Hi,

The subject says it all... I need any books, on-line material, videos, anything any one might know of on general composite techniques and methods.

I'd also like to know more about what can be done with AM's implimentation... I have the feeling there's more. I think I've gone through all the tuts on this that are 2 months old... anything newer I've not seen.

Rusty

Disappointing. Is anyone using AM Composite?

Cheers,
Rusty

It's one of those chicken-egg things. People need more documentation to do more with it, and it won't get documented more until people do more with it.

It seems potentially powerful but there's too much I don't know about it to know if I'm using it right. And when I asked questions about it they were deleted. I was very disappointed by that.

Compositing in any of its various forms hasn't gained much support (at least not publicly) in the A:M Community. This is a shame.

Compositing has a rich tradition dating back to the earliest days of traditional animation and uses tried and true techniques brought forward from film and theater. It can be found at the cutting edge of technological advancement throughout film and animation history. Consider matte paintings, live actors mixed with cartoons, multiplane cameras, limited animation, miniatures/stopmotion/train sets, special effects and a whole lot more and you can be able to fill a great many books. They have!

Compositing is a great time/money saver too. As such it is a subject of great interest to me personally.

There is a perspective on compositing where artists see it only as a cheat. Here anything that doesn't spill directly out of the renderer is somehow inferior, tainted, flawed... wrong. The costs associated with this puritan view to productivity alone is greatly compounded by this skewed perspective. You can see echos of this in 'photorealism'.

While everything can be built in three dimensional space... why would we do this?
Just because we can? Animation is all about reusing images. Why not reuse them.
A:M is built upon a foundation ready to leverage reuseability.
A:M has the compositing tools.

As a productivity multiplier alone compositing should be factored into all 3D work.
Its a tool we should use a whole lot more.

Google alone will return a vast array of information on compositing.
But for general composite techniques and methods I suggest looking at articles on film and special effects. Don't start or stop with computer graphics though or you'll sell yourself really short.

I'd love to see the compositor advanced. If we could get it to the point where we could have as much control over animations as we do with choreographies, then we could get our work looking exactly the way we want it to, but I remembered what martin had said a while back:

Work Halted on the Compositor

I read through many of his posts, and he really tried to sell the compositor; most of the time it resulted in the other party using a different application to do the same thing. I don't even have to imagine how frustrating that would be because Martin usually says what's on his mind.

But we can already produce really top notch stuff!
Its all a bit too technical for the average artist though.

The real secret to compositing in A:M and the key to its future is to use what we already have.

CG artists think nothing of spending enormous amounts of time setting up lighting, creating textures and rendering.
Yet when it comes to compositing... few seem prepared to go there.

Making a simpler version of composite would have its benefits and its drawbacks. Non technical Artists would be able to change simple things like the brightness or color of a light quite easily over the animation time line; however, someone who knows a lot more about the compositor might want to adjust all of the different layers associated with that light such as specular, reflections, diffuse and shading, and shadows, but wouldn't be able to with a simplified version. Though making the process easier to see would be the first step. It's sometimes hard to see what's going on with the compositing area takes up a very small area of the screen.

In a way we can already do this now (well... I can't... I recently broke my v14 CD!)
There are some pretty amazing things we can do in A:M with compositing if we think 'inside' the box. Enough with that thinking outside the box stuff.

Some of this can be accomplished in earlier versions of A:M too but only with v14 and v15 are we getting access to some important animation related elements of our images.
I can only try to imagine what I could accomplish if I had your talent for testing things out Matt.

One way to test the options is to create a simple model consisting of one patch and start experimenting with it. Create an Action. Throw it into a Choreography. Multilayer it with other images, adjust the ambiance... the specularity... anything you can.
Some amazing tools will present themselves.

While its the simplest and currently recommended way... Rotoscopes are not needed for compositing. Neither is A:M's compositor. (Post Effects require them or a camera I think)

Well I'll be damned! I had no idea this decision had been made. AM Composite became part of my work-flow a few months after it became available in a production version of AM and after I learned what it could do. Lately all my time has gone toward rigging and modeling (and will for another 3 months or so) so I have not been working with anything that would involve composite.

I wonder if people really understand what adding the use of Composite buys them? What's the cost of using it verses the cost of not using it. And I wonder if its introduction wasn't a bit too lack luster and the information on how to use it, presented in a way that said too little or nothing about it's advantages and use and, talked too much about what can be done "if you want to" in a complex multi-light project (like jump right into the spaghetti looking at 50 light buffers and here you go... do this, do that, enjoy... why? Search around, you'll find the reasons. Yeah right.). For instance, you don't have to do anything with Composite if you render to exp -- if it looks okay render to what you render to now. But, if something needs tweaking then 'you can' correct things in an easier manner. Initial tuts might have been better starting with something far simpler (like one light and a cube) and more instructive then the Toy scene. There was very little 'selling' of the feature to the users and in this case I think there needed to be... but... why? Did anyone 'ask' for composite? (I don't know?) I can't fault Hash's decision... why should they bust their tails supporting a feature that users don't want and don't use... and why would they want to 'sell' a feature to the users with all the features users 'are' asking for? I have no answers. I'm sorry to see it go though.

