PDA

View Full Version : Split Radiosity Rays From Reflection



Panikos
08-08-2005, 08:20 AM
Its really awfully restricting to have all rays (Radiosity, Reflection, Refraction) into a single button Unseen by Rays.

You want Brighter Radiosity, you burn the reflections.

PLEASE, put an Unseen by Radiosity button. It will cut the chain of this slavery.
:thumbsdow

ingo
08-09-2005, 02:28 AM
I agree loudly ;) I just want to use one skysphere for reflections and another one for HDR-lighting. Pleeeeeeeeaase !

Lewis
08-20-2005, 09:37 AM
I agree also :). We need radiosity On/of per OBJECT and per SURFACE basis - also adjustable in scene editor (for larger secenes wiht many objects).

Weepul
08-20-2005, 03:47 PM
Great minds think alike? :D

I too would like this option.

pixym
08-20-2005, 10:12 PM
So do I,
Please Newtek listen to us.
:agree:

toma
08-28-2005, 05:59 AM
:dito:

toma

JML
08-28-2005, 12:20 PM
:agree: me too

wacom
08-29-2005, 12:36 AM
This could be answered with the new nodal system if they have a ray-type spliter node. Then we could control it on a surface by surface basis too!

Panikos
08-29-2005, 12:47 AM
Wacom, I prefer to click a switch than tweak 20 surfaces :cursin:

Additionally, I think you missed the point.
We are not talking about Radiosity originating from Backdrop, but from Geometry. And currently Geometry is accounted in all (Radiosity, Reflection, Refraction) A Nodal shader cannot exclude an object from the reflection calculation, this is impossible unless the renderer calculates the scene twice, one with the involved geometry and one without it.

:2guns:

wacom
08-29-2005, 01:55 PM
Wacom, I prefer to click a switch than tweak 20 surfaces :cursin:

Additionally, I think you missed the point.
We are not talking about Radiosity originating from Backdrop, but from Geometry. And currently Geometry is accounted in all (Radiosity, Reflection, Refraction) A Nodal shader cannot exclude an object from the reflection calculation, this is impossible unless the renderer calculates the scene twice, one with the involved geometry and one without it.

:2guns:

Well I was trying not to drop program names...but in XSI you CAN do it at the shadder level, and using things like overrides etc. it's easy to set it up for your surfaces. And no it doesn't seem imposable- I've seen it with me own eyes! The nodal shadders don't have to compute the whole scene first, just the branches of that tree before it sends it off to render.

There at least two to three ways to do this in XSI, but I'm just a noob with it so I can't say how many other ways there are. And yes having it on the surface level can be a pain, but lets say you have a steel tube chair with leather strapping...wouldn't it be nice to turn off the radiosity reflections for the leather and not for the steal tubing?

Now getting MR to look as nice as LW's renderer in as short amount of time...that's another story :)

Is this not what you're looking for? In the first pick the floor is seeing the radiosity from the sphere at 100% in the second it is .6% but it still sees reflections etc and the radiosity on other objects is reflected. In the third I have the floor being highly reflective, but the radiosity is calculated seperatly for the floor hence the reflection is seen in the ball, but the radiosity is orange in color and it's intensity is being controled seperatly. Again I'm a noob and maybe I'm not understanding you...but I just thought with a nodal system you could get what you wanted...

Panikos
08-29-2005, 02:39 PM
Wacom, thanks for the reply.

You know, in examples with floors and balls its easy to conclude or misconclude things. I dont underestimate other apps or your knowledge or your constructive will. Thank you for this !
What I face with LW is in real application. If I use a highly lum sphere to illuminate a scene, brightening it works well, spreading around a lot of diffuse light. However the way LW works (where all the rays are grouped together) its impossible to have this bright radiosity without burning the reflections.
LW Reflections are ugly, no control at all. Plugins like G2 or Hypersmooth do not help because the Lum Ball is part of the scene and its not a Backdrop Image.

If you reduce the % of Reflectivity of a Surface, you weaken the reflected ball, but you loose the reflections of the rest of the environment due to the huge difference in brightness.

