I think we are actually on the same page here although I’m more so talking about the process rather than what the end result should be. The workflow makes more sense to have units, I’m not saying it is realistic and produces realistic renders as result - renders never do look realistic most of the time without post as you say. What I’m saying is, that it’s nice to have a reliable method of setting up lights and materials without a lot of guesswork anymore than what we used to 10+ years ago.
Like my comparison earlier with physics in CG. It is much easier to have gravity set to 9.8 and heave typical real world weights in order to get predictable results. Otherwise we’d be using numbers which have no meaning at all. The physics in Maya, again, will never be 100% scientific - ever, but it is great to be in the same ballpark when doing that sort of stuff. The same logic applies to rendering and lighting in my eyes.
After all that we can apply artistic flair to the real world settings.
the line I draw is where it becomes utterly impractical to create the perfect environment that the render engine will accept, and perform in a reasonable amount of time.
I agree!
You simply cannot justify something ‘just’ by saying it’s realistic. If that made sense, then there would be no cg lights, no interpolated GI, because they’re not realistic. Even HDR’s have little basis in reality, because their light doesn’t change as an object moves around. We’re not scientists, we need good looking images more than we need accuracy. And we need to be able to make changes according to artistic direction, which usually has 0 basis in reality.
I agree with what you say here as well. I don’t think any pro has submitted raw renders to clients, as we all know they usually get smashed in post. But as I say earlier, it is nice in rendering to have an idea or what the scene will render like initially.
“Those innacuracies give rise to other innaccuracies. And if you’re shackled into doing everything else “realistically”, you can get stuck with artifacts, or with single frames that take days to render, or have to shoot all your HDR’s over again, then redo all your lookdev, or other drastic measures.”
Does that really not make sense? Of course I want things more realistic. But I have to be able to render them, for that to be any advantage at all. Until all these things are done realistically, then we NEED the flexibility to be able to compensate for them. Most of 3D rendering is still a HACK. You can’t give us a hack, tie our hands, and still expect things to come out photreal.
Yes, inaccuracies and cost in increased render time too! Moore’s law, with faster hardware, rendering times stays the same over the years! I think we do have nice flexibility now with 32bit float exrs passes.
It’s being locked to a see-saw, and clamped, that I object to.
What do you mean here specifically? clamped? We have the ability to still use Maya shaders in most 3rd party renderers which are not realistic (because they can go past a diffuse value of 1.0 for instance). We can also choose to revert to the past and use linear lights with no relation to real world.
Unless you’re working in the scientific or legal field, aesthetics must come before physical accuracy. Every one of the most incredible images you see today are touched-up, tweaked, and color-corrected. None of them are 100% real. If accuracy / realism / perfection doesn’t serve aesthetics, it’s discarded. If you’re working on LOTR, and ‘realism’ turns out looking ugly, do you keep it that way? Not if you want to keep your job, you don’t.
Agree 100%.
On the side I work as an architectural photographer - the amount of cheating that takes places and spending a tonne of time in post certainly justifies your logic of aesthetics > physical accuracy. Could not agree more.
I guess, my main point earlier and during this post is that it’s just easier to work with a base set of measurements and rules that are very predictable in their outcome. It is not scientific or accurate, but it’s reliable. I think reliability is the key - even Vlado these days says he uses Universal Settings in rendering because it is reliable. It may take more machine hours to render, but reliability and predictability means more time to the artist and thus more time can be made with less tweaks, less settings and more on image and post.
Does everyone here think that realism = aethetics??
No way! Not at all.
If that were the case, then the Mona Lisa, and all the master’s works, would have become obsolete, and “ugly”, with the invention of the camera. I wondered which was more important for a little while too, but it soon became obvious. Aesthetics take priority.
It is an interesting point. Back to my ILM reference, Rango is a very non realistic movie in look! But their workflow on that particular job was setup to be physically accurate, not 100% scientifically based or anything - but because it is possible to work like this as it is predictable results in lighting and camera. Cinematography in CG is easier now we have similar ways in which to use the virtual camera like the real world one. After the base of physically plausible settings, artistic flair is added on top.
This my point, I hope I explained it properly. Again I do think we are on the same page. But perhaps you might disagree with the workflow, that’s cool if you do too. Thanks for the interesting discussion.
