Animation with GI that doesn't flicker

[vlado]

Is this too much to ask?

Yes

Seriously though, if there was an easy way to do this, it would have been done already.

Of course, there is already a method that does not have flicker issues - use brute force GI, and you are set. You will need more computing power, but that is more or less straightforward to obtain.

Best regards,
Vlado

[Morbid Angel]

this takes us back to the whole rendering theme in the first place. If rendering was easy any one could do it. But its not. Gi that does not flicker is just an element which takes time learning. I dont have a problem with it, neither do the professional artists who work with it.
All Im saying is, it takes a bit of practice and time to learn and understand.

[tricky]

Yes

haha. LOL. I knew it was. It is just deeply frustrating that having been in this business for over 10 years, and using Vray since its infancy, we still have major issues with moving geometry if we want to use GI. There have been massive advances in the way we render our stuff thanks to Vray, but animation still is the achilles heal (sorry, no idea how to spell that!!!) of 'budget' archvis.

Oh well, 'Dome Of Lights' (Circa 1997) here we come!!!

[sv]

there are also loads of different ways to approach animated scenes, rather than just setting up the animation in max and pushing render, and complaining that without SP3 it's not going to work...
Like having people moving around in a room; you could always split the scene up into BG with cached light and moving camera, render this separately, then render the characters moving separately, then render shadows separately if you need to, you could always break it down into ambient shadow passes and key light passes if you really needed to do that....

all i'm saying is there are loads of ways of getting it to work; it's not necessarily a problem with the V or any other renderer.

This isn't directed at anyone, I've just been reading loads of requests for SP3 recenlty as though the current Vray is full of bugs, because in my experience it isn't at all....

[tricky]

There certainly are lots of ways to go about animation, and we have employed a few ourselves. From using Lightscape and saving out a lighting solution to the prepass animation settings of Vray 1.5. Nothing (yet) has made the process particularly simple or elegant.
It will get solved at some point in the future, I am sure of it.

[MikeHampton]

Gi that does not flicker is just an element which takes time learning. I dont have a problem with it, neither do the professional artists who work with it.
All Im saying is, it takes a bit of practice and time to learn and understand.

I take a little exception to that statement. I think that 99% of the people here in this forum are professionals.. and 10+ pages of comments on this topic (not to mention countless discussions/threads previously) are clear indication that it is indeed not quite so simple.

I see exactly what Tricky is saying. To use the default max scanline renderer as an example, you got exactly what you expected. It's rock solid, super fast (if you arnt' doing a lot of reflections/etc.), and you can be almost 100% confident that when hit the render button at night, you would come back in the next morning and everything is going to be done, rendered and ready to use with no suprises. Any "glitch" would 99% of the time be a user error, like forgetting to hide a wall that the camera is moving through or something. When you are on tight deadlines, smaller budgets, and only have a few computer render farm.. all this is critical.

I don't think there is hardly an issue with doing a static rendering. I think most of us may have a small issue here or there (like DR not working as expected or something), and most, after getting a few animations under their belt, probably have gotten over the hurdles and probably have a pretty good process/workflow for an animation of a static scene (only moving camera), but even then.. "problems" crop up that are frustrating. It's not 99% rock solid yet... but it's getting closer. Of course then suddenly throw LWF into the mix and start that learning process all over again!

But when suddenly you are animating (and not just the camera), is currently where things seem to fall apart. Sure, you can use brute force GI, or you might render somerhing in 12 different layers and composite them together. That might be fine if you have 50+ computer renderfarm, and rendering in layers may be critical on very large projects. But for those who don't have those kind of resources and need to have 1000's of frames rendered overnight, with no suprises, it can be frustrating to think you have something figured out, but then the next project you work on using the same process/workflow suddenly doesn't work as expected on another very similar scene. And you don't have the time to figure out what's going on, other than pulling 20 hour workdays, to still meet your deadline.

I'm not necissarily "blaming" vray for this. Vray gives us global illumination and the ability to render millions/billions of polygons in a reasonable rendertime, I think that vray is nothing short of amazing. I hate when I have to use scanline due to not wanting to jepordize a certain projects deadline. But as with almost all render-engines.. those rendertimes come as a result of finding tons of cleaver little "cheats" to get those results within those render times. But where those "cheats" don't matter on a static rendering.. they add up to inconsistancies and "problems" from frame to frame when doing an animation.

Obviously those inconsistancies, can be overcome. People are doing some amazing animations. But I always have to wonder how many times those frames where re-rendered to fix glitches/problems/inconsistancies/etc.? How many little "tips, tricks, and processes" does one need to learn and imploy to get around all these issues? It's certainly not simple and elegant yet. Not everyone has the luxury to re-render frames several times (or even just twice), or to render each frame in a half a dozen elements to composit together to get around issues. And especially not to figure out what all those little tricks are from one animation to another.