Well what to do... off the top of my head... I guess I'll continue to use AM Composite as long as its there in the program. Is there any harm in doing that? Off hand, don't see any. I'll just have to understand that it is 'as is' (if something isn't there or doesn't work that's just the way it is) and that it might go away someday.

Cheers,
Rusty

PS: Yes I can find tons of material on Composition on the web... I was looking for recommendations.

Yeah- I feel bad...it's been on my list of 'features to try' for quite some while now...I hope I can still find it in V15 when I finally go looking for it.

Its confusing even what it does (it's primary purpose) If it's plainly a 'compositor' like the name suggests, then many of us who use Adobe After Effects (MUCH less than 2k to 100k) will be hard pressed to try it. I'VE been using AE since V1 by C.O.S.A. (1992?)and it is a BIG part of my life...I even teach a class in AE. On the OTHER hand, if A:M Composite is a way to color correct and adjust the buffers you employed on your 64 bit .exr files (up the value of a light here, keyframe the specularity value there, a little less reflectivity on that other thing... etc) all while compositing your many layers to your background...well, that's something BIG! I hope that's what it does...

I use AE (Production Bundle) and also Premier Pro extensively (love them and have 7 years experience with both as well as PS CS) but AM Composite lets you do exactly what you say as far as "adjust the buffers you employed on your 64 bit .exr files (up the value of a light here, key frame the specularity value there, a little less reflectivity on that other thing... etc)" and I don't know how to do this kind of precise buffer manipulation any other way in any other program (AE can approach this but you can't isolate a specific buffer for a light and act only on this. However -- "all while compositing your many layers to your background" I'm not so sure of. If you mean stacking alpha channel plates like Premiere does (AE too) then no, I don't see AM Composite doing this. That might be more the 'NLE' features going in to V15 but I'm guessing.

Rusty

Yes, Composite was number one on the fellows wish list.

Martin was a bit apprehensive about it (if I remember correctly) but one of the guys at soul cage posted a nice example of why we needed it
and convinced him.

So Noel spent a lot of time programming but not many people started using it. (any?) I'm sure if people had taken more of an interest in composite
they would probably still be working on it. I believe they did have plans to add more features to it.

The A:M composite story is an ironic one.

Myself, I just never felt like I had something good enough or important enough to use it on.

David

I seem to recall that the actual request was for "split rendering" which was accomplished by enabling EXR rendering to various buffers. I remember following that discussion closely and the Hash Fellows weren't asking that a new compositing app be developed. They just wanted A:M to output something that could take full advantage of existing compositing apps like AE.

I found a way to get the alphas for the background. It requires that AO be turned on, but does not require that occlusion be set so it will render quickly (this layer can be removed during composites). All that must be done is to use the over post effect on the occlusion layer and the alphacolor layer and then mix the over with the occlusion layer with an input value of -100% (to remove the white occlusion).

Click to view attachment

So using this technique, anything can be added to the background.

-----
It appears that shadows are now automatically rendered for lights in V15.

Click to view attachment



Here's the entire EXR file if anyone wants it. The file's actual size is 79 MB, so you'll have to extract it.

Click to view attachment

Click to view attachment

Didn't they get this in AM composite? or are you saying they didn't need the composite just the open exr format to render too?

A:M composite is a way of looking at and adjusting EXR images rendered by A:M in A:M. But an EXR aware app like AfterEffects could do that too. A:M doesn't need A:M composite to render to EXR.

Correct. The people who wanted EXR rendering already had compositing apps. They didnt expect Hash to make another one.

I was surprised when A:M composite was announced bcause it seemed like a big expansion beyond what was requested.

I hope it doesn't go away - I used it for an image contest entry, and would never have been able to get the lighting right without it.

When TWO is over, and I can get back to playing with models / rigging / lighting / materials I am sure that I will make further use of it. Reflections - I would love to play with tweaking those.

As we know well, A:M has always been driven toward 'all you need to make a movie on your kitchen table'. It would only make good sense for A:M to be able to take advantage of the renders that spill out of it. Can you imagine the surprise and backlash if the opposite were true and A:M could render the EXR's out but not take advantage of it... or even use them?
How happy would A:M users be to find out they had to buy Preimere or After Effects to use those images. Yikes.

A:M Composite was but one step toward a solution that would have eventually included split rendering. Lack of support and interest from those that initially called for the feature effectively killed it. As David said, "The A:M composite story is an ironic one."

