FAO Dalomar - LWF & colour bleeding

[DzineTech_cg]

Damian, I hope you will be able to point me in the correct direction. I've been reading many topics on this site regarding LWF and separate ones on colour bleeding.

I have read that you have said VfSU is already setup for LWF; I assume this is with LDR and RGB correction enabled but what other settings should be set when trying to attain physically correct renders? I have read many topics on the Chaos Group site and LWF is indeed a confusing topic with the variety of methods suggested (Throb's 0.64 colour correction curve, Lele's .255 method, 0.4545 Inverse gamma method etc...) I would prefer to go down the .255 method however I am unsure if I need to change any settings to use this workflow.

• Should I have Correct LDR enabled?
• Should I have Correct RGB enabled?
• What gamma settings would normally be applied?
• Does the sRGB button come into play with the method?

I have been testing with a variety of settings however nomatter what I do, in a current interior model that I am currently working on (white walls, timber floor etc..) I am unable to stop or noticably reduce the colour bleeding from the floor onto the walls without performing a precalc using a white material on the floor ( I have a feeling that doing this would provide an incorrect physical solution with more light being bounced due to the white surface). One of the features of LWF is to reduce colour bleeding but I haven't been able to achieve this yet, even applying the .255 texture multiplier to the floor material. I expected with the darker material that less colour would bleed onto the wall but it wasn't the case. Any suggestions?

Thanks

PS Using IM+LC, linear colour mapping set to 1-1, system gamma both at 2.2

[dalomar_cg]

Re: FAO Dalomar - LWF & colour bleeding

All in all, this is a very complicated question, and one that is a personal decision of how you want to work. V-Ray is setup for linear work flow with the correct RGB and LDR options. Lele's method is very good, and something I think is very promising, however its much easier to implement and keep track of within Max (via maxscript).

All that being said, I'll run down how I would go about Lele's method within vfsu (in principle his method is actually fairly simple, although his explanation and logic for the method goes through a number of complex steps). I must also note, that I haven't gone through and fully tested this workflow, that is been a little while since I've dealt with this, and that I am not currently using this workflow (all be it that I'm not doing that much vis work at the moment).


At the heart of Lele's method is the brightness of the sun/sky, how materials react to that brightness, and how the physical camera exposes the materials and the sky. He argues that the sun/sky system is the basis for photometric lighting in v-ray (in max this may be argued, but in vfsu this is most certainly the case), and that one of the implications of using that system is that the physical camera must be enabled as well in order to expose this brightness.

What Lele gathers through a series of very logical, although not incredibly apparent, steps, is that the relationship of the sun/sky to the physical camera is acting correctly, but that vray materials are "miscalibrated" in a way that they do not work correctly within that system. Therefore, when we adjust the exposure of our cameras, we typically do so with the understanding that we must properly exposes our materials (which are "wrong") as opposed to the sky (which is right).

This leads to two major issues; the darker appearance of the sky, and a physically incorrect reaction of the materials. The physically incorrect reaction of the materials can be seen in the caustics effect of the gargoyle (if you remember the video), scenes that seam to bounce around two much light, and the "gathering" of lighting in certain areas where it shouldn't be. Essentially with material settings as they are the materials wind up "reflecting" too much of the sun's intense light energy. This is because if we keep a very simple model of the shading/reflectance energy, then the materials do not decrease this energy, but rather, maintain that energy at levels that will overexpose standard images (this is the simple equations that lele uses as example at the start of the video).

So this leads us to a situation where we need to adjust the material model that is currently in v-ray. Lele's solution for doing this is to essentially multiply the materials by .255 which essentially darkens all of the materials. To compensate for these "darker" materials, the exposure of the physical camera must be changed to match that (ie we have to increase the exposure some how...lele suggests iso which is a good way). Once we do this, the dark sky that we were seeing before is now at its proper/expected illumination.

AFAIK lele does not suggest this as an alternative to LWF. In fact he retains the gamma correction of all images and colors (inversely corrected) as well as gamma correcting the final image. So in vfsu that wouldn't change (this leaves the sRGB button unnecessary).

So know that all the cards are on the table, how do we make this work in vfsu.
1. All colors must be set through the AColor map...you would set the color as you normally would, and then change the texture multiplier to .255

2. All maps must have a texture multiplier of .255... this causes a bit of an issue with linked materials. Therefore, even though it is still advised to use linked materials to generate proper mapping coordinates, you will have to add another layer ON TOP OF the linked layer with the same texture and a multiplier of .255. That will effectively "remove" the texture from the SU material.

Here are somethings that I am not sure of how they will react in vfsu.
1. Reflections - specifically filter colors. I don't think that setting the reflection value (ie fresnel in most cases) will need to changed. I also don't think that the .255 thing should extend to setting filter colors, since that would effectively decrease the reflections by doing that (this I'm a little more unsure of).

2. Light emitting materials - These are hard to pin down when using the physical camera any way, so I would stick with the mantra of "just set it at whatever works for the scene" (I kinda just made that up, but you get it). My advice would be to know what the exposure of the scene is before you go messing with levels and things. Since there will be more exposure you should probably notice less of a need for very high multipliers (but they will still probably be significant).

