Basic question about render preview

[clagman]

It seems as though they have worked the render preview so that it doesn't add to the render time. I did some test renders and it made no difference even with complex radiosity scene (1 hour + with 8 CPUs) on or off. Does anybody else bother with this thing? I mainly use it because I like to have the visual of the 7 frames displayed, it won't do this without enabling the preview.

[Exception]

What exactly is your question?
It doesn't add to the render time, correct. It does use more memory.

[clagman]

Sorry, you have it exactly. Thanks for the answer. I remember when you had to disable the preview to cut the render times down, so I guess it's been a while. Thanks again.

[Exception]

Yes that was several years ago, but you are right. It probably still adds to the render time, but so little that it hardly matters.

[clagman]

I'm still going bonkers over all the new stuff that 9.3 has vs. the last time I used it at 6.5b. Lots'o catching up to do (and learning the entire cs3 video workflow at the same time).

[Exception]

I wrote some guides on the new AA sytem and GI in 9.3, if that helps. They're on my site under resources.

[clagman]

Thanks, it's nice to see people putting their knowledge out there for others. I think the first thing I delved into was AA and then the new radiosity. I've seen a lot of complaints about speed for MC radiosity renders but compared to what I'm used to it seems really fast.

[Exception]

You can get very speedy renders out of Lw as long as you know how to use it. Same goes for all other render engines. Kray is faster, but only if the settings are correct. Fprime is sometimes faster, sometimes slower, depending on the situation. For me, I can get Lw native to render faster than Fprime in most cases.

[Iain]

You can get very speedy renders out of Lw as long as you know how to use it. Same goes for all other render engines. Kray is faster, but only if the settings are correct. Fprime is sometimes faster, sometimes slower, depending on the situation. For me, I can get Lw native to render faster than Fprime in most cases.

I agree that LW can render faster than FPrime but I think this is very much an animation vs stills question. LW only beats FPrime when using Final Gather (at best) and more and more I think FG is a real compromise in overall light quality. You can only get away with that in animation.

FPrime's simulation of LW's Monte Carlo is so much better than FG that it shouldn't be written off just because of marginal speed benefits.

[clagman]

Ya got me on that. I've yet to use FPrime (although I've seen it a lot). Even if FPrime wasn't the speediest render I would love to have it for previewing. I wish Viper was more like it.

At any rate most of the radiosity rendering I do doesn't require a lot of detail (like an outdoors setup) so FG or even BG rad is fine (I guess animations yes not arch-vis stills). For rendering stills I suppose I could use MC and get TequilaScream to split the frame up and network it out. Get some nice high rez that way.

[Exception]

FPrime's simulation of LW's Monte Carlo is so much better than FG that it shouldn't be written off just because of marginal speed benefits.

No true, Fprime is an institute... I'm just saying, there's so much complaining about LW's speed lately from people that really havn't spent the half an hour in trying to find out why. Fprime for all its qualities also has its shortcomings.

[v1u1ant]

Im rendering a 300 frame scene with 1831047 polys with one area light and one directional light, out to 720x576 with 1.067 far on a G5 quad core mac with following setup:
Classic Cam
3-pass AA
Ray Trace Shadows, Reflection, and Transparency.
using lightwave renderer and getting an average of 6.30 minutes per frame. Well its on frame 294 now after 20 hours 51 minutes etc etc.



Is that a good time? Would i get faster with FG+interpolution??

I get a fair amount of noise as you would imagine with only 3 passes.


I did have some nodal surfaces but remove them afetr watching my render times rocket. Would FG solve that?

[archijam]

Why classic camera, may I ask?

j.

[clagman]

I think the classic camera renders slower than any of the new cameras (like the perspective camera for instance).

[Dionysios]

I wrote some guides on the new AA sytem and GI in 9.3, if that helps. They're on my site under resources.

I just wanted to say thanks a ton. Your GI tutorials and sample scenes really helped me figure it all out. Thanks again.

[v1u1ant]

Really??? Which camera would be recommended these days then?

[archijam]

Really??? Which camera would be recommended these days then?

I think Exception's tutes mention this ... some of the new AA features etc are only available for the new cameras ... perspective is a good start.

j.

[Iain]

With the exception of Perspective, I've found the new cameras to be slower than Classic.

[Exception]

Really??? Which camera would be recommended these days then?

All of the cameras except for the classic camera use the new render engine. This new render engine on first glance looks very much like the old one but was pretty much written from scratch.
The new cameras handle complex scenes much much better. A scene with a million polygons or something in classic cam might take 10 minutes, it might only take perspective 1 minute. It all depends. The heavier the raytracing (reflections, refractions transparency etc) the faster the new cameras will be.
for flat 2D work and simple scenes classic is usually still faster.

The new AA system of perspective cam is usually faster than classic, and enables the much improved new motion blur and DOF, as well as more advanced AA settings.

FG+interpolated is going to be extremely much faster than just plain FG. However, you will need to learn how to control the settings, or you might get ugly scenes. On the other hand, there's no grain.
I think your scene might be able to go down closer to 2 minutes, but not knowing the content, that's hard to say.