What I've asked is to simply separate rays so you have a global switch to select which geometry is used by Reflections and which geometry is used for Radiosity. As simple as that.

Unless you encounted this awful problem you cant understand how @#!#%$^ bad it is. Any shader solution to this not only it doesnt help at all, but even IF it helped it would have been a pain in the ***** to tweak all reflective surfaces of a scene and they arent 40..50.. 60 (in my long term project) but more. :devil:

Bloody ****, I needed this functionality like 2 years ago.
And I compromised badly expecting with hopes "Oh, well maybe in the next update" and that never happened.

I've been asking for a Camera Shader System since LW6.5 and its promised in LW9. :thumbsdow

JML
08-29-2005, 02:50 PM
we should just be able to turn off or modify radiosity intensity per object AND per surfaces.

JML
08-29-2005, 02:54 PM
it's funny how lw radiosity takes forever with raytrace reflection,
and most of the time you don't see it (on heavy transparent raytrace surfaces.)

let's say a building with lots of reflective glass , you can not see
the radiosity in a 90% transparent glass and it will take forever in LW right now.
being able to turn off radiosity for all the glass in the scene would save
so much time in rendering and for the same end result in that case..

Panikos
08-29-2005, 02:58 PM
Jml, the channel that is sensitive to Radiosity is Diffuse.
Usually for Reflective/Refractive Surfaces Diffuse is low.
The Reflective/Refractive surfaces are called Non-Lambertian and they account for caustics. :D
Diffuse only surfaces are called Lambertian. "natural surfaces create scattered reflection in all directions. If the energy reflected is exactly the same along all directions, the surface is said to be Lambertian. In this case radiance and reflectance are both independent of the viewing angles qr and jr."

JML
08-29-2005, 03:01 PM
are you sure?
so why a surface with radiosity will take 10 times longer if it has reflection on it ?

Panikos
08-29-2005, 03:24 PM
In reality, all surfaces are reflective. Stand in a Dark room, you have some things surrounding you with colours but you cant see them cause there is no light to reflect. Camera and our eyes see reflections of light and NOT the objects. Your senses are trained to realize their volume, texture, smell, taste, sound. Your eyes gather light reflections, only one aspect.

Diffuse is like extremely blurred reflection that is simulated like Diffuse+Radiosity on top, i.e the Cheap way to calcute to avoid the cost of rendering an extreme blurred reflective surface.
Reflective surfaces need to reflect the environment in focus without any blur, so they have to trace the surrounding environment.
Specular is again a simulation of reflection with out of focus Reflection of a bright light and zero reflections of the environment. Specular again is a cheap way of faking it. (BTW, currently the way that LW treats Radiosity and Reflection together, simulates a default Specular Reflection without the ability to tune the bloody thing ... BAD, this is what the thread is about.

If you have a surface with Zero diffuse, and of course Zero Reflection, it wont receive any radiosity. If you have a surface with 400% Diffuse, it receives Radiosity x 4 and also any light facing this surface will bounce at a factor of 4.
Unnatural, but just in case you need it.

The necessity here is the ability to exclude the Radiosity Source from the reflection, and use an alternative one. More Control, more natural results.
:thumbsup: See the Worley's Art of Noise Document !
:goodluck:

Weepul
08-30-2005, 05:15 PM
With all your talk of diffuse vs. reflected, bear in mind that the way LightWave calculates radiosity is that it basically sends out probing rays in all directions from the surface. The incoming radiosity light is applied based on the diffuse properties of the surface. However, the light that's calculated as incoming will include anything that will render - reflections, highlights, transparency, shaders, as well as a surface's diffuse shading.

Try this: make a reflective ring on a flat plane (like is so often used to demonstrate caustics), put a luminous polygon in the scene to light it, turn on raytraced reflectons, and crank the radiosity quality way up. You'll see caustic reflections on the ground plane.


The necessity here is the ability to exclude the Radiosity Source from the reflection, and use an alternative one. More Control, more natural results.

Or, to use a normal light for the first diffuse hits, radiosity for bounces of that light, yet being able to have bright objects in the scene for sharp reflected non-Specularity-generated highlights, but not having the luminous object cast any radiosity into the scene which would manifest itself as ugly speckles (with lower radiosity quality) or "replacing" the light from the regular light (with ungodly high radiosity quality).

