GI example chart

Started by N-drju, April 12, 2017, 09:08:45 AM

Previous topic - Next topic

N-drju

Since wiki editing is disabled, I decided to post something in here, though I am aware that this section of the forum has not been used for some time.

Below is a GI settings example chart. It is intended as a resource for Terragen users who would like to obtain a clean and self-explanatory reference chart for the global illumination settings available in the software.

A similar resource, created by Martin a.k.a. dandelO was available back in time but unfortunately, disappeared from the web much to my (and probably others too) dismay. I have gone to great lengths trying to find this GI tutorial that he made. Unfortunately it is nowhere to be found.

Therefore I decided to recreate his work using the similar scene and setting-by-setting approach that dandelO used. I have been preping a very similar resource for my personal use so I thought it would be nice to post it in here. It is now up for grabs for you. I hope that it can help understand GI and grasp some of its underlying principles or at least serve as a per-case reference if you want to check your settings before you go and hit that render button for the last time.




The test scene against which various settings are being checked consists of the following features and objects (left to right):

- "Test room" made from five thin, grey cubes.
- Red, non-reflective, non-luminous sphere sitting atop a flattened, glass sphere object.
- Box with luminous sides.
- Grey and slightly reflective "plutonium sphere".
- Sand-colored, non-reflective and non-luminous "mentos sphere".
- Overhead light (sunlight with about 1.5 power).

The render settings are:

Resolution: 1275x717
Detail: 0.68
Anti-aliasing: 4





Examples:

[attach=1]
As you can see, there is hardly anything to say about "global illumination" here... Light does not seem to affect anything that is not directly below and shadows are black. With settings zeroed out, only direct illumination is taken into account and all light-shadow intricacies are being omitted. Good for experimental and bootleg renders.

Render time: 0:06:41
GI cache detail: 0
GI sample quality: 0
GI blur radius: 8
Supersample prepass: NO

GI surface details: YES
Occlusion weight: 1
Bounce to the ounce: 1
Radius: 24
_____________________

[attach=2]
Default Terragen 3 settings. For the most part, these settings are well-balanced and offer decent lighting effects overall. However, as you can see if the light sources and surfaces are "tricky", the lighting effects start to go crazy ape bananas. This happens because, there are simply not enough samples (or light-calculating dots as I like to call them) to correctly capture all the lighting in our picture.

Render time: 0:06:24
GI cache detail: 2
GI sample quality: 2
GI blur radius: 8
Supersample prepass: NO

GI surface details: YES
Occlusion weight: 1
Bounce to the ounce: 1
Radius: 24
_____________________

[attach=3]
Another example of 2-2 render with increased blur radius... which is not always the way to go as you can see from an increase in unwanted illumination areas all over. A render with a caption that says - "More is better, not always"!

Render time: 0:06:28
GI cache detail: 2
GI sample quality: 2
GI blur radius: 16
Supersample prepass: NO

GI surface details: YES
Occlusion weight: 1
Bounce to the ounce: 1
Radius: 24
_____________________

[attach=4]
As you can see, those settings are quite decent on overall. Notice that this time, light coming out from luminous cube has been accurately cast on the red and grey spheres. There are still some unwanted effects on the walls but they are not as striking as before.

Render time: 0:08:33
GI cache detail: 3
GI sample quality: 3
GI blur radius: 8
Supersample prepass: NO

GI surface details: YES
Occlusion weight: 1
Bounce to the ounce: 1
Radius: 24
_____________________

[attach=5]
Now we're talking! As you can see the cast and bounced light is now almost flawless. Some mistakes do appear but more as a result of the specific nature of that scene rather than a GI lighting error. Notice that light bounced of the "mentos" and "plutonium" spheres is visibly represented on the room's walls now.

Render time: 0:08:35
GI cache detail: 3
GI sample quality: 4
GI blur radius: 8
Supersample prepass: NO

GI surface details: YES
Occlusion weight: 1
Bounce to the ounce: 1
Radius: 24
_____________________

[attach=6]
The 4-4 settings offer a very good quality as you can see. These can well be used if you are looking for some extra quality with regards to your lighting.

Render time: 0:10:28
GI cache detail: 4
GI sample quality: 4
GI blur radius: 8
Supersample prepass: NO

GI surface details: YES
Occlusion weight: 1
Bounce to the ounce: 1
Radius: 24
_____________________

[attach=7]
These, are very high-end settings that you will probably rarely use. The render is very clean in terms of lighting as you can see but it also takes considerable time. Almost twice as much as with the previous settings!

Render time: 0:21:14
GI cache detail: 6
GI sample quality: 6
GI blur radius: 8
Supersample prepass: NO

GI surface details: YES
Occlusion weight: 1
Bounce to the ounce: 1
Radius: 24
_____________________

[attach=8]
If you think that overheating and melting down you CPU maxing out the GI settings is worth giving a try, you will probably end up with something like this. These settings are ridiculously high and render takes forever to complete but you will probably agree that the image is flawless in terms of illumination. Good for very, very, very high quality projects with complicated illumination system.


Render time: 0:43:03
GI cache detail: 8
GI sample quality: 8
GI blur radius: 8
Supersample prepass: NO

GI surface details: YES
Occlusion weight: 1
Bounce to the ounce: 1
Radius: 24





Well, that is all (for now) but given some free time I will probably also upload more diverse settings with supersample prepasses, radius changes etc. One thing to remember - I am not saying that any of the settings presented above are inherently "bad" or "good". It mostly depends on the situation and of how much and complex light you have in your scene. However, they can offer you some insight of how given settings may influence the render on average. The choice of option and "mode d'emploi" is yours to make.
"This year - a factory of semiconductors. Next year - a factory of whole conductors!"

luvsmuzik

I hesitate to comment, but thank you for this! I too slap my forehead looking for old files that may even be on another processor lurking in a tutorial folder.
You may delete this comment or I will to keep the information foremost. :)

Dune

I remember the old illumination box, might even still have it somewhere, but this is a very good replacement. It will surely benefit a lot of users.
Btw, did you also do a GI detail 4 and sample 3 test? Might be interesting to put that next to the 3-4 test.

N-drju

#3
@ luvsmuzik - You're welcome! I think there is no need for that! I intend to post "portions" of GI examples and I have nothing against them being separated with comments and suggestions. Actually... I think I will need to reload the images... I posted them using another computer and now that I look at them I see that .jpg conversion on that machine sucks big time! :-\

@ Dune - Yes, I am planning to add more cris-crossed GI examples in a week or so. I might even add GI test on landscape renders vs. "box"!

________________________

Awww, shucks... I messed one attachment up. One example less for a moment will reload them soon.
"This year - a factory of semiconductors. Next year - a factory of whole conductors!"

Oshyan

Wiki editing will be back soon. I hope at that time you will add this information there as we do not intend this area of the forums to be used long-term. But in the meantime thank you very much for the contribution.

- Oshyan

N-drju

Quote from: Oshyan on April 12, 2017, 03:30:01 PM
Wiki editing will be back soon. I hope at that time you will add this information there as we do not intend this area of the forums to be used long-term. But in the meantime thank you very much for the contribution.

- Oshyan

Yes, thank you Oshyan. I was somewhat confused and uploaded the chart here in the meantime. Hope it did not cause trouble. I promise I will move it to where it belongs! I will have time to resave the pics in better quality format too.
"This year - a factory of semiconductors. Next year - a factory of whole conductors!"