VRay Sun, Sky and Physical Camera video tut + Bonus Script!

[DaForce]

Bloody jumped up italian

[studioDIM]

Aye, by definition!

Lele

[DaForce]

too right

[priad]

Thanks a lot Lele!
Great video!

[wxyz]

AE Settings?

Fantastic tutorial. Does anyone know how to tweak frames in AE7 to match the V-Ray Frame Buffer? Saving Cineon files worked, but zi cannot seem to match up hdr's.

[mmikee]

Thanks again ! This is a great script.

[jonnybefree]

great script lele!! thanks alot.

[zeronove]

well done, very useful videos, thanx a lot lele, but ..only a few questions...

please correct me if I'm wrong, I understand that using vrcolor map is someway similar to use cc plugin to darken color swatches (or not?), so
1. in which way it is different from the lut correction that is available in latest max releases (I'm still on viz 2005, so I can't test it on my own)
2. the ISO values resulting from your test seems to me a little unconfortable if compared with the common ones, I mean 100 ISO is a sort of minimum in "real life", isn'it!?

thanx


a.

[percydaman]

loved your videos Lele. thanks a bunch

[studioDIM]

please correct me if I'm wrong, I understand that using vrcolor map is someway similar to use cc plugin to darken color swatches (or not?)

I personally loathe the CC plugin, and never use it, but multiplying 8bit rgb integers by a float value leads to a float value, so in a way it should be identical.
If, that is, the CC plugin was coded correctly, and i am definitely not sure it is (hence overbright and invalid pixel values are returned through vray).
Also, using the VRay colour map leads to precise end colour values, which isn't the case for CCing them.

1. in which way it is different from the lut correction that is available in latest max releases (I'm still on viz 2005, so I can't test it on my own)

LUTs are for mapping of 10bit cineons to 8bit images.
Here we work in 32bit float (~4 million values for 0.0-1.0 versus 1024), hence with a much higher precision.
My guess is that the LUT option is there for integration with flame and inferno systems (which lack support for 32bit images) used in high-end film production.
But don't quote me on this

2. the ISO values resulting from your test seems to me a little unconfortable if compared with the common ones, I mean 100 ISO is a sort of minimum in "real life", isn'it!?

Indeed.
I touch only ISOs for the sake of simplicity: that is a linear way to change exposure (i could have done it with exposure times, linearly as well, but already that is the reciprocal of 1. It may be more difficult for some with less of a mathematical mind).
Of course, feel free to lower F-numbers, and keep the ISOs in a "confortable" range.

I'd like also to set a few points about this physicality fashion we're all after.
First and foremost i hate Dslr.
I tried them all at work (stopmotion for kids for channell4 in the uk) and not a single one of them (from top of the range kodak and Canon to cheaper models) was able to shoot the same picture twice with the same results.
The digital light measurement is so very flaky it hurts.
And just so we exclude my ineptitude from the equation, I was on the set of "The corpse bride" of Tim Burton, and they had the exact same issue.
Stop motion has to be reprocessed in post for light balancing to get a usable sequence (we talk of taking the same set picture twice in the space of half a second...).
So I personally keep only analog devices in mind when comparing with my digital realm.

So, if we move on to the analog, I'd say we have a few more issues to take into account, that the VrayPhyscam at the moment does not.
Namely, the "loss" of F-Numbers when a Higher number lens is used, and the variation of that "loss" between makers of lenses.
I have a nice, very old, very lovely camera which has a 50mm, a 110mm zoom and a focal doubler.
Shooting the same thing with the 50 and the 110 means i have to compensate with the F-Stops to get a similar brightness. All the more so when i screw on the focal doubler (yeah no bayonet lenses...).
Since that piece of kit is a cheapo, i lose 2 F-Stops of brightness.
On top of that, film has development, that can be compared to post.
Talking only of ISOs makes little or no sense.
A 100 isos film from maker A will have a grain type, that from maker B another, roughly of the same proportional size. Both will have a smaller grain compared to the 200 ISOs of the same maker, and so on, but there is no rule cast in stone.
A picture shot within a church with no flash can be extremely bright given enough exposure time even with a 100 ISOs, and that can further be enhanced during development of the film and printing, thank the god of photography (or i'd have thrown so many pictures away, lol).

