PDA

View Full Version : Radiosity Guide updated to 9.3



Exception
08-22-2007, 12:07 PM
Hey dear bedfellows,

I updated the radiosity guide to reflect some of the important changes that came in 9.3.

As always, suggestions are welcome!

cheers,

http://www.except.nl/overig/radiositytute/radiositytute.htm

dballesg
08-22-2007, 12:49 PM
Thanks for your work and the update! :)

David

roboken
08-22-2007, 01:06 PM
Thanks for the update. Nicely written and extremely helpful to those of us who are just getting up to speed on radiosity.

scratch33
08-22-2007, 01:25 PM
Thanks for helping us exception.:thumbsup:

Simon
08-22-2007, 01:34 PM
Thanks!! That's so useful and far better than I've been able to understand from the manual. Especially the part about MES and MPS, I've always been confused about how they work together or when one works while the other is ignored.

colkai
08-22-2007, 01:40 PM
Many thanks Exception

Exception
08-22-2007, 02:48 PM
Thanks!! That's so useful and far better than I've been able to understand from the manual. Especially the part about MES and MPS, I've always been confused about how they work together or when one works while the other is ignored.

That's seriously true, you know, I even don't understand the manual, it doesn't describe the MPS setting at all, I feel tolerance is totally wrong, making me think MPs is not engaged unless tolerance is set, and there's some plain wrong statements in that hint section...

brrr... I'd rather reread my guide ;)

Russel
08-22-2007, 11:21 PM
Thank you, Exception!

I am impressed and appreciate your hard work for creating this excellent tutorial.
-russ

:thumbsup:

zapper1998
08-23-2007, 12:47 AM
Cool ...
Thanks Exception..

colkai
08-23-2007, 02:10 AM
I even don't understand the manual, it doesn't describe the MPS setting at all, I feel tolerance is totally wrong, making me think MPs is not engaged unless tolerance is set, and there's some plain wrong statements in that hint section...

brrr... I'd rather reread my guide ;)

As would pretty much everyone else sir. :)

Matt
08-23-2007, 02:43 AM
Good work (although check your spelling of radiosity in the diagrams!) ;)

Should be in the manual as standard me thinks!

Exception
08-23-2007, 02:55 AM
Good work (although check your spelling of radiosity in the diagrams!) ;)

Radioisty's just fine thanks. (5am... falls off chair onto bed... zzzzzz)

Exception
08-23-2007, 03:08 AM
fix it. sort of. :)

Matt
08-23-2007, 03:38 AM
Radioisty's just fine thanks. (5am... falls off chair onto bed... zzzzzz)

I know how you feel!

starbase1
08-23-2007, 05:59 AM
Many thanks for that clear explanation, much appreciated.

At the risk of abusing your help, I am running into problems with radiosity whereby I get a large scale mottled look very often - see attached image.

I've also had a lot of trouble making metal still look like metal when using radiosity - the specular highlight just vanishes, and if the metal is not clean the reflection is not appropriate. The end result is my metal looks more plastic with environmental lighting.

At the moment I am faking things with a separate specular only light, but that gives off sharp shadows in the specular component.

Any clues for the clueless?
Nick

Exception
08-23-2007, 10:55 AM
Any clues for the clueless?
Nick

the mottled look is because or too little samples. So, you need to increase the alloweable distance with which they will blend (MES/MPS combination) or increase the number of rays. Most likely increasing the number of rays will be most effective. Are you using MPS, if not, you should, and do you have tolerance set? Use 0.3, that's a nice tradeof for quality/speed.

If your radiosity starts to look too bland because you want speed it helps to add an occlusion shader to provide the details.

Concerning reflections/specular... I'm afraid I don't quite know what you mean. do you have an example for those?

starbase1
08-23-2007, 03:51 PM
Thanks, I was at work when I started looking, but have since got home and tried some of the tutorial stuff - you are right, increased rays helped. I got much better quality and a shorter render time from your tips, thanks!

One general thing - like most tutorials and tips, yours concentrates on room scale objects - my example has a landscape which I suspect is rather different. From what you said, I took the large ground plane, made it luminous, and turned it off for radiosity. It's not actually in a position to receive or cast shadows, but the large size may well have been slowing down radiosity a lot.

As for the metal, what I mean is that the standard way to make metal look metallic is to go for low diffuse, high specular, and tint the specular to the surface colour. With radiosity you don't get a specular highlight, so old style metal surfaces no longer look very metallic.I'll try and knock up a pair of comparison images to show what I mean if that's not clear...

Thanks again,
Nick

starbase1
08-23-2007, 04:17 PM
Oh, and thanks you your tutorials - it never occured to me that just switching to the advanced camera without playing with the fancy settings could double the render speed!

I must be rendering 4 times faster at double the quality now... Anyone who thinks radiosity is too slow, take a look!
:thumbsup:

A Mejias
08-23-2007, 10:38 PM
Thanks for the update. Very helpful!!!

Exception
08-23-2007, 11:14 PM
As for the metal, what I mean is that the standard way to make metal look metallic is to go for low diffuse, high specular, and tint the specular to the surface colour. With radiosity you don't get a specular highlight, so old style metal surfaces no longer look very metallic.I'll try and knock up a pair of comparison images to show what I mean if that's not clear...

Thanks again,
Nick

Well, I make metal usually without specular and with blurred reflection. I know this is quite the render hit but using adaptive sampling really narrows it down to the grainy areas, and if it's for animation, you can get away with quite a bit of grain... also, using a bump map and reflection works a bit faster than using blurred reflection but it might flicker a bit, depending on your settings.

Make sure you set your ray recursions as low as you can, sometimes you can get away with just 3 or 4, and when you use reflection in those circumstances it's really not a bit render hit.

Specularity is really just an old fake way to do fast approximation of reflections. You couls also of course just add a few lights, turn them off for diffuse, and voila, specular highlights. If you then also turn off directional rays for radiosity you get 0 render hit! :)

aquinde
08-24-2007, 02:14 PM
Excellent work Exception - and much appreciated.

starbase1
08-24-2007, 03:42 PM
Here's an example using the tutorial techniques for anti aliasing combined with radiosity... Before I used them it would not render at all, (which I now realise was down to a ground plane with 20 mm between evaluation points!)

Now it's 2.5 minutes.

Exception
08-24-2007, 03:56 PM
Yeah you love me!

:) In cases like this MPS works quite well too. It makes those incidents with infinite planes have more samples near the camera than far away. Works really well.

Just a little addition here.. I've ran into financial trouble. I'll be adding a paypal button to the tutorials site. Strictly voluntary. If you won the lottery or something. Cheers :)

Exception
08-25-2007, 10:48 PM
Just a little addition here.. I've ran into financial trouble. I'll be adding a paypal button to the tutorials site. Strictly voluntary. If you won the lottery or something. Cheers :)

Seems i killed the thread with that one... :)
Allright cowering LWers in the corner, I know most of us have it tough, so don;t sweat it :thumbsup:

eagleeyed
08-25-2007, 11:19 PM
Hey Exception, I tried to give you a donation (aren't I good) but I dont think you have put the right code on the button. It just goes to the PayPal homepage. It usually supplies code that directly links to your account.

Anyway, once you get that fixed I will give you something, you have helped me alot.

Exception
08-25-2007, 11:42 PM
(aren't I good)

Yes you are!
Angels will smile upon you.

thanks for bringing this to my attention. I've updated the code. Let me know if it doesn't work for you.