At least, that's how I'd use it. :D

wacom
08-30-2005, 06:55 PM
So you just want to be able to say that one object doesn't get it reflections caluclated in radiosity, and another does? You don't want any control over how much of each they get?

I get what you are saying, I just think it would be better in the end to do it at a surface level, and with overrides (where each object has its own shader, but for reflections and photons/radiosity you override groups of surfaces).

I do feel like the village idiot here though...so I'm probably way off...

So you want objects to be able to recieve reflections- and give/get radiosity, but have the ability to seperate the refelctions out so that they are not calculated in the radiosity solution...yes?

So what you'd like in an ideal world is the ability to tell an object NOT to transmit, but only to recieve photons? Like the floor in these images?

JML
08-30-2005, 09:21 PM
.........

as the images shows, radiosity+reflection = very bad render time.
(when diffuse is 100%)

thank you for the diffuse tip, it saves a lot of time on very transparent object such as glass.

but for everything else which is not transparent but reflective, radiosity improvement would help a lot..
---------

after doing the last render,
I guess it's not radiosity and reflection which make slow render time,
it's when there is transparency added with it..

Panikos
08-31-2005, 12:30 AM
JML, any value greater than zero in Transparency or Reflection forces renderer into more complex calculations, either 0.1% or 200%.

I believe that the more we discuss these, the more confusion.
Its natural that not all people carry the same experiences and not all people work in the same way. Its a waste of time trying for all to meet at the same "point"

I hope Newtek people understand what the feature request is about.
I wont post anything else here.

toma
08-31-2005, 02:06 AM
I think what we'd like to have (either per object or per surface if it can be overide at the object level) is something like this :

(taken from cinema 4D v7)

toma.

wacom
08-31-2005, 09:57 PM
I think what we'd like to have (either per object or per surface if it can be overide at the object level) is something like this :

(taken from cinema 4D v7)

toma.

That's basicly how it works in XSI too...so Panikos is right- this, like instances is a needed feature that most packages support.

Panikos
08-31-2005, 10:58 PM
Ok, my extra word !
Here is an example.
I have a character and his eyes have a reflection amount, something like 10%.
My character stands in a room with windows.
When he sees opposite to the windows, I want to see the reflections in his eyes. So far 10% works well.
If I turn him towards the windows, due to the very bright illumincation of the geometry that emmits radiosity outside the room, his eyes turn "burned white"

This is because Reflection is not "normalized" or is not "proportionaly scaled"
to the the average illumination of the scene. So even 1% looks burned, and such a small amount of reflectivity misses all indoors reflections.

(This weakness is visible in all reflective surfaces of the scene.)

If I have two geometries outdoors, one for radiosity and one for reflection,
one "Unseen by Radiosity" and one "Unseen by Rays" I can have separate amount of radiosity and separate amount of Reflection brightness for the entire scene, with only 2 mouse clicks I am done. No surface tweaking necessary.

;D

Weepul
09-01-2005, 06:25 PM
So you want objects to be able to recieve reflections- and give/get radiosity, but have the ability to seperate the refelctions out so that they are not calculated in the radiosity solution...yes?

Well, not for me. What I'd like is a way for an object to not appear in reflections, but still will cast radiosity light into the scene - and equally important if not more so, the opposite: for an object to be able to appear in reflections but not cast any radiosity light.

Panikos
09-01-2005, 07:14 PM
weepul !
BINGO :agree:

wacom
09-01-2005, 10:17 PM
:thumbsup:

Panikos, you are mistaken as to how much work it would be for this in a nodal system- it would be less than just selecting the objects in some instances!

You use a raytype spliter to control how the objects surface looks to other objects surfaces- not the other way around.

In this simple scene (on light, one luminouse polygon, a reflective sphere, and a checkerd floor) I think you can see exactly what you guys are looking for.

On is illuminating a scene, but not causing reflections, while the other is causing reflections (that I blew out on purpose) but not illumination, yet each scene is using photons.

wacom
09-01-2005, 11:13 PM
I'm hope'n that NewTek is going to put a node like this into 9.

