How to work the camera and lighting.

[tussilago]

Hello everyone!

So, I have been thinking a lot about this. Just wanted to check how your thoughts are on exposure and lighting of a scene.

With the Vray physical camera, you have all the same settings you would in a real world camera, but the main difference from reality being that you also have the setting for ALL othe lights, and also throw scale of the world in to the mix.

Say i am photographing a livingroom IRL. Then I can't simply dial in the sun if i don´t get enough light, i have to use the camera exposure, and maybee some extra lights. This is what gives me a headache with Vray. There is no constant, no default value that I know to be true for the "world" light.
The difference in HDRI maps is huge, and you can't just set it to 1 in the render-multiplyer and know that that is a default value.
I notice how the automatic exposure function in Vray NEXT uses the ISO value to correct exposure, but is that not missleading? On a real camera, using a ISO over 800 would give
extremely noisy results. When Vray for example tells me that i need an ISO of 3000 then I guess my scene light is insufficient, so i raise the value of the HDRI multiplier, but if it gets to a really high value, then I guess I will loose information in the HDRI as it seems to be all just white.

Sorry for the confusing post, let me try to boil it down to a few questions.

* Is it a good idea to try and get a default light setup that is somewhat accurate to the real world? Vray sun value 1 (that always seems way over bright) HDRI render multiplyer 1 (that is always way to dark for interiors)
* Does the ISO value have any effect at all on the grain / noise of the render?
* Do I loose "information" in the HDRI, and thereby my shadows/details in the render if i multiply it too high (lets say 30, then all is white?)

I am working on an interior right now. Big windows in a scyscraper. So I am thinking that in the real world this would not be any problem at all, If anything I would get too much light in to the scene. But in this case I use an HDRI (PureLIGHT HDRI, though I tried a lot of others as well) and to get the right exposure (F-4, ISO 100, shutter 180)in some places i need to raise the multiplyer of the HDRI
to 35. This does not seem right?

Glad to get some input!

Thanks
// Jonas

[DanSHP]

I follow some real world camera settings for the scene I'm working up, highsun/dusk/overcast/interior etc (there's a chart on here somewhere). I then increase the HDRi multiplier as needed to get enough light into the scene (varies from supplier to supplier) I will literally go as high a i need to. I then alter the RBG value of my maps (in output) so that they are not too bright or too dark.

This works for me.

When you say you loose all the information in the HDRi, do you mean it just goes white? This happens in the material editor (it's not an issue, if your camera is exposed correctly) when it renders it should be ok. If HDRI sky is coming out too white in the background of a render, if you are working in 32 bit ext/tiff (16bit should be fine too) you can just adjust this in post really easily and reclaim so of the lost information detail.

In my opinion, you limit yourself using 100% scene real world values. From experience attempting this myself, you always end up tweaking them anyway to get the render looking as you'd like it.

[tussilago]

Yes! That is the way I go about it too. My point being for example regarding the HDRi, that if you multiply it enough times it will eventually turn all white. At that point, have you not lost all the advantages of using HDRi lighting? Both regarding reflections but also the hues, contrasts and complexity. At that point it must be all white light coming from all directions with the same intensity?

So basically what I wonder if; if I render an interior scene for example, and use an HDRi with value of 1 or 2 (say that that looks about right in the viewport) and then use the ISO setting in the camera and crank it up to about 3000 to get a descent exposure. Would that render look EXACTLY the same if I instead increase the HDRi to lets say 60 in the render multiplier, and use like 100 ISO on the camera?
In the real world (if you would be able to do such a thing I would imagine that you get totally different results...

[glorybound]

One of my biggest frustrations is how different HDRI's are from each other. Some are fine at a multiplier of 1 and other is as high as 30. I do the same thing. I use real-world camera setting and I adjust the HDRI multiplier until it works. Without the camera being the baseline, it would be a mess.

[DanSHP]

Yes! That is the way I go about it too. My point being for example regarding the HDRi, that if you multiply it enough times it will eventually turn all white. At that point, have you not lost all the advantages of using HDRi lighting? Both regarding reflections but also the hues, contrasts and complexity. At that point it must be all white light coming from all directions with the same intensity?

So basically what I wonder if; if I render an interior scene for example, and use an HDRi with value of 1 or 2 (say that that looks about right in the viewport) and then use the ISO setting in the camera and crank it up to about 3000 to get a descent exposure. Would that render look EXACTLY the same if I instead increase the HDRi to lets say 60 in the render multiplier, and use like 100 ISO on the camera?
In the real world (if you would be able to do such a thing I would imagine that you get totally different results...

What are you referring to specifically re the HDRi coming out all white? The material editor preview? Or the way it renders out in the backplate of a render? Or just the way the HDRi might blow out a scene?