I like to quote Craig Zerouni from Digital Domain when he says that he "Has been coaxing images from computers for over 25 years". I think many of us are hoping that we can quit "coaxing" consistant GI animations out of our computers to focus more on just being creative.

I also don't doubt we will get there... either through more cleaver "cheats" that solve most of the problems with GI and moving objects/chaning lighting conditions or processing power becomeing fast enough to move to brute force (what I call the 'Maxwell' solution). It's just some of us are hoping to get there sooner rather than later.

Simple and elegant is why I think you see so many amazing things being done with Zbrush. Not that it doesn't have it's technical issues also with specific things, but at it's core.. just being a brushed based modeler... it's simple and elegant, can handle millions of polygons like it's nothing, and people have just taken it and run with it (that is if you can get over the frustrating UI of the damn program, my model is a "tool"? WTF? ).

[Morbid Angel]

I see exactly what Tricky is saying. To use the default max scanline renderer as an example, you got exactly what you expected. It's rock solid, super fast (if you arnt' doing a lot of reflections/etc.), and you can be almost 100% confident that when hit the render button at night, you would come back in the next morning and everything is going to be done, rendered and ready to use with no suprises. Any "glitch" would 99% of the time be a user error, like forgetting to hide a wall that the camera is moving through or something. When you are on tight deadlines, smaller budgets, and only have a few computer render farm.. all this is critical.

I don't mean to bring anyone down, or to say that some one is unexperienced but, you can't be serious in this statement?
If this was true, there would not be a need in other renderers.

[tricky]

I don't mean to bring anyone down, or to say that some one is unexperienced but, you can't be serious in this statement?
If this was true, there would not be a need in other renderers.

Not sure I understand your question. If 'what' were true there'd be no need in other renderers?

[Morbid Angel]

that the scanline is rock solid, super fast and etc. Yes it can be, but then it lacks all of the things we use in vray today. Without them, it cannot compare on many levels, hance it cannot be brought into this discussion as an example.

[MikeHampton]

If this was true, there would not be a need in other renderers.

I stand behind what I said there.. but my message was already too wordy so I deleted a couple of paragraphs. Let me just continue the thought.

"Of course we all use vray because you (unless maybe one is a dedicated lighting specialist) could spend days/weeks tweaking lights and still not get the same beautiful results you can get within just a couple of hours with vray. Not to mention being able to render tens of millions/billions of polygons, utilizing lot's more reflections, and dozens of other features that you can't do in scanline. But none of that matters if you can't meet an immovable deadline because you run into problems of inconsistancies in GI solution of the frames of the animation."

[MikeHampton]

And really my entire wordy message could be boiled down to this..

If it were so simple... there would be a 5 step process to render an animation that has moving objects. You do 1, 2, 3, 4, 5 and hit the render button. Maybe there might be a 6 or 7th step in certain situations (chaning lighting conditions, dark scene, whatever). But 10+ pages of this thread proves that it's not so simple... if it were this thread wouldn't last past 2 pages.

Not that there AREN'T workarounds and ways to do it... it just seems like it's more like there are 100 things to try and you have to figure out what combination of 75 of them are going to work for a given animation. And the next animation you work on needs a different combination of 75 of those 100 things... just sayin

[Morbid Angel]

If it were so simple... there would be a 5 step process to render an animation that has moving objects. You do 1, 2, 3, 4, 5 and hit the render button.

this is the case with sp3.
The gi animation there is far superior.

[tricky]

that the scanline is rock solid, super fast and etc. Yes it can be, but then it lacks all of the things we use in vray today. Without them, it cannot compare on many levels, hance it cannot be brought into this discussion as an example.

Of course it can - the point is not comparing the quality of renderers is it? Maybe I misread it. I thought he was talking only about simplicity and elegance.

I understand and agree with most of what Mike is saying. But I think he would agree that this is not Vray bashing; the team behind vray have done amazing things with this software and it has been allowing many people to produce truly stunning imagery. I am still stunned that we have been asked to pay no more money over the years of development (don't tell the Chaos team I said that!). However, a more straightforward (elegant?) way to create a GI animation is certainly on my wishlist.

Dear Father Christmas, I have been a very good boy this year...

[MikeHampton]

this is the case with sp3.
The gi animation there is far superior.

Well.. let's hope so. But I also thought that the "animation" mode for IR maps was going to take care of these issues, and it does help tremendously, but obviously it's not the magic bullet yet. And even when it does work adequatly for a scene... having to render an IR map for every frame AND the render time hit of merging several IRR maps together to actually render each frame, I find is often a deadline breaker.

I've been out of the loop for a few weeks.. when is SP3 available? Is it avaible now (as in not just in beta)?

[vlado]

It is still just in beta.

Best regards,
Vlado