Panikos
09-02-2005, 02:48 AM
Wacom, I cannot excite for something I havent seen yet. :argue:
Does it hurt that much to have an extra switch ?
:devil:

wacom
09-03-2005, 12:48 AM
Wacom, I cannot excite for something I havent seen yet. :argue:
Does it hurt that much to have an extra switch ?
:devil:

I don't see any reason why something like this raytype node couldn't be used as a default on object settings and with an on/off checkmark AND a slider to control how much it is ON when it is ON. Then for all the nodal phobic they could just tick something. I was only showing examples to show that it can be done... :D

Panikos
09-03-2005, 04:27 AM
Wacom :hat:

Currently FPrime's Radiosity doesnt cross Transparent Surfaces with 1 Bounce.
In order to have this, you need to have 2 bounces (see rendering penalty)

If you set a Transparent Surface like a window "Unseen by Radiosity", 1 Bounce works

Node-Based is cool, but is not a panakeea.

:D

Gui Lo
09-03-2005, 06:22 PM
Radiosity control at the channel level within the surface editer would be excellent.

Just wondering.
This sounds like the real world where the camera has a fixed aperture.
For example I watched the England vs Wales Football match last night. It was a sunny day and some of the pitch got direct sunlight while most was in shadow. When the camera looked at the sunlit pitch it was blown out and couldn't see the action properly until it adjusted.

Would a more controllable camera also solve the reflection problem?

Does anyone know if aperture control is part of the new camera shader?

Gui Lo

Panikos
09-03-2005, 06:54 PM
Gui Lo

You are shooting in a room with a window at daytime.
Usually the outdoors look white due to the difference in light intensities inside and outside. To be able to view both conditions you have to illuminate the dark area (the indoors) and then close the camera iris cause you cant beat the sunlight otherwise. (I speaking in real terms but its the same application in animation too)

The promised Camera Shader system deals with the fashion that the camera samples the light, i.e Lens shape, wide angle, telephoto, orthographic etc
It has nothing to do with the amount of light that enters the Camera.

You can study some HDR related material. Use google for that.
In LW terms, you can use HDR_Expose, Exposer, FP_Gamma.

Exposer is cool. It allows adjustment of the amount of Light that enters the camera and allows animation of its variables via envelopes plus Lagging.
For instance you are shooting the sun and then you pan to a shadow area.
You close the Iris when shooting the sun while you open the iris while you are panning towards the dark area to adjust to the new light condition.

This space is small to describe all these. But once you get familiar with the principles you will love it :thumbsup:

Anttij77
09-06-2005, 06:59 AM
The necessity here is the ability to exclude the Radiosity Source from the reflection, and use an alternative one. More Control, more natural results.
:thumbsup: See the Worley's Art of Noise Document !
:goodluck:

This would actually make it UN-natural.
Because, if you have a surface that emits light, and a surface that reflects light 100%, then the reflected amount should stay the same as the incoming light. That is physically correct, and natural.

toma
09-06-2005, 07:39 AM
This would actually make it UN-natural.
Because, if you have a surface that emits light, and a surface that reflects light 100%, then the reflected amount should stay the same as the incoming light. That is physically correct, and natural.

You are right but this should be possible with 3D… actually there are very few directors that like "that real lighting", all the images we are use to look take them a lot of effort to avoid that look. When shooting outdoor they have to use deflectors because otherwise the shadows will be too dark and so on…

What's the point of using 3D rendering with all the unatural things it can do, if the render engine is locked to "true real physical light"? All the photographers, film makers etc are cheating because no display, no paper, no camera can capture and render the light with a natural look (you don't need sunglasses to look straigh at the sun if it's a photo, even a super-over-exposed one…)

What we want is the possibility to tweek the lighting without having to use deflectors or multiple lights : we want to use the full capacity of our tool, 3d is another medium, it doesn't have to mimic the others.

One can think it is some sort of lazyness, but after all the tool exist, unfortunatly it is locked. I think we should be able to render all sort of light situation, real or not.