You can't have one (the renderered/filtered images) without the other (the viewer/editor).

Matt,
That looks impressive.
Your method is better than mine.
I'm afraid to admit to it but my piddly experiments continue.
In it I attempt to set up a project for split rendering by using Named Groups.
In assigning a specific name to those groups they can be automated, manipulated and removed from any scene or animation.

For my part its more of a personal exploration into the process than anything. I don't have the technical knowhow to even begin to know what I'm doing. That makes me more appreciative of those that do.

A:M already has Split Rendering. You render to EXR, you get a separate buffer for each light and each aspect of each light.

You do not need A:M Composite for that to happen.



I'll say it again, the Hash Fellows didn't ask for a compositing app. Saying they did doesn't make it true.

If you can show me that A:M composite was on the Fellows feature request list, I'll stand corrected.

I'll remind you also that the Hash fellows were asked by Martin to come up with a list of 10 features, of which Hash would choose three. Hash could have picked any they wanted, they chose to do Split Rendering. They must have thought it had strong merit or they wouldn't have done it. If they wanted to add A:M composite, that was their decision. But to claim it was a bad idea from the Hash Fellows... the politest thing I can say about that is to say that it's revisionist.




Yes you can. Split rendering is just an enhancement for what many people were doing already... rendering parts in A:M and assembling using them in fully developed compositing apps. Just like people have been rendering footage in A:M and editing it together in fully developed NLE apps. For years now. It isn't expected that A:M be a complete beginning to end path for everything you could possibly do with it.

Of course I should have fully completed the sentence and said "A:M Composite was but one step toward a solution that would have eventually included split rendering of objects." just to be clear. Split rendering of Objects and split rendering of Lighting being different aspects of the main feature.

If we were to split hairs, as I recall the Fellows interest in the feature was in split rendering of -everything-. It was going to take a little time to get there. But... A:M Composite was an early step in that direction.

Compositing has been one of A:M's most powerful features since the introduction of Alpha Channels. That should be close to 20 years now.

Keep on talking about compositing Rusty.
There is a lot of ground to cover.

Sounds like there are different definitions being assigned to the term 'Split Rendering'.

'Split Rendering' definition 1 (from this thread above) The ability to render buffers which will allow post-render tweeking of lighting and such(using the .exr file format THIS HAS BIG POTENTIAL and is most likely the FUTURE of 3D compositing)...

'Split Rendering' definition 2 (from this thread above) The ability to render objects separately.

'Split Rendering' definition 3 (from another current thread) The ability to render separate passes for specularity, reflection, matte, flat fill, doughboy, shadows etc. so that all can be built/tweeked in a compositing application. (Maya does this by rendering each to a separate image sequence) Photorealism can be achieved thru this manner in an app like AE...you can take an element like the specularity pass, apply an 'Unsharp Mask' filter to it...throw a smart blur onto your reflections...adjust the 'Levels' on your shadows... there are dozens of extremely usefull filters that come with AE, and hundreds of 3rd party filters. Here is a link to examples of work made in this manner: http://www.gracewild.com/divisionx/index.html

If def 1 and 3 are the same...GREAT!

Sounds like the 'Fellows' were requesting approach #3 or similar... and Hash's programmers researched it and found many benefits to using .exr and went that way with it...as well as making an application (A:M Composite) to handle the resulting .exr files (Hash doesn't have much 'swing' at Adobe) the way they needed to. The entire industry is slowly coming around to the exr...a new compositor called 'Nuke' is making waves in this approach, and may be the next 'Big Thing' like the 'Flame' is currently (Big ticket item, possibly what Martin was referring to...$$$)

Exactly what I thought...
Both things are very helpful... Buffers for the lights etc. are, but I think we need #3 too...
Some of the bigger competitors can do the #3 way already. So if Hash can do both, they are in advantage.
I think it isnt even very problematic to programm this because A:M already does everything needed but putting the images out separatly.

As far as I know, an image is represented by a matrix which can be manipulated till it fits what you need. So all A:M have to do is to manipulated at first the original matrix and than save the difference between the original matrix and the manipulated one to a new matrix which represents the image-file (for example spec-pass).

I am quite sure there is more to it, but I think this is the approach behind it, right?

*Fuchur*

The different pieces could have been done by simply adding the EXR format to A:M, but that would have taken far longer to do it in one fell swoop, and the management for an animation would have gotten very hairy.

A:M already puts out the different layers. Each light gets three different components: diffuse, specular, and shadow. The entire image gets several other layers like alpha, ambiance, occlusion, diffuse, and reflections; it is important to note that reflections are rendered for the entire scene, so if one light is adjusted, it will not affect the reflections.

Hi all! I must admit that I've been always very intrigated by this feature but for some reason it seemed too me like a very complicated thing and the learning always was postponed for one or other reason...