If you set the exposure correctly the camera will only let so much light in based on it's settings. Right? So you're not loosing anything by cranking the multiplier up, you just control how much of that HDRi light information is let into the camera.

If I set my camera up to some general real world settings, for say an interior, and there isn't enough exposure, i will just crank up the multiplier until there is enough and then I will probably have to alter my map/material RBG values, so nothing is over exposed. I rarely change the camera settings.

As Glorybound said, the camera is your baseline, everything else in the scene should feed into it.

[tussilago]

What are you referring to specifically re the HDRi coming out all white? The material editor preview? Or the way it renders out in the backplate of a render? Or just the way the HDRi might blow out a scene?

If you set the exposure correctly the camera will only let so much light in based on it's settings. Right? So you're not loosing anything by cranking the multiplier up, you just control how much of that HDRi light information is let into the camera.

If I set my camera up to some general real world settings, for say an interior, and there isn't enough exposure, i will just crank up the multiplier until there is enough and then I will probably have to alter my map/material RBG values, so nothing is over exposed. I rarely change the camera settings.

As Glorybound said, the camera is your baseline, everything else in the scene should feed into it.

Ok. Maybe I am just thinking backwards about everything and worry too much

But at the same time I cant really wrap my head around the ISO setting of the camera. As I asked before, is there really no difference between using the HDRi multiplier or the camera ISO setting in a render. I get worried (as would in real life) if the ISO number exceeds 1000, but maybe it makes no difference on the quality of the render?

Should probably make some tests on the subject when I get some time...

[jackieteh]

hi do you mean vray next use only the ISO value to correct the camera exposure, what about others setting like f-number and shutter stop, didn't change?

[tussilago]

Hi!
Yes, it seems like VrayNext only corrects with the ISO value if you use the automatic exposure.

[tricky]

I've often wondered if increasing ISO on a VRay camera would increase noise. I posted a question some years back on this, but I can't recall the response. Of course, upping the ISO (or 'gain') on a real camera increases noise, though modern large sensor cameras deal with this much better than 10 or 20 years ago.

My understanding is that having your VRay camera set at a really high ISO means that there isn't really enough light bouncing around the scene, so your GI calculation and rendering will be slower. VRay 'struggles' with dimly lit scenes, right? For this reason, I try and keep my VRay camera at pretty much the default and then adjust the lights until it looks about right.

It is weird, having real world settings but then complete BS controls too Some forum user's signature reads (or used to read) something like "Why on earth do we need physical camera settings in a virtual model?" and I understand that, but for those familiar with photography, it can help in understanding a scene exposure.

(why the default physical camera is set in EV is beyond me!)

[Vizioen]

I've often wondered if increasing ISO on a VRay camera would increase noise. I posted a question some years back on this, but I can't recall the response. Of course, upping the ISO (or 'gain') on a real camera increases noise, though modern large sensor cameras deal with this much better than 10 or 20 years ago.

My understanding is that having your VRay camera set at a really high ISO means that there isn't really enough light bouncing around the scene, so your GI calculation and rendering will be slower. VRay 'struggles' with dimly lit scenes, right? For this reason, I try and keep my VRay camera at pretty much the default and then adjust the lights until it looks about right.

It is weird, having real world settings but then complete BS controls too Some forum user's signature reads (or used to read) something like "Why on earth do we need physical camera settings in a virtual model?" and I understand that, but for those familiar with photography, it can help in understanding a scene exposure.

(why the default physical camera is set in EV is beyond me!)

Haha omg I'm not even going to begin answering this, but this is just wrong on so many levels.

[tricky]

Haha omg I'm not even going to begin answering this, but this is just wrong on so many levels.

Erm, care to expand?

[Undersky]

Erm, care to expand?

He's probably just meaning that no, ISO in rendering and ISO in camera film response is not the same thing. The noise you get in the renderer is due to not enough samples. The renderer isn't simulating what happens when light hits the camera sensor (or the film) with regards to a specific ISO setting.
The reason why the auto exposure in the Vray Physical camera is because it's the one setting of the three (f/stop, ISO and shutter speed) that only affects the exposure without altering the image like the other two settings do.

The way I go about lighting a scene is to start with a VraySun. I set it to somewhat the kind of lighting I'm going for and exposes from that. Then I delete it (or just deactivate it if I need the position for aerial perspective) and load in an HDRI, that I change the multiplier on until it matches my exposure settings. I usually use a grey override shader for everything aswell.

Some forum user's signature reads (or used to read) something like "Why on earth do we need physical camera settings in a virtual model?"

I remember those days, but took another route, which is I'm infinitly happy about. It was actually Chaosgroup's decision to move physical camera settings into the CG-camera that prompted me to get my first DSLR and learn photography. "I might as well keep with the times and see if I can benefit from this and become better at 3D." I thought to myself. And now, some eight years after I can say that learning photography and composition has been one (if not the most) of the most important and rewarding ways to improve on my creative eye and vision. But enough off-topic.