The way LightWave deals with light now should just be a "realistic light preset"…

I could do a analogy with material settings : why having a specular & gloss setting, in reality it is just reflections and microbump…

toma.

Panikos
09-06-2005, 07:42 AM
Scientifically speaking YES. "Natural" here means visually not scientifically.

The problems begin with some preconditions and assumptions.
Not always Backdrop is used as a Radiosity source.
And if the source of Radiosity is let say a geometry of 400% Lum intensity, the "0-100% range of reflection sensitivity is not adjusted/multiplied by 4/rescaled to accomodate all the HDR of the environment.
Reflectivity appears burned due to the additive mode that Reflections work in LW.

See what I wrote in the previous post about 2 different light conditions in terms of brightness.
Its impossible to tweak the Reflection Curve or its Gamma, there is no way.
There is No HDR_Adjust for the Reflection channel !
Reflection channel is associated with Color and Diffuse. If you weaken the Color and Diffuse in order to collect the weakest reflections of the environment you get very dark grey look.

Being able to exclude the Radiosity Source from the Reflection and using an alternative source, gives the ability to tweak the reflection with a smaller HDR range.

Anttij, I am sure that you are very familiar with HDR Range.
If I had the ability to Globally define the lowest and highest reflection range I would have been very happy. There is no such a thing, as simple as that :help:

Thats a cool idea for a Pixel Filter that would process the Reflection channel ...
But is more feasible and easier to ask for an "Unseen by Radiosity" button, than a HDR_Process for Reflections :D

toma
09-06-2005, 07:58 AM
:agree: That's why most of the time I have to bump the radiosity intensity up to 200%, use a "sun" for diffuse, another one for specular, and even sometime do another rendering just for the reflections…

Still, having the possibility to raise the radiosity intensity is the proof that splitting radio and reflections rays is partialy there but without control, just a master control.

It's better than nothing, but raising the radiosity to 200 300 or 500 % (for indoor scene) will lead to very grainy images. the only possibility now is to use an smoother environnement… that will ruins all the details in the reflections :( … so definitely Yes, we need more control…

toma.

Panikos
09-06-2005, 02:44 PM
Oky, here is a photo with a real reflection.
(Not only that, you get some decent SSS, Global Illumination with countless light bounces and a cool photo of Cyprus beach) :D

The reflection of the sunglasses is under shadow. If we interpret this surfacing it means that it has some Diffuse value in order "read" the shadow.
Also out of the overall ambience of the photo we can see a bright sky and bright sun too.
The Reflection of the sunglasses is too dim, but we can easily see the reflection of the beach as well as the sky on it.
If we had to duplicate such reflection in LW, it would have been impossible.
Why ? Because the sky reflection would appear FAR too bright compared to the reflection of the beach, cause the Sky here works as a Radiosity emmitter. Dropping the reflection level (let say 2%) to have such reflection of the sky, would make the beach reflection BLACK and invisible due to the difference in Illumination value.

See, the real reflection auto-adjusts to balance between the two opposing light conditions, where in LW this is impossible.

Panikos
09-06-2005, 02:58 PM
Here is a closer shot.
No photoshopping :D

toma
09-06-2005, 03:04 PM
Sorry if I mess up everything, but I think the glasses reflections can be done via the color highlight settings…

ohhh I feel bad ! sorry Panikos, I'm with you (really) but I think Newtek will say something like that…

I think that we need to emphasize the fact that UN-real lighting is needed even when you have to do a "photoreal" rendering…

tomas.

Panikos
09-06-2005, 03:08 PM
Toma, no need for sorries.
I dont need anyone's support if this what you mean.

ColorHilights will simply push the RGB Reflection towards the Surface color however it will maintain the reflection brightness.

In the example above the Reflecion is not tinted with a color.

If someone finds a cheap excuse to justify not to implement it,
forums are history, write down the date and my request.
:thumbsup:

toma
09-06-2005, 03:21 PM
;) works too without colors = here's an exemple with a simple sphere and a hdri environnement (beach probe)…

the sphere is set as this : 0% diffuse and 100% reflection with a color set to reb = 0, 0, 0… and as you can see in the second sample the reflection is "tinted" with black (the question remains, is black a color ?)…

toma.

