this is the default setting I have for the sky texture
I noticed that the texture generates way too much light and the exposure needs to be dialed back
Enscape and similar apps do this automatically. Now I am curious who to best explain this to students.
I recemember somebody having said that the sky texture emits the correct amount of light we are just not used to it and need to adjust / tonemap it to make it feel more the way how our eyes adjust.
after getting your eyes dilated everything is much brighter too as an example.
does somebody have some good solid explanation for what math / energy wise is working here?
I mainly adjusted things by feeling - but when explaining it to students I would like to add a little bit more technical background.
Maybe i wander from the subject, but i never got this thing right: exposure (and aperture) “originally” from photography… and most of the tips say something like:
Exposure can seem complicated, but it is one of the most important technical topics to know if you want to take high quality photos. The best thing you can do now is go out and test the suggestions above for yourself. Play around with your exposure settings, … [https://photographylife.com/what-is-exposure]
This sounds nice:
The main effect of photography exposure is to capture the image with a determined brightness. When you shoot a perfectly exposed photo, you are capturing all the details in the highlights and the shadows. [https://capturetheatlas.com/what-is-exposure-in-photography/]
but it doesn’t explains the physical background…
And now looking at this i even found:
Anyway: Even if there is some physical model/math it would be nice to have some energy attribute in the sky texture node to choose an almost correct exposure (as some photographically tutorials suggest to use the auto mode…).
Or just choosing exposure in that way to have an overall brightness/luminosity/intensity or gamma of IDK value??? (i’m thinking of something like Imagemagick identify -verbose picture gives as image statistics…)
That’s nonsense unless you control the lighting. Lighting with the sun & sky you don’t have control, and no camera can capture all the detail in a single shot. In rendering, the trick isn’t to get the correct exposure, but setting up the scene in a way that looks good while preventing exposure problems - as well as finding the correct ratio between sun & sky and artificial light sources but that’s a whole different story.
I.e. in real life it’s impossible to get a single exposure to light up the interior of a space, and at the same time have the outside look perfectly exposed with no loss of detail. For that the photographer may use “window pull” tricks in post to balance it out somewhat but without overdoing it. Or choose to fill in the shadows with flash.
Exposure only becomes complex if you try to design some automatic exposure function, trying to guess what the correct exposure should be. I.e. shooting bright snow, we know to correct for this guessed value and compensate by adding in exposure compensation - expose more than the camera thinks it needs, otherwise the snow would turn out quite grey.
I.e. if shooting a subject in sun & sky conditions, if we try to get the details in the sky (like clouds), the shadows on the subject and subject itself becomes to dark. There is no exposure tricks that can remedy this. Getting this “perfect exposure” in this case would involve lighting up the subject with additional lights. Either by large lighting rigs or bounce boards, whatever suits. Watch how outdoor movies are made, there is often significant gaffer work involved unless going for the harsh look of natural light only. I recommend watching some cinematography and gaffer tutorials - completely unrelated to rendering - just to get an understanding of how the professionals thinks wrt lighting a scene to get their “perfect exposure”.
Also, “perfect exposure” is subjective. You may have an image that is dark and moody without any highlights at all. It’s part of the theme, and while it may look underexposed, it’s actually perfectly exposed.
i think to be fair here he means an evenly exposed photo that tries to balance out details in shadow and highlights. Thats also why you shoot in RAW if you want to develop the image.
Thats could be… but it was a citation with given URL… so please don’t quote me like that. And i even said that i don’t get the explanation at all and did found something that says it sucks…
On the other hand: speaking of “perfect” is subjective… welll that’s exactly the question or the declaration of this tips/tutorirals because their conclusion is practice, practice, practice…
Maybe a better question: Is there a potential measure methode to get a “not too bad” exposure?
(By the way: once i read something about old b/w photographs with steam locomotives which where “natural/lifelike” and they where additional lighted… already back in the day …)
Just noticed, why do you have negative sun elevation as standard? One thing to keep in mind is you need an outside to bounce the sun into the ceiling, as the texture doesn’t have anything below ground level. And don’t be afraid to tweak the knobs.
We don’t have auto exposure in Blender, and if we did, it would just like in a camera be a best guess. But unlike in a camera, we can tweak our exposure at hearts content without loss of detail, whereas in a camera anything above and below cliplevels would be lost forever. I wouldn’t worry too much about pre-setting exposure correctly for a “correct exposure”, but rather tweak the rendered lossless data with exposure or in post.
One thing people seem to forget is that the sun is a warm light, particularly at lower angles. Whereas the sky is cooler. Use blackbody node to set interior light sources’ color temperature. Now you have the correct relation between indoor lights, sky, and sun at any elevation. And whups, you now face the same issues a real photographer have
For a camera in clear sky and sun frontlit normal subject (i.e. a face), the norm is to use sunny/16 rule and then know what settings this translates to in manual mode - the subject will not be overexposed. For rendering, we just need to figure out what gives a close approximate. Around -4 to -4.33 maybe? At -4 and sRGB checker with 0.8 and 0.65 is barely distinguishable, using a 30 deg sun elevation, but 0.8 is a quite high albedo. A lot more leeway with Filmic obviously. That setting will not be enough to floodlit an interior, maybe up to -1 or thereabouts. For sky only, maybe +3 to +4? Need to experiment with what works. Interior lamps can compete with and complement daylight, but cannot outperform sunlight other than provide localized fill.
