Rendering In Radosityanyone have any Tips and tricks

[Paul B1974]

Hi Guys,

I have posted in the feature request forum about the fact I have found the Lightwave render engine is a little Slow..

I render mainly Architectural scenes so Radiosity/Global Illumination is essential for getting the right effect... These are mainly still images so a tyipcal render is between 4000-7000 pixels


So does anyone have any suggestions about making it quicker or any experiences on the subject .... please I'd like to here about it

And please no comments about getting faster computer that ain't going to help much ( I have a dual Zeon 2.8gig Machine : )

Regards

Paul

PS I Have the Lightwave 7.5 lighting book which is good by the way :)

[Dodgy]

You should really look at the different settings. Some will drop the render times (but keep up the quality) much better than others.
Tolerance seems to make a big difference in time, so you might want to fiddle with that first, but like anything you have to play with the settings to suit.

Also, negative intensity lights can be used to suck out a colour from the scene...

[kevman3d]

I've had FAST and good results from a few things.

Namely, dropping the rays per evaluation really low (3x9 is usually good), making sure 'Shading Noise Reduction' is active and using an enhanced AA option (or an AA option with NO Adaptive Sampling).

Why AA? And why NO adaptive sampling? Simple - this will force a different GI result each AA pass for your radiosity - It may look a little 'noisy' each pass, but once all AA passes are complete, the final result is a blend of ALL the passes together... The slight difference between them all helps 'smoothen' the noise out quite well...

You can hone down the performance of the Radiosity so its pretty quick this way. Even with the extra AA passes, you may find its hours faster then doing a high-sample Monte-Carlo render (and finding out after 8 hours that it looks awful).

Also, use a little 'Ambient' value in the GI panel - Like 1,2% only though. Ambient works in GI in a different way to the 'fake' effect it does normally... A few people have pointed this out for better looking radiosity.

I've also had some pretty snappy results from using Interpolated mode radiosity with a 0.5 to 1.0 tolerance. However I do find this mode a lot more 'splotchy', hence I tend to usually stick with Monte Carlo...

Finally, if its a walk-through animation you want, you can also Bake your radiosity solution back onto the textures of your 3D environment. Providing its just a moving camera in the scene, this should work great!

Oh, yeh, and get a faster PC! :) lol! Just kidding!

[richpr]

Make sure Radiosity Caching is on for the multiple passes...

Surface baking for fly-throughs, although I don't expect you need this since your images' resolutions are that high...

The combination of interpolation, interval, the evalutation matrix (5x20, etc) will improve quality, but also adds to the render time...

You could look into network rendering... and split the image across multiple machines for rendering... although that has it's own issues (can be difficult to set up or to keep running smoothly)...

[Dodgy]

MMM some nice hints there dudes :)

Should you use cache rad with your method kev? If you want to force a recomp every AA pass, should cache rad be off???

[richpr]

I would assume so, or else it won't be recalculated on every AA pass... With caching it shouldn't be recalculated for every pass...

[richpr]

Pages 399-428 (Chapter 9: Realistic Lighting Environments) of Inside Lightwave 7 touch radiosity, caustics and surface baking...

[Zafar Iqbal]

You could also use Backdrop Only under GI settings to drasticly speedup render - combine it with normal lightsources to breakup even lighting from all ankels - and use Backdrop Gradient to set the intensity for the GI.

Dont be afraid to make a gradient that wont do much for the picture itself, because you can always mask it away and replace it with something nice in Photoshop or other 2D program.

If you are indoor, then you can turn up the Ambient light, but in the worst case you have to use alternative way of calculating GI (Monte Carlo or Interpolated).