Linear Workflow

[Micha_cg]

Linear Workflow

.... One workaround for this is to use textures for everything. Even for simple colors use a small colored patch and apply gamma correction in the bitmap loader. ...

I use this workaround for my current project. Thank you for the hint, Gijs.

[Micha_cg]

Linear Workflow

I ask me, make it sense to use the gamma 0.4545 for bump and reflections maps too?

[flipside]

Linear Workflow

I don't think so, since it's just the values that count there, it will look different however

[Micha_cg]

Linear Workflow

I see, the answers is difficult. Maybe Vlado know it.

Other question: if I set my textures at 1/2.2=0.45, than my images show to much saturation at textures (output 2.2). If I set the output at 2.6, it looks better. I have the feeling, the gamma correction dosn't work perfect. What is your experience? Gijs recommend gamma 2.6, but in my other render engines I never have used it.

[Gijs]

Linear Workflow

@Micha: I do not recommend gamma 2.6, it is just that an uncalibrated monitor has a gamma near to 2.6

if you are serious about color (and i assume you are) Then you should do yourself a favor and buy a hardware calibratin device to measure and profile your monitor on a certain whitepoint and gamma.

One other thing that pops into my mind is that I remember you saying that you should set secondary multiplier of GI on 0.8 because of the max. 80% reflectance rule. If you want to use the multiplier for this purpose that it would make more sense to me to leave the secondary multiplier at 1.0 and decrease the primary to 0.8. The reason is that this will reduce the energy of the first bounce only, and not the subsequent bounces as in your method. I am confident that this will also help you to correct your saturation problem.

[Gijs]

Linear Workflow

As for changing the gamma of your bumpmap: I think it makes sense to change the gamma for a gray scale image, because it influences the midtones. you can question however if you're going to see the difference...

[Micha_cg]

Linear Workflow

@Micha: I do not recommend gamma 2.6, it is just that an uncalibrated monitor has a gamma near to 2.6

if you are serious about color (and i assume you are) Then you should do yourself a favor and buy a hardware calibratin device to measure and profile your monitor on a certain whitepoint and gamma.

It's on my personal wishlist since a few weeks and I will buy it soon.
I would buy a ColorVision Spyder. Can you recommend a device?

One other thing that pops into my mind is that I remember you saying that you should set secondary multiplier of GI on 0.8 because of the max. 80% reflectance rule. If you want to use the multiplier for this purpose that it would make more sense to me to leave the secondary multiplier at 1.0 and decrease the primary to 0.8. The reason is that this will reduce the energy of the first bounce only, and not the subsequent bounces as in your method. I am confident that this will also help you to correct your saturation problem.

Some posts befor you can see an example with primary 0.8 - the LC pass dosn't match with the final rendering. The LC pass is darker. But I like to use the LC pass for fast preview and adjusting the lighting of a scene. The truth is, we are fumbling with this partial solutions.
What we need is a general color and texture multiplier 0.8 for physical correct workflow!!! For example Maxwell user know the physical rule (it is like mandra for beginners), but can not set partial multiplier (this is good so) and must set all colors and textures manual at 80% (except bump and reflection maps). I think, we could make is easier.
We can not set colors/textures at 80% for GI only. This is not correct. We would get a wrong ratio between GI light and direct light and object color/ object reflection. If a white color should be a 80% grey, than it must be a 80% grey for anything. So, the way of the maxwell user is physical correct. OK, I could set my textures at 0.8 and could correct any colors per Acolor multiplier 0.8. But a general multiplier could make the workflow easier. Also, the manual correction dosn't work for Rhino materials.

I would like, if we could make, that Vray can be used from everybody in a physical correct way. I would like, if we could set global options:
* gamma output - 2.2
* texture and color gamma - 0.45
* color and texture correction - 0.8
So, every new user could use the standard colors and textures and anything will be good. He must not think about gamma options and 80%.

Gijs, I would be happy, if we could discuss this topic until this 80% problem is solved. It's so important.

[Gijs]

Linear Workflow

OK, now in order to get things together we must at least agree about what is physically correct and what is not.

1 Do you agree that for physical correctness, both primary and secondary multipliers should be left 1? (and rather correct things at material level)
2 What you are saying about the 80% for colors and textures is related to diffuse color only right? (as opposed to using it for everything including fog color, reflection color, filter colors, etc)

if both 1 and 2 are a yes, then I suggest that it is better to make a prominent note in the help files or a tooltip that guides you. Anyhow we should make a recipy for making physically accurate renderings and still leave it up to the users if they want to make use of it and where they want to step away from these rules to achieve a special effect.
As for correcting textures, how could you tell me that a certain texture doesn't already has the right HSV values? I think it is 'dangerous' to correct textures automatically by multiplying them with 0.8.

