PDA

View Full Version : More advanced Renderer (Part II)



cavalos
02-20-2003, 02:53 PM
And speaking about LW accurate and "brute force" motion blur method, Iīve been wondering if other algorithms can do a better job. Take Final Renderīs aproach for example. They called itīs solution a "True 3D motion blur". So, is Cebas saying that we have a "False 3D Motion blur"?

Psyhke
02-20-2003, 05:16 PM
CIM, what is the difference between Global Illumination and Radiosity, as you understand it? I've been under the impression that Radiosity is one element of a set of ideas called "Global Illumination". In other words, caustics, bounced lighting(radiosity), subsurface scattering, etc. would all be called elements of Global Illumination... as I undertood the term to mean. What's your take on Global Illumination when you mention it in your post?

Halsu
02-20-2003, 05:29 PM
Originally posted by CIM

- Fast global illuminations and not radiosity--it's just too damn slow.
- One-pass, quality antialiasing. The aa passes in LW add ages to render times.
- Quality motion blur and dof that doesn't take tons of aa to look decent.
- Ability to save shadow maps and photon maps.

If you are willing to sacrifice some quality, you can get LW to do reasonably fast global illumination. There's no free lunch. Another option would be faking or baking it.

There's just no such thing a s one pass high quality anti-aliasing. The other render engines that *seem* to do thi in one pass actually take more than one sample / pixel, just like lightwave. The biggest difference is that in LightWave, the user can actually *see* the passes.

Again, motion blur without many passes will always be a hack. It may work nicely visually, but it's not "true" motion blur like what we have in LW.

Why would you wanna save shadow maps?

deemon
02-20-2003, 07:05 PM
>Why would you wanna save shadow maps?

1. Viper (or G2-type plug-ins) could use the saved files, correctly showing shadows

2. No need in recalculating them every time you load the scene

3. Probably 3rd party plug-ins could replace them with their own shadow maps

4. Probably more efficient memory usage, with keeping in memory only those maps that are visible in the current frame

Panikos
02-20-2003, 08:59 PM
Unfortunately Quality and Speed are opponents.
You can get fast results but you wont have quality.

Having both is a challenge, that each and every developer is chaising.

The knowhow of the user is an extra factor.

As far as Global Illumination & Radiosity, they are not the same.
Think of this simple sentence : Global Illumination is the Light, Radiosity is the reply of nature to Light.

You can have GI only.
You can have GI and Rad.
But you cannot have Rad without GI.

Psyhke
02-20-2003, 10:37 PM
Still not getting the concept of GI as a distinct lighting scheme such as radiosity. Can you describe it? Where does the light come from? From traditional light sources, or somewhere else? Help me out.

Panikos
02-20-2003, 11:02 PM
I apologize, I used wrong term for this concept. Its actually "illumination"

Illumination is generated by LW-lights, Skylight, Reflections, Specularity etc.

PhotonMapping is the procedure that collects/accumulates the amount of illumination that exists in your 3d-world.
After this procedure, your geometry is shaded considering your surfacing plus the secondary illumination coming from the surrounding geometry.
If you read the LW-manual, for every surface there is a "collecting" hemisphere that gathers the illumination of the environment.

If you use GI+Rad, every surface attribute plays a significant role, especially Diffuse which is mutually related to the rest of surface attributes. Generally speaking surfacing becomes more sensitive to light.

I highly reccomend to read some books about Light and its properties. Even your elementary-school-physics book can give you a much broader idea.

Psyhke
02-20-2003, 11:23 PM
Hmm. Well I've always understood radiosity in the sense that you describe it as applying secondary illumination using 'hemispheres' to collect data. That's bounced lighting. That's physics. It's clear, understandable, and makes sense. "Global Illumination" on the other hand...

Whenever I've asked this question, I can never get a clear answer in the same way that radiosity can be clearly understood. Alas, GI never seems to emerge (for me) from its seemingly eternal station as a mystical, right-under-your-nose, out-of-the-corner-of-your-eye, through-a-glass-darkly, fuzzy, mysterious concept.


:D

Panikos
02-20-2003, 11:33 PM
Global Illumination is Step-One for Radiosity.