[fco3d]

To my understanding, from the ground up V-Ray followed Physical values and kept a consistency between Physical camera, exposure and V-Ray sun/Sky and photometric lights.
So if you need something to have as a reference, as mentioned earlier you can use VRay Sun, adjust your camera exposure to fit your scene and go from there.

Having said that I know many people have a misconception of how light 'should behave' when in real life is not what we think. I've seen many questions regarding exposure of large rooms with small windows and the only light source is the Sun/Sky, people question why is so dark when in reality, that's the way it will be in real life, that's why we have artificial lights in every room.

Also, you need to consider that the human eye exposes very different and more efficient than a photographic camera, and V-Ray follows Photo/video cameras systems and not Human Eyes.

To add to this big mess we have hundred of different quality of HDRI around, many people belive that to do a physically correct render you need to use an HDRI to light your scene. This is no true, this depends of the source of your image, online you can find many very bad quality HDRI some of them barelly fit the standard, that's why you see so many bluish renders, is not because the HDRI is giving you the Sky color, it is rather the HDRI is short or lower in information and is clipping the colors to a bluish tint. Of course, this can create some interesting results but that will fall in to the artistic aproach for your image.

When you load an HDRI the first brightness you get is the default 'Stop' that the photographer chooses for that image. that's why you'll have some HDRI that project by default a lot of light and other don't.
Correct HDRI panoramas should be able to produce Shap shadows if the sun is up a bright. If you have a sunny HDRI and you don't get sharp shadows then that image doesn't have enough light information or is not well balanced between the Sun and the sky.

So to resume your original question, you should use Physical camera with common photographic values, for instance for a sunny exterior day Ftop 16, ISO100 Shutter 1/100 or EV 16, and adjust as you need from there.
If you HDRI image is too dark, then just increase the intensity, if your HDRI doesn't have enough light information, then the colors will clamp and won't generate more light information, you may get some strange colors spills in some cases too.

Hope this helps you to clarify your questions.
Best luck.
Fco.

[joconnell]

Short answer for the minute:

There's three parts to the final results in an image - camera settings, amount of light, brightness of materials. If you have one part you're in pretty much the same situation as inventing everything by trial and error. If you've got two parts then you can work out the other variable.

Reference or data is key if you want to get accurate results and there's huge benefits for working in this way - if your material brightness is accurate then you can happily use them across projects knowing that they'll look pretty good right away. Here's what you can use to measure each value:

Camera settings - exif data from any photo you take
Material values - either buy a colour sampler like a nix or cube sensor where you can put the sensor over an object and press a button to get an rgb value, or buy a colorchecker chart which has known values and take a photograph of the chart beside whatever the colour you want to match is - if you colour correct the photo so the chart's values roughly match it's known values, the object you put it beside should be quite close
Light values - these are tricky since you often have light from the sky through windows or lots of bulbs on at the same time but a lux meter is cheap and can be used to record a light value. If you had to match the light in a room from a single ceiling light for example, you could put a lux meter underneath it, measure the distance from the lightbulb to the luxmeter and record a value. If you make an area light in vray that's the same shape as your real world bulb and put a vray light meter helper the same distance from it as your real world measure, you can adjust the vray light multiplier until the light meter helper gives you the same value as your lux meter.

A big problem is that photorealistic kinda of means make something look like a photo. A camera isn't designed as a scientific measuring device to give you really accurate results, it's a thing to make pretty pictures. Likewise your eyes don't see colour accurately, they put a bit of exposure adjustment on things too so the colour or brightness that your eye or your camera sees can be a little bit different from what it actually was, and all of those tiny differences put together can end up giving you results that are different enough to make your render look unnatural!

[^Lele^]

John put it very well, as usual.

The poor man's rendering equation:
R = A x L x C

where

R is your rendered image,
A is the shaders' Albedo
L is the lighting falling on said shaders
C is the camera exposure

With a modicum (modicum!) of maths, is trivial to figure out that setting as many variables to 1.0 will make evaluating the others simpler.
For example, set lighting to a white (1.0f L becomes 1.0) dome, turn exposure off (ie. set C to 1.0), and whatever value you pick in the rendered image will correspond exactly to your shader's albedo (in fact, it's a perfect test to verify if a shader is energy preserving. also called Furnace Lighting, or GanzFeld Lighting).
Having unknown values for two of the three multipliers isn't conductive to figuring a scene's ideal setup.

As for what is a "1.0" for the camera exposure, it's very easy to find out:
Render a perfectly white background (1.0f), with whatever camera exposure you have.
Pick the value, and divide ISOs by it.
The camera's now "normally" exposed, your background back to white 1.0f.