As for the spyder: I don't have experience with those, I have used i1color and they work good. You should however google around for the best choice, because I once read a test that had different recommendations depending on the brand and type of monitor.

[Micha_cg]

Linear Workflow

1 Do you agree that for physical correctness, both primary and secondary multipliers should be left 1? (and rather correct things at material level)

Yes, this would be good.

2 What you are saying about the 80% for colors and textures is related to diffuse color only right? (as opposed to using it for everything including fog color, reflection color, filter colors, etc)

Right.

As for correcting textures, how could you tell me that a certain texture doesn't already has the right HSV values? I think it is 'dangerous' to correct textures automatically by multiplying them with 0.8.

I never have seen physical correct textures. Most commercial texture bundles are simple textures with white = 255. Everywhere I see color values around 255. I suppose so, the most user use textures like this. The global diffuse color multiplier is most important for Rhino object based materials.

Do you agree, a diffuse color multiplier make sense? The user could choose to let it at default 1 or at 0.8.

[Gijs]

Linear Workflow

Maybe there could be added an automatic Rhino to Vray material converter to solve the Rhino material issue. Or let mcneel add a multiplier button

Because there is already the option in the bitmap loader to change the multiplier. I still don't think that a global texture multiplier makes sense. In fact when you use a texure, you should actually be able to look up the reflectance value of that particular material and base the multiplier on that. So suppose you have a wood texture with an average reflectance of 50% and the average of the bitmap is like 70%, then the multiplier would be 0.71. I don't think that all textures have the same 'error' as they are all photographed or scanned often with autocalibration.

btw, just a few post back I mentioned this:

And also it (maxwell) will have a texture correction internally without doubt.

here is the proof (look for Tom's answer)
http://www.maxwellrender.com/forum/v...ic.php?t=17275

[Micha_cg]

Linear Workflow

I have the feeling, that textures are allways captured to bright, but very seldom to dark. So, a global correction of 0.8 could be the right direction. We don't know, how much correction we need, this is true, but 0.8 could be a good approximation. The error between 0.71 and 0.8 is not so high as between 1 and 0.8. I have the feeling, don't touch the value and let it at 1 allways produce a higher error than a default usage of 0.8.

An other example: color system like NCS or RAL are working in non physical correct mode too. For example RAL 9016 Verkehrsweiss give me a intensity of 96%.

If we can use options like global GI multiplier, than we should get an option like a global general diffuse color multiplier too. I don't understand, why the one option makes more sense, but not the other.

I think, the problem is, that classical raytracer used 100% white colors and it was right so. If somebody has used the 100% color system, than classical texures are right too. But we have a new situation, the new engines like Maxwell and Vray need to be physical correct below 80%. I think, a global correction could help to go in this direction and could help to avoid to much work on texture/color settings and would help we can better use the Rhino materials.

So, the only disadvantage seems to be, that a coorection of 0.8 would not be the right factor for any case. But most we don't know, which correction is the right, - we only know, we need a correction.

PS: I'm curious, how maxwell will handle this problem.

[Micha_cg]

Linear Workflow

Gijs, do have seen this here. It seems to be Maxwell do an internal correction, but for my taste to much.

[Gijs]

Linear Workflow

Yeah, I've seen that one. I think your assumption is right (about the internal correction). The main difference between Maxwell and Vray (apart from the rendering speed of course) is that in Maxwell there are certainly made a lot of decissions behind the scene. That is what makes Maxwell produce quite high quality right out of the box.
I think that they have been brainstorming about what can go wrong if someone just starts using the software. One of them has probably been the 'too high reflectance problem'. My hard guess is that colors are not clipped, but that they put a curve on the colors that is let's say linear up to 70% reflectance and then goes asymptotically to something like 85% reflectance.

What I would probably find better is if there was a checkbox in the diffuse color / diffuse texture that is on by default (but can still be overridden) that says auto convert colors to physically correct values. It reminds me of a checkbox in the new vray blend material (in 3dsmax) that has a shellac button with a warning that it can create invalid results.

[Micha_cg]

Linear Workflow

Yeah, I've seen that one. I think your assumption is right (about the internal correction). The main difference between Maxwell and Vray (apart from the rendering speed of course) is that in Maxwell there are certainly made a lot of decissions behind the scene. That is what makes Maxwell produce quite high quality right out of the box.
I think that they have been brainstorming about what can go wrong if someone just starts using the software. One of them has probably been the 'too high reflectance problem'....

We could try to help, that Vray can be easy used as possible too.
I'm curious what Joe and Corey say.

[joe_b]

Linear Workflow

Gjis was involved in a lot of UI / overall V-Ray for Rhino usage discussion with Corey and Damian. They've brought me up to speed on some of it. We're completely open to his ideas and have already had several meetings about it and other topics (don't worry Micha, Linear Workflow was a big topic )