My radiosity test

[sdds]

Hey all. Been working with some radiosity today just need some criticism.

 

The render setting are:

 

accrucy = 95

Prepass = 1/1

Diffuse Depth = 1

Stochastic Samples = 500

Min = 2

Max = 10

 

I also used AO, total render time was 25mins at a resolution of 700p by 500p. I think it looks ok but it lacks quite a lot, not sure what it is though.

[sdds]

anybody?

[edub]

I usually have max/min set to something like 10/100.

[STRAT]

well, for the quality of lighting and the general lighting atmos, thats up to your artistic abilities. but those gi settings could be optomised.

 

try -

 

accrucy = 70

Prepass = 1/3

Diffuse Depth = 1

Stochastic Samples = 500 (i'd maybe try 750-1000 if you want to experiment)

Min = 50

Max = 350

 

and a slight touch of AO

[STRAT]

I usually have max/min set to something like 10/100.

 

how long is a peice of string? you cant generalise with settings. every scene is different. 10 for mins is awfully low (as is 100 for maxes). and dont worry about a mathematical proportion between mins and maxes either. it doesnt matter.

[sdds]

Using the settings you suggested strat, also changed the lighting a bit. Used only 2 omnis this time with area shadows, because I am used to faking used about 7 lights in last render.

[sdds]

Ok I am really struggling with this, my scene looks dead do I need a diffuse depth of 2 or something?

[STRAT]

nope. 2 will only make it brighter still, and drastically slow rendering down.

 

approach lighting from a different angle -

 

try just using the sky object as your main gi caster, with 0 lights. then, individually, insert local or ambient omnis where needed.

 

the second image you posted here looks the nicest so far, but even that looks like too many lights. also, to save render times, go easy on area shads and dof.

 

you're still deeply in the habbit of faking gi, ie, using a fair few omnis. take them out and start again. less is more.

[sdds]

Thanks for your input strat.

 

Its ok to use ambient only lights then with gi? That will defiantly remove some of the grayness. DOF is done in post so thats fine. Sorry to keep bugging you but you've got all the answers.

 

I have used 1 area shadow, do you use them at all in your renders?

Also I would really like to see one of your interior renders if that is at all possible.

 

Thanks strat

[STRAT]

ok, here's a few interiors i've recently submitted for a health spa with some lighting descriptions. note, the depth and vividness of the lighting comes from your point lights and fill in lights, NOT the quality of the gi settings. the gi and ao quality just aides the main lights. thats where the art is.

 

img01 -

 

 

took slightly longer to render due to area shadows and blurry reflections.

in here is 1 main sun light shining in, a gi emitting sky object, and localised omnis and spots where needed. no extra ambients in here.

 

 

img 02 -

 

 

again, 1 main sun shining in and a gi casting skyobject, and general local omnis where there'd be actual lights in the scene.

 

 

img 03 -

 

 

this has 1 sun, a sun and a gi sky object and the usual point omni lights, but i've used 1 omni as a general ambient light too placed in the middle of the scene. allot of light and colour in here is being cast from the strong value of the swimming pool surface gi emmittance.

 

 

each image has an accuracy of about 70, a dd of 1, stoches of about 500, mins 50-70 and maxes about 200-300. there's also an AO pass to help with things. you can also add render tags to individually tweek gi properties, as you can with individual materials.

 

these rendered @ 3000 pixels at about 2-3 hours each.

[sdds]

Thanks strat, really like your images, gives me something to aim for. Will have a play with what you suggested and see how I go.

 

cant thank you enough m8.

[edub]

how long is a peice of string? you cant generalise with settings....

 

 

Double the length from one side to the middle

 

just funnin'

 

I hear ya... I was just too rushed to elaborate on my previous post , when I mentioned the 10/100 (the 1:10 ratio is a legacy from the Fluff tutorials, btw).

 

I'm actually trying out your settings a.t.m., and I'm liking what I see, so far.

 

i wish there was a way to intuitively know what the interrelationships are between DD- samples-min/max... but I guess that's where the artistry comes in (as in having done trial&error [x8sideways] ).

 

LOL - I'm juiced on C2H6O and C8H10N4O2!!!

[STRAT]

(the 1:10 ratio is a legacy from the Fluff tutorials, btw).

 

yup, i wouldn't take too much notice of that. alot of peeps put faith a min max ratios, but in the real world you work to the scene. is a ratio cleaner/faster? i doubt it. i havent seen evidence of it in my years of using it. more a hinderence to me than an advantage. some scenes i render have my mins are higher than the maxes, making the maxes irrelevant.

 

 

i wish there was a way to intuitively know what the interrelationships are between DD- samples-min/max... but I guess that's where the artistry comes in (as in having done trial&error [x8sideways] ).

 

experience i suppose. i'll ALWAYS use a dd of 1, to keep render times normal. but then infil omnis must be used to compensate for lack of bounces. but inserting these omnis is an art in itself.

Generally, when the accuracy is about 70%, a stoch sample rate of 500-1000 should suffice. as for mins and max, for mins 25-100 is usually ok, but maxes could be anything from 100-750 maybe.

as you say, trail and error in your scene.

 

those settings are for 'normal' gi usage too. if, forinstance, you have alot of gi emmitting objects or area lights or things like that you'll need ALOT more samples to accommodate.