If you are only using the nishita sky to light your scene, when you change the “exposure” in the rendering settings it has practically the same effect as changing the strength of the background shader (increasing the amount light). I see no difference.
Edit: the colour node shows the “raw colour” values of the materials.
I just realized I was adjusting the exposure in the film tab, if you adjust exposure in the colour management tab the ratio is different but the effect the same.
I suggest keeping Nishita sky background to 1 (for a normal sky) as a fixed reference, and tweaking interior lights to feasible values; floodlit with sun these aren’t gonna matter much, they complement and fill sky only, and exposure shouldn’t change much between sky only / sky only + interior / interior only. That makes the interior lighting assets far more re-usable.
Yes I agree, I was trying to say that the exposure settings directly compensate the lighting amount.
What confuses people with the Nishita sky (or a high quality outdoor hdri) is that lighting values are extremely high. This is actually normal as in the real world sunlight is blinding (don’t look directly at the sun!)
Our eyes compensate by retracting and dilating the pupil and our brain does a few tricks as well, but in extreme situations we have to squint or put our hand up to block the sun that goes straight into our eyes, or squint when we go out of a dark interior (even if we are not facing the sun) to give our eyes time to adjust to the new conditions.
The Nishita sky does a pretty good job in giving the correct values but depending on the image you want to achieve you normally have to compensate turning down the lighting with the exposure settings, especially for exterior renders.
If you ONLY use the sky for lighting you can do the same by turning down the background strength.
That is why I am looking at a method to teach lighting better to the students.
We can calculate the lighting via formulas or use apps like DiaLux. But this is not very visual.
I am interested in stimulating lighting via rendering. I know that this is a nearly impossible;e task.
Displays need to be calibrated
Materials need to be correctly setup
Render engine well - does it really realistically calculate light too?
sofar we always balance an image more from an aesthetical point of view
I would like to be able to set sunlight sky and interior lights and have a process that would generate a somewhat realistic result
There was a debate what the default intensity of the node should be. Let the sun intensity be similar to the real world thus requires exposure offset or defaults to a value that requires no offset. Core devs were involved and no consensus was reached. At one point the creator considered writing an essay to ton just for this particular design decisions.
As for the specific parameters I asked the creator.
The Nishita Sky node has a Sun Intensity value which is a sun intensity multiplier, 1 means default real world sun intensity. The Background node (which the Sky Node is attached to) has a Strength input, that is simply a multiplier of the whole output of the sky texture node.
Teach in reference to what? Lighting engineers or rendering out art? Use DiaLux if teaching for lighting engineer; the key point for them is to adequately light working surfaces and spaces wrt incident light. In Blender all we get is wrt reflected light. While light reflected does count into light received for indirect lighting, in Blender we cannot check incident light.
False color in DiaLux is used to indicate incident light.
False color in Blender is used to indicate reflected light.
Two different worlds, and two different uses.
Go outside in sunlight with a DSLR and set it to manual exposure. Sunny 16 rule dictates front sunlit face should be shot at f/16, ISO100, 1/100th second. May go a little dark based on conditions, but shouldn’t be overexposed. Now go inside and shoot a space - it’s going to become under-exposed unless you have a hellufalot of sunlight pouring onto near perfectly white surfaces.
Ops never heard about this. In Interior Viz we pace lights and render and adjust all well artistically - which is fine but for beginners hard to know if it is correct or not.
So my curiosity is if there is a way to use rendering as a tool to somewhat preview of the lighting I would like to implement could work.
In reality we might do a RCP and do a light calculation on paper but for bigger projects this type of a task is often given out to specialists.
since we are visual people I am interested in seeing if I can tie all this together into a workflow that is somewhat believable and not as complicated for students to learn and use.
Talk it over with the architect/lighting engineer and let them make the call. For home spaced I’d expect interior lighting to be the driver, since you need to be there during dark, with sun & sky just being what they are. Don’t take my word for it, I’m not into real archviz. I’ve seen plenty of awesome archviz artists doing tutorials I would say was plain wrong from my background. You’re there to make it look beautiful, to sell the idea. You don’t have any control if the future owner of the space would use your decoration assets.
As a non archviz artist, my approach would be relatively close interior light assets using 1W (for Cycles) and control it via nodes (blackbody for color and IES/fixed value for strength). Look at realistic images (not tweaked for sale ones) of how bright indoor lighting appears compared to window size and how much (non sun) daylight they let in. Remodel the space and tune your indoor lighting assets to match and save out the asset and it should be good to go for any re-use.
But now you need to start thinking like a real photographer doing indoor spaces. Use fill flash? Do window pulls? We have better tools for this than real life. We can make the transparent shader for camera rays darker than for everything else, thus “dimming” the exterior which would often appear overexposed otherwise. Same as the photographer, don’t overdo it.
Similarly, check out professional cinematographer tutorials and how they think doing indoor light, faking everything along the way. Their intent is often beauty and exposure control, not ultrarealism realism - plausible is good enough. They’re also working with a media that don’t store all information, so exposure control is a tad more critical.
Instead of looking for a formula to do lighting all the time, instead set it up naturally first, and since that may end up looking like garbage, students should learn how to improve the bad situation. Maybe all you need is a flash in the middle of the room and lower the exposure? Maybe all you need is “gel up” the windows (with shader tricks)? I would assume there are differences to every project that forces you to think new.