[edit]. to Newtek : those simple settings, color Highlights and radiosity intensity should be more powerfull and versatile, and we should be able to "override" them at the scene level… untill then, the only way to go is G2, but even if I like it's power, I would rather have a way to do it in the Lightwave's way : straightfoward.

Panikos
09-06-2005, 03:59 PM
Toma, except from the sky, the rest looks black :)

Weepul
09-06-2005, 04:14 PM
Scientifically speaking YES. "Natural" here means visually not scientifically.

Quite right. Any scene that uses Phong highlights, Lambertian diffuse shading, anything but infinite bounce radiosity and reflection, etc. etc. etc. is physically inaccurate. It can be close, but it'll always be fakery. This control would be a great help in setting up scenes, photorealistically styled or not, as is necessary with today's computing power and tools.


Its impossible to tweak the Reflection Curve or its Gamma, there is no way.
There is No HDR_Adjust for the Reflection channel !
Reflection channel is associated with Color and Diffuse. If you weaken the Color and Diffuse in order to collect the weakest reflections of the environment you get very dark grey look.

Actually...I believe additive reflection is physically correct (assuming you've modified diffuse and transparency so no more than 100% of incoming light is reflected or transmitted), however, that's dealing with light intensities - linear gamma, it's sometimes called in graphics. LW renders with linear gamma.

Add the FP Gamma plugin, and you'll get nonlinear reflections with roughly correct intensity relative to the surrounding brightness, amongst other real-world effects on the image. You'll be able to see the beach and the sky in reflections. ;) Of course, your materials will look off, unless you've designed them to work with gamma adjustment from the start. (Or, use a custom shader, which I'd have already made if I knew how to program! :D)

It might be possible to approximate with a pixel filter, too...but that requires programming in C also...there can't be LScript pixel filters, to my knowledge.

Panikos
09-06-2005, 04:57 PM
If we select the Default backdrop as a source of reflection, it looks great on any reflective surfaces.
However LW (was first) to introduce HDR image illumination and Radiosity.
This left the Reflection options static and the necessity for more control is obvious.

Since the default backdrop is weak as a source of radiosity, if we boost it,
all the reflective surfaces are burned. :cursin:

As I wrote before, Reflection should have a Gamma control either per surface,
or a global Pixel Filter that would adjust the RAW RGB Reflection LW internal Buffer.
This is hard considering the priorities of Newtek currently.

Unseen by Radiosity switch and a layered Radiosity/Reflection is more possible.

Panikos
11-20-2005, 05:56 PM
aaah, everytime I find this in front of me, I throw stones :compbeati :bangwall: :cursin: :2guns:

Exception
11-20-2005, 06:30 PM
Guys I'm with you on this...

Not only the blowout, sometimes you want to use a different relflection map than the HDR maps you have lying around, and still illuminate the scene with an HDR. I run into this daily... There's no workaround...

heyhey I'm with it!

habañero
11-20-2005, 07:22 PM
I wote aye! I was hitting myself witht a smurf beer opener all of last week over this limitation. Still hurts! :P

Nice Rig Wacom. I thougth dragon32 was all the rage, but then you come here with a Z80 ...

Using luminosity for illumination, I have this trick that works fairly well with lamps:

http://iihhrrr.web.surftown.nu/test2.jpg

Now this Fprime render is lit solely with luminous objects, and not in a very long time. The main object which is actually a large sphere centered on the other light sources, and set to unseen by camera. If I could set this to unseen by reflections as well I'd have quite a few more options. This lightball showing up in reflections sortof breaks the illusion as it is now.

Just to clarify, using just the small lights there is close to impossible in LW classic and I did not manage to get any results in Fprime either

There obviously are other routes to that result, but notice this is a monte carlo full caustics render (don't mind the other one, it is from a try with just the small lights that I don't have at hand):

http://iihhrrr.web.surftown.nu/rings.gif

While LW caustics supposedly aren't consitent great for animation, and slow, these I find to be rather fast and they are consistant and independant of scale.