Well, at least until you have roused my curiosity again, and searching for more info I've found --> THIS <-- incredible thread that have really caught me for all the night... With it, now not only I start to understand what that Composite feature exactly means but I'm sure it can be INCREDIBLE useful in production workflow terms working inside A:M itself, so I'm right now completely fascinated with this incredible find and, in the other hand, completely sorrow cause the fatal fate that seems to looms over it...

But well... anyway I only wanted to share all that stuff just in case you don't know it existed, and recommend to take a look cause I'm sure you won't regret it, even if it takes all your night too , good night alll...

Thanks Ramón López

I pdf'ed the first three pages if someone wants them

Click to view attachment

Click to view attachment

Click to view attachment

Yup.

Don't go into this with an eye on any advancement to the feature.
Its already been established that this won't happen.

Go into it with efforts focused on what we can already do.
Success and advancement will follow the effort.

Thanks for the collected tutorials John.
Very useful!

Well, I think at the very LEAST, this has been some healthy discussion. We've established a few facts- among them:

-This is a very useful feature.
-OUR lack of enthusiasm and support for it has prompted Martin to altogether drop further development.
-This is a very complicated feature, and hard to describe, document or showcase.
-The feature is still available for those who dare try it in it's (working?) early condition in V14 and V15. I know I will.
-Until a standard is set amid the very small usergroup of the .exr file format, the possibility of working with A:M rendered exr files in a 3rd party app like AE, NUKE, or whathaveyou is...for now...a 'pipedream'.

Earlier today, as I researched the 'Maya' way of exporting separate passes (in the 3rd definition of the 'split rendering'-see above) I called one of my fellow animators who happens to use Maya and can easily get AMAZING photorealism out of it... He said that rendering separate passes for spec, ref, shadows, etc was... in his words... the 'secret sauce' to photorealism. He also acknowledged that the .exr file format held untold advantages and that as far as he knew Maya had no plans to persue this path.

An excellent book to get on the subject of Lighting is Jeremy Brin's "Digital Lighting & Rendering". He also has a section on compositing where he has this to say related to our current discussion.

(He then goes on to list some of them. See: http://www.3drender.com/light/compositing/index.html)

Brin's book is one of the many books I brought to the desert with me.
I brought more books than clothes.
You usually find clothes even in a desert but good books on animation... a bit harder to find.

Of note here is that of the two (rendering in layers and rendering in passes) my primary interest has long been rendering in layers. I understand the importance of the data contained in the filtered passes though and it is considerable but having the capability to control what is in a scene separately is more important to me. Once the object is isolated the details of that object (or even the object itself) can then be changed.
If the objects aren't isolated then you are stuck and have to remove them the hard way; manually. Still images aren't so bad... but animation.... yikes!

There are many interesting sources of information on the subject of compositing on the internet. Here's one I don't remember seeing: http://www.mathiaspedersen.com/tutorials_starry_night_2.html

Note: The program used is different but the techniques fairly universal.
In the tutorial the end product is a still image... not animation.


Edit: I see some fairly large sections of Jeremy Birn's book is online. Cool.
Post updated to reflect some info you'll find via the link.
http://www.3drender.com/light/compositing/index.html

I've worn out the text in Brin's book but that was several years ago and I see that a 2nd edition is out. I'll have to look at the new version and see if its worth getting. The book (1st or 2nd ed) is a must have. I also have Gallardo's book "3D Lighting" -- it was the required book for a class I took. Brin's book is better in that it provides more info you can use right now! But, IMHO, Gallardo's text has some awesome info that Brin's does not like some exact steps to creating different types of lights and environments.



Well I'll start referring to these techniques in this way. You can also look at it as vertical and horizontal (vertical stacking of alpha layers and horizontal blending of color channels). In any event I aways 'render in layers' -- its second nature to me. I even have an EXCEL sheet for planning the layers. But, for myself, before the AM Composite feature delivered separate light buffers(!), 'Rendering in passes' amounted to specific rendering options (there's a nice plug-in for this) for each render pass and then re-combining and tweaking in AE using various blending modes and filters. Beyond that you toyed with the color balances, channels filters, levels and curves. Nothing I have, at least to my knowledge breaks out these buffers from the exr files... I have AE V6 (a couple of versions behind) and it can't do anything with exr files. Perhaps they are supported in later versions.

Some of the most valuable techniques and ideas I've picked up have been from "The Making Of" or 'special features' sections of DVD movies. Lord of the Rings (all 3) blew my mind, I watched it twice taking notes the second time. I used a bunch of this stuff in the Horrorthon and Mantis projects (and I think I 'layered' every scene in both of these).

Keep on talking about compositing Rusty.
There is a lot of ground to cover.

Cheers,
Rusty

David Rogers was one of the biggest advocates for split renders and compositing, but what happened to him? According to his profile, he hasn't posted in 18 months.