3. Refraction/Fog - I really don't have much of an idea on this, so I'm not even going to postulate.


Keep in mind that this system requires complete adherence. It either all or nothing because that one material that doesn't adhere has the potential to really through things off. There also isn't an automated way to take care of all of this...its basically you going through and checking out every material, so be aware when you make them and when you modify them.

Within this system, I feel that the camera is key. Be very mindful of setting exposures correctly. Also, one of the most underrated features of the physcam IMHO is white balance. You'd be surprised what you can do/adjust with using that feature (its a very good way to adjust any color bleeding that may occur).

Your main issue was color bleeding, and I think this is something that should help that, since the preservation of that light energy will no longer be as significant as it was with "standard" materials. Your current method for decreasing color bleeding isn't bad (I've used it myself a number of times), but also remember that you don't have to go all the way to pure white to decrease the effect. Sometimes just a less saturated version of the color/material is enough to make it work. Interestingly enough I've seen some very significant color bleeding in real life that I would have deemed "undesireable" if I was doing a render of that situation...just an observation.

Hope this helps...I'm sure you'll have questions and I hope you can post some results so that we can see how it works out. I think this method shows promise, its just that its implementation is kind of hard to swallow (in a number of senses).

[DzineTech_cg]

Re: FAO Dalomar - LWF & colour bleeding

Thanks for the lengthy reply. The only area that I didn't quite follow was

"you will have to add another layer ON TOP OF the linked layer with the same texture and a multiplier of .255. That will effectively "remove" the texture from the SU material."

I'm not 100% sure what you mean by this - do you mean duplicating the diffuse layer?

As I dont have a background in photography my knowledge of correct physical camera settings are based purely on the settings on ASGvis tutorials. In http://asgvis.com/index.php?option=c...sg9873#msg9873There you describe a method of setting up a scene if attempting to use the .255 method. From my what I recall of Lele's 3 video's, the .255 number (which he admits is only a guidline number) is the factor whereby the pure white test scene does not have any float values higher than 1.0 else clamping would occur (I think!)

Would I be correct in thinking that what we are trying to do is found out what multiplier value we need apply to our texture maps and Acolour maps following a process of trial and error until no clamping occurs for whatever configuration of camera settings I have chosen. If this is the case, for an interior shot, would you set up your desired view and do the above trial and error process? From my early attempts of using a vray plane, applying a vraymtl of Acolour set to white with .255 multiplier, I was still getting clamping until the multiplier was at about .15ish.

[dalomar_cg]

Re: FAO Dalomar - LWF & colour bleeding

I'm not 100% sure what you mean by this - do you mean duplicating the diffuse layer?

Not a duplication in the sense of the same thing, but a recreation of the layer. When you create the ]
Would I be correct in thinking that what we are trying to do is found out what multiplier value we need apply to our texture maps and Acolour maps following a process of trial and error until no clamping occurs for whatever configuration of camera settings I have chosen. If this is the case, for an interior shot, would you set up your desired view and do the above trial and error process? From my early attempts of using a vray plane, applying a vraymtl of Acolour set to white with .255 multiplier, I was still getting clamping until the multiplier was at about .15ish.
[/quote]

Essentially, Yes. Lele's value of .255 is a general guideline, and the important thing about what he was suggesting was adjusting the materials instead of the lighting/sky. I think developing this through a process of trial and error is the way to go, but on the scale of a whole scene, not an individual view. I would imagine that after a number of scenes you would find a range of values that would work well for how you set things up and the nature of those scenes.

From what I remember, my tests produced a multiplier that was in the range of .15. That may change between an interior scene and an exterior one, but I imagine that as long as you find a multiplier that works, then you'll be able to rely on the output.

[kwistenbiebel1]

Re: FAO Dalomar - LWF & colour bleeding

I have been reading this thread with great interest as I also had questions about linear workflow and a correct balance between sky and texture appearance.

With what ratio should we increase the ISO value of the physical camera when using Leles method?

Dalomar suggested making a second diffuse layer set to 0.255 for the 'linked' materials.

Unfortunately, for bigger elaborate scenes containing dozens of materials (happens to me all the time) it is a real pain that VfSU doesn't 'autolink' the current SU materials so we can already start with a list of linked mats, instead of going through the tedious linking process per material (By the way linking a mat takes too much computation time imho).

There was an update promised looong time ago which would adress this 'workflow' issue of the linking of materials.

Could someone of Asgvis team please give us a hint on when we can expect it (if ever..)?
I want to order VfSU but I was waiting for the update to come....

And a second thing,
VfSU is priced quite high. Is there a chance the price will get a little lower as I am confident that this would make the customer market way larger?...

For the moment I can't use VfSU anymore as my demo period expired so I'd love to get on the boat again for next VfSU update....


Cheers,
Biebel

[thomthom]

Re: FAO Dalomar - LWF & colour bleeding

DzineTech: can you post some links to the articles you refered to in the original post?

[rv1974]

Re: FAO Dalomar - LWF & colour bleeding

There was an update promised looong time ago which would adress this 'workflow' issue of the linking of materials.