Ultimately both in the real world, and in the digital one, the only absolute i care about is the quality of the finished image.
The idea of getting to it in a simple way, somewhat faster and more precisely than previously possible is what thrills me about these "physical" tools.
I really care little about rematching perfectly a real life picture in the 8Bits of a Monitor. There is a LOT of room for correction and error both in the digital and in the physical world, most of which is out of my hands.
What i wanted to point out with the tutorial was that simply whacking sun and sky and physcam in a scene wasn't nearly good enough to get the renderer to work in the right ballpark, as light shines on surfaces and within mediums.
So that those too have to be brought within reasonable ranges as well.
The perfect render is closer, but no cigar time just yet

Lele

[gilicom]

Hello Lele
I would like to thank you for the very good tutorial.
You explained a lot of stuff for me.
I havea question about the vra sky..
it is still only .5 in the color float , is there a way for getting it up to around 1.0 and still have the same predictable results? or it dosn't matter that it is still around .5
gili

[studioDIM]

Hello Gili and thanks for the kind words
In reality the tutorials isn't "correct".
It only shows a mean to an end.
Feel free to ddarken the colours further,and raise the ISOs (or lower F-number, or raise exposure time) to get a brighter sky in outdoors, if you so desire
The script to mass darken materials and maps has spinners to allow for a different number than 0.255 as RGB output level, so to leave freedom to the user.

Have a Merry Christmas and a happy new year, you all!

Lele

[efx2004]

Thank you so much for the tutorial.

Regarding the script, when I use a mask material on diffuse slot, the script changes it to vraycolor material. Is there a way to fix that?

Best Regards..

[Micha]

Hi Lele,

today I have seen video tut 3. My english is not so good, and hearing english is difficult to understand. Maybe you could explain the 0.255 method in some words here.

My thoughts:

- physical correct the brightest real life white reflect max. 80% of the light
- if we use 255, than we set 100% reflectance and this is not physical correct. Light between white surfaces bounce to much from one surface to the other. This cause flat lighting at interior renderings.
- a correction is necessary
- if we set 0.255, than the brightest white reflect only 25% of the incoming light - dosn't seems to be physical correct. This cause, that indirect light bounce much less than in reality. Side effect: render times are lower and less GI noise. I suppose, interior renderings will show to dark shadowed areas.
- if we set all color at 25%, than the relation between white and heaven environment is changed to. OK, we get a darker background, but this not physical correct again.

So, what are the problems we have:
(1) white should reflect 80% only
(2) the environment should be visible in the background

(1) here is a 0.8 multiplier physical correct
(2) The human eye dosn't work linear. If I right informed, it work expontial. The same for classial film material. Cheap old CCD cameras work approx. linaer and catch low contrast only. So, if we like to catch high contrasts, than an expontial color mapping is recommended. The "Reinhard" mapping do a great job here, the user can shift between linear (burn 1) and exponetial (burn 0). Burn 0.6...0.8 is a good start point.
Also, I'm not sure how 3DSmax handle it, so far I know, Vray works internal linear. If we are seeing renderings ot computer screens, than a gamma correction of ~2.2 for output is necessary. If we don't do it, than the contrast of lighting is to strong.

I have used Maxwell Beta for a long time: my experience for photoreal images is:
- don't use white colors brighter than 80% (each RGB channel should be below 80%)
- use gamma 2.2 output (vray user must manual correct input color/textures with gamma 1/2.2=0.4545 - Maxwell user get it per default)
- use the burn value to catch high contrasts

Merry christmas,
Micha

PS: Maxwell seems to use an internal correction for 100% white colors, so to much GI bounces are avoided per default.

[studioDIM]

Thank you so much for the tutorial.

Regarding the script, when I use a mask material on diffuse slot, the script changes it to vraycolor material. Is there a way to fix that?

Best Regards..

Will check the mask material.
There are some inconsistencies in the maxscript side, so i'll try and cater for that type of map as well.

Lele