If you like this effect or not, I like to have options and being able to use luminous objects more for lightning especially with Fprime is an option I really would appreciate. Puleze Newtek, just this lil button! :)

Panikos
11-20-2005, 09:26 PM
Bright reflections coming from the radiosity emmitter, they are so ****ing bright.
I use Imageviewer FP to monitor these values.

Image Filters Plugins that are HDR sensitive, take these values as the highest and generate invalid results.
(Usually the Radiosity emitter is hidden from the camera)

The problem reached to the point I cannot stand it any longer.
Nobody from Newtek is willing to discuss it. Its dissapointing.
I am a member of other apps forums, and its a pleasure to find mutual solutions or in the worst case, painful compromises.
If I am considered a newbie and my idea is silly, there are some other people here asking for this.
If I dont see a solution in the upcoming release, I am sorry to say that I
am planning not to compromise any longer with such insufficiencies.

****
:thumbsdow :bangwall: :cursin: :twak: :2guns: :compbeati :mad: :screwy: :argue:

Ztreem
11-22-2005, 03:34 AM
I want this too.

Panikos
11-24-2005, 07:56 AM
Here how Pros handle light simulation.
http://www.debevec.org/Parthenon/Images/LightProbe.jpg
Modern Layered control !
3 Balls, White, Black, Mirror collecting Irradiance, Specular and Reflection.

In LW, Irradiance is sent to both Diffuse and Reflection.
Diffuse receives Radiosity, Reflection burns due to wrong info.

Reflection needs color hue, not irradiance strength.
:cry:

We need layered control, Unseen by Radiosity Option

habañero
11-24-2005, 09:44 AM
What is they are building in the background? Some sort of theme park? :P

This is an important request though, I have been had by this lots of times and I couldn't figure it. Respects to Panikos for bringing it up!

Panikos
11-24-2005, 05:51 PM
The building behind has been there for thousands of years.
Its the Parthenon.
http://www.debevec.org/Parthenon/

habañero
11-24-2005, 05:55 PM
Panikos I was just kidding, I have been there ...

Panikos
11-24-2005, 06:14 PM
:D ;) :hat:

Energy Conservation
http://graphics.cs.uni-sb.de/Courses/ws0203/cg/Literatur/Dutre_GlobalIlluminationCompendium.pdf

Panikos
12-13-2005, 04:23 AM
I spent long time trying to solve this puzzle.
Seems that the problem forces towards the use of ImageWorld/Texture Environment plugin. Not a big deal, however ...

Polygons with luminosity that act like Radiosity sources, burn the reflections one way, the other, in all ways :cursin:

The most serious problem is the lack of documentation. More like a button description than analysis. Some features simply cannot coexist with others.

Still an awful weakness

:2guns:

habañero
12-13-2005, 08:24 AM
I agree a lot with that, I think the manual could benefit a lot by putting more speak on concepts in there.

I understood everything in that paper, it has to do with rays ...

I am not sure if I understand your sentence #2 and #3 equally well, you are saying that luminous light sources are sortof broken, we need to fix this in post with several passes etc, or use the "full broken" imageworld/textured environments? Is it the global one or on surface per surface basis? And that the broken luminous objects are actually more correct than the unbroken ones, that so?

"Some features simply cannot coexist with others."

This sounds bad to me. Are you saying we can't have "the button"?

I would hate to give stuff away to Maxwell if it is at all possible not to do it.

I suspect the new team are already on this, but it'd still be nice to see it added to the 9 feature rooster in cryptical, vague terms ...

Panikos
12-13-2005, 09:37 AM
Habanero

Radiosity is generated by Environment and/or Geometry.
Environment emmits radiosity without aid, geometry emmits radiosity when light bounces on it.

Radiosity emmited from geometry burns the reflections.
This is my conclusion after countless period of time I spent experimenting.

Why this happens ?
Environment Plugins allow boosting intensity or Boosting backdrop.
This "Boosting" goes only to diffuse channel of the radiosity recipients and not to their reflection.
If you build your own radiosity source consisting from polygons, Luminosity influences both the Diffuse AND the reflection of the recipient surfaces.
I know it sounds complex but its a fact.

