Mythbusting (Part Two)

NAB roundtableIn Mythbusting (Part One) I made the case that it is the workflow, not the results, which prevents After Effects from being a more common choice for big budget feature film visual effects work. This topic is firmly on my mind at the moment having just completed work on a feature entirely using software other than After Effects (Shake), and I’m starting work a new edition of my book, all about doing this type of work in After Effects.

An in-depth comparison of After Effects and Shake seems a little pointless, given that Apple seems to have dropped Shake entirely – and I do mean, entirely. Last June we found out that the product known as “shake” (not officially capitalized – I’m just doing that for clarity) was officially End-Of-Life, that 4.1 would be the last version, and that it was even possible to license the source code for $50,000 – remember that fire sale?

With NAB 2007, the other shoe seems to have dropped. There was no mention of a Shake successor whatsoever, but heavy promotion of Motion 3 (which, while looking fast and cool, has until now – to me, at least – been more about eye candy than the rigors of creating a realistic looking shot). A visit to the Foundry booth revealed a few new faces – veterans of Nothing Real and the Apple Shake team. Ron Brinkman announced he was leaving Apple. Other rumors held that key Shake developers still at Apple, those essential to any succeeding version, had been pulled over to the Aperture team.

I remarked on a roundtable panel at NAB that the current situation for effects compositing software is reminiscent of where 3D software was at in the late 1990’s – remarkably up-for-grabs, with no clear current leader. All signs currently seem to point toward Nuke moving into the pole position, but how quickly and universally this happens depends on the timing, pricing and features of version 5, the first one which will show where Foundry wants to take it. If the UI is massaged to be nicer looking and friendlier, the license costs fall more in line with the competition, and version 5 appears before the end of the year and proves as stable and fast as its predecessors, watch out.

And where does this leave After Effects? This is a bit of a half-full, half-empty type of question. Motion is trying to provide all of the motion graphics features of After Effects but with a simpler, more real-time experience. Nuke is focused on providing all of the next-generation visual effects features, such as camera projection and full support for EXR files, which are unavailable in an application like Motion and available but less-than-straightforward to use in After Effects.

All of which brings us back to the topic of this little “Mythbusting” series – where is After Effects at a true disadvantage relative to the competition? I could name a long list of items, but one could do that for any piece of software. Fundamentally, there are two longstanding disadvantages for After Effects which continue to loom large:

1) UI responsiveness
2) Pre-composing

Responsiveness: Nuke is a super-responsive and highly interruptable application, even compared with Shake. With After Effects, one can end up waiting several seconds (or longer) just for a frame to stop redrawing (even if one enables caps lock or the option/alt key explicitly for this purpose). Whether or not this actually leads a shot (or a whole show) to take longer to complete in After Effects, the point is that the application feels so much slower is many cases that artists at all levels – beginners and experience pros alike – become biased against its overall performance. In 2007 it should never be difficult to adjust a simple slider control because the machine is busy doing something else.


Pre-composing
has long been the bane of After Effects; it seems to be the price we pay to work in the Timeline. With each new version, features are added to ease the pain; CS3 offers a prominent new button in the Timeline which allows you to see and select other compositions in which the current composition is used. Nevertheless, pre-composing is a pain because it adds steps and obscures vital data.


Oddly enough, Motion 3 (on the motion graphics side) and Nuke (on the effects side) both aim squarely at these weaknesses in After Effects. I have only seen the Motion demo – it isn’t even publicly available at this writing – but its features seem less obscure and its image processing faster (thanks to heavy use of the GPU) than After Effects. Whether that comes at the price of sophistication and control remains to be seen, but I doubt that Motion would ever be used to complete an effects shot on a feature. Nuke, meanwhile, is also astoundingly responsive with huge shots, and it has the familiar advantage of any node-based system: no pre-composing.


We’ll know more later this year about how all of this shakes out. Meanwhile, it seems pretty clear that in the long run, shake’s out.

29 April 2007 | after effects, punditry | Comments

6 Responses to “Mythbusting (Part Two)”

  1. 1 Marijn Eken 30 April 2007 @ 3:03 am

    Wow… Your comments on After Effects responsiveness and pre-composing have always been something I’ve had trouble dealing with. Especially the pre-composing drives me nuts at times. For example: you have a compositing with 3d layers and a camera. For some reason you need to pre-compose a 3d layer, but other 3d layers will stay in the composition. You would have to duplicate the camera and pre-compose that along with the layer to retain correct motion!!! Ouch…
    I’m currently enjoying the pains of After Effects renders. How come a that I can look at a frame in the timeline, but when I start a render starting at that frame, it tells me: Couldn’t create 4234 x 2490 (or something similar) image buffer!?!?!?! What the @#!^%?
    I somehow think that Nuke would never do that to me… I’ve been looking for the next step in my compositing and I think Nuke might just be it. I don’t hear a lot about Fusion 5, although at first glance it seems like a wonderful program…

  2. 2 Alex C 30 April 2007 @ 10:50 am

    Hey Mark

    This website has some good insight on this question :
    http://www.digitalgypsy.com/vfxlog/archives/2005/10/after_effects.php

    And a comparison of the main packages in use in the VFX industry

    http://www.digitalgypsy.com/vfxlog/archives/2006/07/nuke_shake_flam.php

    As always, your insights are a great read

    Alex

  3. 3 Josh Ngwee 2 May 2007 @ 11:18 am

    Enjoyed your studio techniques book and looking forward to the book you mentioned in the post – about time someone put out an AE book on real world style compositing examples rather than the endless AE intro type books around.

  4. 4 Jonas Hummelstrand 8 May 2007 @ 11:22 pm

    I’m sorry to see that while many of the GUI responsiveness bugs in AE 7.0 are now fixed in CS3, the GUI still feels unresponsive.

    I’ve always accepted the “truth” that 7.0’s GUI is slower than the 6.5 GUI, but I decided to do some tests. I installed 6.5 Pro, 7.0.1.11 Pro and CS3 on my Mac Pro under Windows XP SP2 with the Bootcamp 1.1.2 beta drivers.

    To my amazement 6.5 is actually slower to navigate around he TLW, but it _feels_ snappier than 7.0 and CS3! I can only attribute this to two factors:
    Factor A: 6.5 is better about interupting previews to give back the control to the user, where the newer versions often disregards interuptions, or just stops at the first click/key press, forcing the user to click/press key again.
    Factor B: The scroll wheel on a standard mouse is almost four times as effective in 6.5 than in later versions, making them seem slow when they really are faster GUI-wise.

    Let’s hope for a soon to be released CS3.01! :-)

  5. 5 Tony Agliata 30 July 2007 @ 11:55 am

    Mark, thanks for taking the time to shed some light here. Makes me wonder how much more of my time should be spent in Shake. Maybe a look at Nuke is in order. Although my first love is AE.

    Digging the classes at fxphd.

    Tony

  6. 6 mc 31 July 2007 @ 9:26 am

    One thing I didn’t mention here is that many large studios took Apple up on the offer to buy the Shake source code when the product was EOL’d – thus development of it CAN continue, amazingly. It’s hard to picture a world in, say, five years where a dozen or more post-4.1 versions of shake have continued to be developed at places like ILM, and yet it’s not out of the question if Nuke somehow doesn’t turn out to be the holy grail that everyone is hoping it is/can be.

google

google

asus