Could someone of Asgvis team please give us a hint on when we can expect it (if ever..)?
I want to order VfSU but I was waiting for the update to come....

And a second thing,
VfSU is priced quite high. Is there a chance the price will get a little lower as I am confident that this would make the customer market way larger?...

For the moment I can't use VfSU anymore as my demo period expired so I'd love to get on the boat again for next VfSU update....

It was promised in the end of summer but the exact year wasn't noticed.
Let's hope it'll happen in 2008 :-* ASGvis

[dalomar_cg]

Re: FAO Dalomar - LWF & colour bleeding

I have been reading this thread with great interest as I also had questions about linear workflow and a correct balance between sky and texture appearance.

With what ratio should we increase the ISO value of the physical camera when using Leles method?

I believe this is how he calculated it...don't quote me though

currAvgSceneValue    desiredAvgSceneVal
-----------------  =  ------------------
currISO         resultingISO

[quote]
Dalomar suggested making a second diffuse layer set to 0.255 for the ']

In our internal builds, the materials are handled in a much more integrated way that is also much easier to manage which makes linked materials much less needed as far as workflow goes.

Could someone of Asgvis team please give us a hint on when we can expect it (if ever..)?
I want to order VfSU but I was waiting for the update to come....

I wish I could...sorry, but thats all I can really say...it will be out though.

And a second thing,
VfSU is priced quite high. Is there a chance the price will get a little lower as I am confident that this would make the customer market way larger?...

I'm not here to talk about the price, but anything is going to seam like a lot when you can download the base application for free.

For the moment I can't use VfSU anymore as my demo period expired so I'd love to get on the boat again for next VfSU update....


Cheers,
Biebel

send me an email and I'll see what I can do.

[DzineTech_cg]

Re: FAO Dalomar - LWF & colour bleeding

I've been playing about with this however I thought it would be wise for me to follow the tuturials in SketchUp as Lele uses in Max; so firstly I wanted to test that the float value of the SU sun matches 385 as Lele suggests the vray sun in max does.

I created a plane, applied a vraymtl with a pure white diffues colour to it, set the SU sun settings to 12:00pm (date july 1st), colour mapping linear (dark and bright multipliers set to 1), input gamma 1.0 - output gamma 2.2 . Using a standard camera or a physical camera (with exposure off) the float values I expected would be circa 385. The maximum float values I get are 125,130,130 (decimal places removed!) So, why am getting about 1/3 sun intensity as Lele's vray for max test does? The only thing I can think os that the sun position isn't perpendicular to the plane but I have tested various settings and the float values wouldn't go any higher.

Any idea's Dalomar? I cant think where I have gone wrong so is there something different in the VfSU build compared to VfMax?

[dalomar_cg]

Re: FAO Dalomar - LWF & colour bleeding

So, why am getting about 1/3 sun intensity as Lele's vray for max test does? The only thing I can think os that the sun position isn't perpendicular to the plane but I have tested various settings and the float values wouldn't go any higher.

I would say thats it right there. At the beginning of the video Lele says that the only reason he is able to make that jump is because the sun is perpendicular. Once there's an angle (or the plane is not white), then shading algorithms get involved and it will decrease that intensity.

The sun model itself will also change its intensity based on the angle of the sun coming in, so the intensity of the sun actually decreased as well as the contributing portion from the sky (you didn't say you had it on, but you shouldn't at this stage)

Also, I think lele started with completely uncorrected gamma scene 1.0 input and 1.0 output...not sure if that would really matter though

[thomthom]

Re: FAO Dalomar - LWF & colour bleeding

And wouldn't the geographic location of the model also impact this?

[DzineTech_cg]

Re: FAO Dalomar - LWF & colour bleeding

I would say thats it right there. At the beginning of the video Lele says that the only reason he is able to make that jump is because the sun is perpendicular. Once there's an angle (or the plane is not white), then shading algorithms get involved and it will decrease that intensity.

I was pretty sure that would be the case but thanks for confirming it.

The sun model itself will also change its intensity based on the angle of the sun coming in, so the intensity of the sun actually decreased as well as the contributing portion from the sky (you didn't say you had it on, but you shouldn't at this stage)

You are correct, I set the environment to none and gave it a colour from the swatch.

Also, I think lele started with completely uncorrected gamma scene 1.0 input and 1.0 output...not sure if that would really matter though

He states 1.0 for input only, 2.2 for output and system gamma. As we don't have a system gamma setting I dont know if that affects us trying to replicate the method.

[dalomar_cg]

Re: FAO Dalomar - LWF & colour bleeding

He states 1.0 for input only, 2.2 for output and system gamma. As we don't have a system gamma setting I dont know if that affects us trying to replicate the method.

I wasn't exactly he said, but it probably won't matter with bright float values... I'm not sure what you mean as far as not having a "system gamma setting"

[DzineTech_cg]

Re: FAO Dalomar - LWF & colour bleeding

Oops, I posted the reverse input and output settings but I did have the input set to 2.2

My description wasn't exactly helpful but by system setting I mean the display gamma setting. The only reason I mention it is that I am trying to apply the rules of LWF from max to SU.