Finally, if you read the "Art of Noise" document on Worley Labs site, it states that HDR_images generate more noise.
He suggests to scale them down or blur them and drive this new image to the diffuse and the original HDR_image to the reflection.
This canalization is applicable only in empty ground-floor scenes and not in general.

I admit that Node-Based surfacing can provide it, but I prefer a global canalization than a per surface.
Global is one button, if you change your mind, it takes a few seconds to change it.
Per surface requires visiting all surfaces.

Not directly related but is serious.
Currently Backdrop/Spherical refletion/refraction maps are considered as extra diffuse and emmit undesired radiosity.
This doesnt happen in FPrime, only in LW.

Also, when using Raytraced Reflections + Backdrop its an accumulation of 50% Raytraced rays and 50% of Backdrop.
The backdrop is reflected de-facto even if is not accessible by the reflective surface, ie even if the reflective surface is enclosed in an other
geometry and doesnt see the backdrop.

Anyway, I can write a book about LW rendering engine, twice as fat as the current LW manual.
Its 2006 and we are using Reflection/Refraction algorithms of early 90s and I am literally sick of adding
a stack of plugins in order to solve a LW weakness.

jeremyhardin
12-13-2005, 09:51 AM
i still agree that this is a serious need.

habañero
12-13-2005, 09:52 AM
Well that clears up a few things.

I did read the thread from the beginning before posting.

Weepul
12-13-2005, 06:40 PM
I disagree with your reasoning, Panikos, but still 100% agree that this would be a very useful feature.

wacom
12-14-2005, 01:40 AM
I admit that Node-Based surfacing can provide it, but I prefer a global canalization than a per surface.
Global is one button, if you change your mind, it takes a few seconds to change it.
Per surface requires visiting all surfaces.



Ah, this is where an override comes in- you simply have the group and apply an override to a certain feature. Stop the nodal fear Panikos and embrace your future!

Panikos
12-14-2005, 02:06 AM
lol, If I was afraid of the future I wouldnt push the technology, I would satisfy with the current status. Nodal is great, I use it with sabre & TB_Shadertree.
Wacom, you have a lot of LW9 expectations. I do as well.
But, I dont have anything solid yet, just a press release.
I am very tired with this thread, too much official silence, give me a good reason to rely.

Exper
12-21-2005, 10:51 AM
Hey Panikos, still battling on this one? :D

Panikos
12-21-2005, 03:25 PM
Hehe Exper :D

Its not about battle at all.
Actually I am very excited for my conclusions.

Using the Low Dynamic Range version of the ImageWorld-image as a spherical reflection map brings peace ;D

My suspicions were 100% correct.
The Reflection channel needs a switch for Limit Dynamic Range as a native option, OR Splitting Radiosity from the rest of rays for use for the benefits of other cases.

I am so reliefed that I was proven right.
Now its Newtek job to provide a solution to this. I am anxious :hey:

Panikos
12-21-2005, 03:42 PM
Here is the proof.

The HDR image uses a HDR image in ImageWorld with boosted intensity.
The LDR image uses Raytraced reflections + Spherical map of the Low-Dynamic-Range version of the imageworld image.

I dont ask features by accident, nor for my ego.

(Sorry for the noise, its FPrime at Level 1)

Exper
12-22-2005, 04:58 AM
I dont ask features by accident, nor for my ego.I know for sure (lean consolation). Ok... :stop:... hold my tongue. :)

Panikos
01-02-2006, 08:12 PM
Here is clean version :D

The character is in rest-pose, to check illumination only :)
Pitty, Sasquatch is missing :oye:

Castius
01-02-2006, 10:59 PM
I wouldn't mind seeing more control over different types of rays inside LW. It should be somthing on surfaces and object controls.

This might help in the mean time when you're not using Fprime.
http://www.evasion3d.com/hs_lw_intro.html

Pavlov
01-03-2006, 03:58 AM
I agree with the feat-request.
i think ray control should be splitted as much as possible, giving:
- separate recursing control for every ray's type
- giving access to GI and ray control on surface basis
- more control over ON/OFF tags on shadowing, reflection, refraction on both OBJ/surface basis
- OBJ/surface based control of GI (% emitted/%received, not just ON/OFF)

Paolo Zambrini