Chaos Corona Forum

Chaos Corona for 3ds Max => [Max] Feature Requests => [Max] Resolved Feature Requests => Topic started by: pokoy on 2014-07-14, 20:01:32

Title: Light intensity behavior not plausible?
Post by: pokoy on 2014-07-14, 20:01:32
Finally had some time to have a look at A7 and love the new Nitrous support - great!

However, there are some issues with the way lights behave, it seems they're not physically plausible (using that term here because it really makes sense when working with lights):

- Light illumination is increasing when you increase light's size, I think this is wrong. Instead, the illumination should decrease when you increase its size because the same power is emitted from a larger area, meaning that less light will illuminate the same area. The current behavior is the exact opposite of real world behavior. Sorry if illumination is not the correct term, I hope you understand what I mean.

- When switching between W/lm/cd/lx the intensity should be converted to display the same power. Right now, when I create a light with 50W and switch to show Candela instead, it'll use a value of 50 cd when rendering, resulting in a way darker render. I'd expect the light to keep the same visual intensity and the intensity value to be converted accordingly.

- Viewport display won't respect switching units. Same example as above: create a light with 50W, switch to lm. Render will be darker (correct when it's going from 50W to 50lm), viewport display will keep the intensity of 50W. If the value was converted to display the same intensity in another unit, viewport AND render result would both be correct, right now they are different.

Hope this makes sense.
Title: Re: Light intensity behavior not plausible?
Post by: maru on 2014-07-14, 20:16:36

the illumination should decrease when you increase its size because the same power is emitted from a larger area, meaning that less light will illuminate the same area. The current behavior is the exact opposite of real world behavior. Sorry if illumination is not the correct term, I hope you understand what I mean.
What? :D

Quote
- When switching between W/lm/cd/lx the intensity should be converted to display the same power. Right now, when I create a light with 50W and switch to show Candela instead, it'll use a value of 50 cd when rendering, resulting in a way darker render. I'd expect the light to keep the same visual intensity and the intensity value to be converted accordingly.
I'd like that, too.
Title: Re: Light intensity behavior not plausible?
Post by: romullus on 2014-07-14, 20:24:21
- Light illumination is increasing when you increase light's size, I think this is wrong. Instead, the illumination should decrease when you increase its size because the same power is emitted from a larger area, meaning that less light will illuminate the same area. The current behavior is the exact opposite of real world behavior. Sorry if illumination is not the correct term, I hope you understand what I mean.

No, it's not wrong: radiance (http://en.wikipedia.org/wiki/Radiance)
Title: Re: Light intensity behavior not plausible?
Post by: pokoy on 2014-07-14, 20:45:53
Not really wrong, more a case of what you define as intensity, but I see what you refer to.
I find it more correct if the total emission was fixed regardless of the light's size. In that case my suggestion would make sense.
I guess I have to do some tests tomorrow. I wonder how emission in material works.

What about the different units having no effect in the viewports, do you agree that it's not working as expected?

Title: Re: Light intensity behavior not plausible?
Post by: pokoy on 2014-07-15, 10:42:33
Ok, after some tests I realized that it's W/m^2, so it makes sense that the illumination increases when the lights size increases as well. Same for lm, cd and lx if they're meant to display illumination of a surface. Sorry, I've been expecting to see the same behavior like in how Brazil handled lights.

Still, there's the inconsistency with changing units where the render shows the correct intensity (for example 50W to 50lm) and the viewport does not. Best would be to convert the value to result in the same intensity so it works correctly for both the render and the viewport. Should I post this to the bug section?
Title: Re: Light intensity behavior not plausible?
Post by: Ludvik Koutny on 2014-07-15, 11:35:08
I would not expect most things to work like in Brazil. Brazil went in completely opposite direction than Corona. More fakery, and less physically based.
Title: Re: Light intensity behavior not plausible?
Post by: pokoy on 2014-07-15, 12:00:07
Actually, it was more about flexibility, there was a checkbox to allow for both scenarios. No need to dismiss an idea when it may be useful, even in Corona.
Title: Re: Light intensity behavior not plausible?
Post by: Ondra on 2014-07-15, 12:23:09
some of the other emission units should give you the behavior you want
Title: Re: Light intensity behavior not plausible?
Post by: Ludvik Koutny on 2014-07-15, 12:36:17
Actually, it was more about flexibility, there was a checkbox to allow for both scenarios. No need to dismiss an idea when it may be useful, even in Corona.

Of course i wasn't implying that. My point was, that if you try to apply Brazil workflows in Corona, you may fail. And as Keymaster said, there are multiple emission unit models to choose from :)
Title: Re: Light intensity behavior not plausible?
Post by: pokoy on 2014-07-15, 12:39:24
some of the other emission units should give you the behavior you want

Yeah, that's what I thought. They do not show the correct behavior in the viewport, though. Is it possible to implement correct viewport behavior for all light units?
Title: Re: Light intensity behavior not plausible?
Post by: Ludvik Koutny on 2014-07-15, 12:44:49
some of the other emission units should give you the behavior you want

Yeah, that's what I thought. They do not show the correct behavior in the viewport, though. Is it possible to implement correct viewport behavior for all light units?

Oh, sorry, my mistake then. This is for Keymaster to answer. AFAIK, viewport lighting is difficult task, as Autodesk sucks at providing correct documentation or support for developers when it comes to Nitrous. But we will have to wait for Keymaster to confirm or deny :)
Title: Re: Light intensity behavior not plausible?
Post by: Ondra on 2014-07-15, 13:54:58
doing the correct intensity in viewport is sometimes pain in the ass, sometimes impossible... I'll look into it though
Title: Re: Light intensity behavior not plausible?
Post by: pokoy on 2014-07-15, 14:23:18
Thanks for considering this!
Title: Re: Light intensity behavior not plausible?
Post by: romullus on 2014-07-15, 14:44:22
In my opinion, viewport lighting is just a waste of time, not worth even considering. Interactive rendering,  that's the real deal. Can't wait for that implemented in Corona.
Title: Re: Light intensity behavior not plausible?
Post by: pokoy on 2014-07-15, 14:54:21
I guess it depends on the kind of work you have on your desk. When working with a client in front of my PC, I want a basic lighting just to be able to tell where the light comes from, where the shadows are etc. It would be a missed opportunity not to use Nitrous with its capabilities for that, and I constantly find myself thinking how much time it could save me when all the stuff was working in the viewport. Not having to render something just to tell how bright my lights are and where the shadow will be - nothing wrong about that.

When it comes to finalizing the image/shot and working on my own, then IR is the more important thing, I agree.

Both can be a huge time saver, even more so when combined ;)

Edit:

Some thoughts why I think it would be important, rather sooner than later:

- already mentioned above, I find myself loosing a lot of time rendering stuff only to set up light intensities and to quickly preview shadows
- it's very hard to implement a smooth interactive workflow due to SDK shortcomings, max has just poor support for any interactive solution. Based on how much demand there is for this workflow, we may see improvements from Autodesk here. Even if they'll work on it, it can easily take another 2 years until max provides a good foundation for interactive engines. I wouldn't want to have to wait for so long with only an incomplete support for all the nice viewport features that we got with Nitrous.