For outdoor scenes, GI is the simulation of skylight, i.e Light coming from Sun plus the secondary reflection from the sky, since atmosphere reflects light. In plain words, light coming from everywhere, in LW terms "Backdrop only radiosity"

For indoor scenes, where there is no direct exposure of your geometry to the backdrop, your lightsources play the role of small suns.

Unfortunately, its hard to describe all these without using tech-terms, since many words mean the same thing in plain english.
The most important, is to understand the principles, and how these are applied in LW.

Psyhke
02-20-2003, 11:48 PM
The way you just described GI is in line with the way I have decided to understand the term (as I mentioned in my post to CIM): as an overarching principal, versus a distinct, singular scheme or algorithm. When people refer to it as a replacement for radiosity, I wonder if I'm missing something.

Okay... moving along now.

Thanks for your comments. :cool:

JMarc
02-21-2003, 08:26 AM
Originally posted by deemon
>Why would you wanna save shadow maps?

1. Viper (or G2-type plug-ins) could use the saved files, correctly showing shadows

2. No need in recalculating them every time you load the scene

3. Probably 3rd party plug-ins could replace them with their own shadow maps

4. Probably more efficient memory usage, with keeping in memory only those maps that are visible in the current frame

I would add this fifth reason for saving shadow maps and radiosity calculations but I would make it the first reason.

5. The ability to use these saved files over a networked render and attain the consistent results minus the redundant calculation time.

cgolchert
02-21-2003, 11:38 AM
Originally posted by Panikos

For outdoor scenes, GI is the simulation of skylight, i.e Light coming from Sun plus the secondary reflection from the sky, since atmosphere reflects light. In plain words, light coming from everywhere, in LW terms "Backdrop only radiosity"



I'm pretty sure this would just be enviroment lighting. (or the effects of...)



Conventional rendering systems include only "local illumination"—the light coming directly from light sources. A "global illumination" approach additionally includes the complex lighting effects that occur between all of the surfaces in the scene, as light is reflected from or transmitted through surfaces.*



This, by definition, would include SSS, radiosity and caustics.




*From the Global Illumination mailing list.

Panikos
02-21-2003, 11:55 AM
well, everybody interprets things from his own point of view.
I am more interested in understanding, than playing with words.

cgolchert
02-21-2003, 12:41 PM
Originally posted by Panikos
well, everybody interprets things from his own point of view.
I am more interested in understanding, than playing with words.


Sure but this is a little different than the tomato/potato pronunciation. This is something the industry had definitions for. :)

Panikos
02-21-2003, 02:57 PM
oky, then ...

to me, "Global Illumination" as I wrote before means "Light from everywhere". If this doesnt make sense, then the industry-guy that decided to name it like this is, ment something else ...

this is my last word on this issue.

cavalos
02-21-2003, 05:30 PM
Back to the original question, theese days Iīm very interested in make a comparision betwen LW and other hig end solutions in DOF and motion blur performance. Final render seems to have a very good algorithm to deal with "true 3d" motion blur.
For some jobs Iīam using Brazil but Iīve never touched Mental ray or Final render so I would love to hear other experiences.

Best
Christian

Panikos
02-21-2003, 05:35 PM
Sorry ...

my last addition, obtained from a Siggraph paper ...

"The computation of all possible interreflections of light in the scene and thus provide a full solution to the rendering equation, are called global illumination"

;)

If I had the choice to follow the developer course, or the artist course, would have been a painful dilema. Terms are often complex. I am happy with my interpretation of them, at least I communicate well with LW :D

cappie
02-22-2003, 10:43 AM
Concearning the radiosity part of the Lightwave render engine:

I'd like to see an option to use "discontinuity meshing" on the rendered surfaces (for more info, see http://www.cs.kuleuven.ac.be/~pieter/bekaert/pieter/professional/mt/)

And, on top of that, maybe some more advanced caustics.. (or atleast a preview mode so its easier to finetune the caustics)

Lightwave's render engine could also use a better network render solution.. a simplified freeware multi-platform render client would be nice... (I still miss Linux support!)

But, thats all (for me).. 'nuf bitchin'.. Newtek, you guys did a great job so far!! :)