physical Sky. With 1.6 you can also use Skydome together with the sun.

That was actually one of the ideas I had, but even if I could find any star HDRs, I'd like more control over the light directions. Is there a way to make my own HDR images that would work for this?

[WillMartin]

I know that some of the plug-ins allow a Turbidity setting of 1.0 (such as LightWave's). I guess I was just kind of assuming yours would be one of them. Yeah, a setting of 1.8 won't work for a space scene from all I've tested; indeed makes things look too atmospheric. Sorry about getting your hopes up like that. Hopefully others here can offer options or workarounds then.

[Bubbaloo]

Use sun and sky, check alpha, and put your own background in in photoshop.

[Mattia Sullini]

Try with MVerta...he is the guru for everything, but for what spaceships is concerned he is a demi-god. Go check his personal site if you do not believe me!

[mverta]

Thank you!


About Space Scene Lighting:

Traditionally, CG space scenes just use a parallel light source with no fills, ambient light, and completely hard shadows. The only problem with this setup is it that looks-wise, no matter how "real" it is, it usually sucks.

The collective "minds eye" for what a spaceship "should" look like is a studio model filmed under studio lighting. That's how we've always seen them represented. There's always at least a touch of softness to the shadows and almost always some fill. The true, "no ambient" lighting look is almost always disappointing, and if you look at any of the major recent sci-fi franchises, you'll see plenty of bounce and other studio-typical lighting effects.

So the bottom line is that I approach these scenes like any other studio setup, and often a small, high-powered emitter or two plus a little fill gets the job done just fine. For principle's sake, I'd like to see Maxwell support parallel/angle adjustable light sources, but I can't say I'm in dire need, and I do a lot of space work.


HOWEVER...

One trick that I have used is to use a Sky Dome, set to pure black, with the sun emitter turned on, and positioned directly above the scene like noon lighting (the least coloration). Then I rotate the entire scene 90 degrees on the X or Z axis. Most space shots have the sun source coming from directly to one side or slightly above or below the horizon line of the ship. Since putting the sun at the horizon turns it yellow, I turn the scene on its side, instead, leaving the sun directly "above" and preserving as much of the intended surface color as possible, and then white balance in post as necessary. Putting the camera above the rotated ship just looks like the scene is aligned normally, but you get the traditional parallel light look.


_Mike

[tom]

Mike, wouldn't embedding an emitter into a tunnel box serve as parallel rays for this purpose?

[mverta]

That's one of the "studio lighting" type things I do sometimes, yes... though the rays aren't truly parallel, which I don't mind. But you can still end up revealing the scale/orientation of the emitter sometimes, especially on big huge scenes. So sometimes it's just easier to do the sun trick. Depends.


_Mike

[sandykoufax]

I'm really waiting to see the result.

[polynurb]

Hello all,

maybe that is also an approach:

I did some tests recently with Parabolic reflectors in rhino, locating a very small
emitter square in the focus of the revolved parabola.

after some tweaking of size ( the emitter was 0,002m in size) and meshing resolution of the parabola (the higher the better for accuracy) I was able to cast rays almost parallel over the distance of 1km.
the parabola itself was 5m in diameter. same as the "red spot" which was hit.
You can still see that the emitter was square.
I used a very low power skydome to show the grid in the render, so there is no scattering of the emitter.
I think that system should be "scalable" with changing the emitter power at the same time; haven`t tested more yet.. like noise clearing etc.

By polynurb at 2008-01-23

[Bubbaloo]

Cool test!