[Steve Bentley]

K is right, but that looks more like Processing to me. Its the “history” of the objects that’s going to kill you. They never really go away and just fade. In Processing this would be just a pixel array having an algorithm run on it, so the memory required would be the same from start to finish (the number of pixels don’t change). Then there’s the (possibly intentional) “poor” quality of the video – a classic Processing “artifact”.

I’m not sure Plexus could keep everything connected forever. I find it jumps around due to inconsistent “dot” counts and proximity.

[Steve Bentley]

Snake and dragon boat in blue are definitely 3D – you can see an aspect change on the head of the dragon boat and the snake just… well, has volume, that we get to see different sides of.
The sword slash to the neck (brown section right after the boat) is also 3D as there is an aspect change on the hand and we get to see both sides of the hand – you also get to see more of his eye area inside the helmet. While some of this could easily be 2D with Duik or similar (or more likely with their own software – it is a software company after all) its probably a mix of both. I would agree that doing some of that “sketchy look” stuff in 3D would be simpler in 2D due to intentionally ratty edges but some of the forced perspective stuff on limbs is easier in 3D – doable in 2D? For sure! But why bother if you have the assets – which they probably do from the game.

The background people have all the hallmarks in the animation style of 2d puppet tool stuff.

Many of these companies use custom software that can combine 2D and 3D so you get the best of both worlds without the render hit of doing 2D in a 3D package.

AE is 2D only. Yes you can fake 3D with angled 3D planes and you can set these flat planes in what feels like 3D, and you can use something like Element 3D to get true 3D, but as amazing as Element is, doing true 3D is easier in a package dedicated for it. Heck, even Particular isn’t truly 3D (at least by the time the objects show up in the comp window)
That’s not to say you shouldn’t try in AE. We just finished one in fact.
You can also extend Ae with C4D – the combo of Ae and C4d now is pretty cool. You can bring one into the other to get the benefits of both and keep everything “live”.

There is no one way of doing things. While its cool to be able to say “that’s actually all 2D” and have people be amazed for a day or two, there will be someone else who produces something that looks just as good, but did it in half the time for less money and has the rest of the day off because he did it in either 3D or a combo of 2D and 3D.

There’s a plug in for AE now for offset layers to give the illusion of depth – stacking – you can even do faces turning – I forget what its called. I don’t think the shield is a that, I think its 3D. But you could do it with a 3D plane and a curved top matte on a separate layer. The plane rotates into view and the curved matte tracks with the arm to keep the correct curve for the shield face. You can also warp surfaces as they turn to reduce what looks like an edge profile.

If it makes sense to use one or the other go for it. But in the end, the client doesn’t care how it was done. Most of this is a mix with the 3D stuff having textures that are hand-drawn so it looks 2Dish. Or it could be all 3D to keep things simple in one package, using some 3D objects and some just as flats.

[Steve Bentley]

The first half probably has some 2D comping in it (like AE) but that’s pretty much all true 3D. And it is a game company so they will probably have used the 3D assets and software from the game.
The second half though… there’s your parallax. Could easily all be AE there. 2D illustrations just being dragged about at different depths for parallax.

[Steve Bentley]

Hang on, is it the AE file size? or the size of the movie that AE is spitting out that you’re wondering about?

Final movie size is pretty easy to figure out.
NumberOfFrames * FrameXPixels * FrameYpixels * 3RGBChannels (or 4 if there’s an alpha) * 8bits per channel gives you total number of bits in the file. Then divide by 8bits in a byte to get bytes (then divide by 1000 to k’s and then by a thou again to get MB etc). So really, the 8’s cancel each other out and you can stop the math after you multiply by 3 (but I thought I’d show my work).
Now this only works with an 8 bit RGB uncompressed file. (use 10s in place of the 8bits per channel when its a 10 bit file of course).
Also, with Animation best you are going to have some Run Length Encoding (RLE) which losslessly compresses some of this size down. But this only really makes a difference when you have an alpha channel or if you have lots of flat color with no noise and keyframes spread far apart. But at least the math can get you in the ball park so you know you aren’t crazy.
Keep in mind very few systems will play back a file at those pixel dimensions in AND in Animation Best AND in real time, so its probably going to have to be compressed or subdivided for the video wall processor at some point. Or perhaps the player sucks it in and compresses it as it does so.

As for the sluggishness in the render, the frame blending will certainly add a significant amount of time (do you really need that?) FB will only make previously rendered footage smoother if you are making it last longer than it did originally and usually there are better ways of stretching time than frame blending.
Large frame sizes take longer to render, and its exponential – generally a frame size twice as big will take 4 times longer to render (all of this really depends on what it is and what frame size you are doubling – its all about what fits in RAM).
More layers take longer to render.
More frames take longer to render and this can be exponential too depending on what’s got to be held in ram from beginning to end of timeline. So 10 frames might render in 2 minutes but the next ten frames might take 20 minutes as the RAM gets used up. It also depends on what’s happening in those frames. AE’s render time guess doesn’t know what it’s going to have to do 20 frames later. The guess is based on what it’s done so far and AE assumes all the frames to come will be like the one’s its seen already.
Certain effects take much longer to render than others.
Certain source elements take longer to calculate than others (EXR’s with motion blur and depth channels take much longer than a motion blur put on a jpg sequence)
Higher bit depth comps take longer than lower bit depth comps to render.
Not having gobs of RAM will take longer than having more RAM (the more you can keep in RAM the faster it will render, so the more RAM you have the more you can keep there) All else gets shunted to the disk cache which is usually slower especially with all the swapping going on.

As for the AE file size, number of layers or comp length doesn’t really increase the file size that much – its the keyframes and number of files in the project panel that really add to file size. Size of comp really has nothing to do with AE file size.

When you get into minute long (or longer) AE projects, consider doing some of the work in an NLE. Longer elements usually rip out of NLE’s faster (but they are of course simpler). Having a bunch of prerendered “element” movies stacked inside AE usually runs faster than doing all the work in AE – at least at those lengths.

[Steve Bentley]

In my experience it’s the bigger comp sizes (4k plus) that really need all that RAM. To be able to hold the frame in memory and all the assets that need to be integrate (but not necessarily all the layers in the comp since some of those will get shunted to the disk cache). There are times when depending on the effect and frame size, you just can’t render on 32GB.

AE is both really good and bad at making sure no matter how much ram you have it can usually work with the disk to render – it can just take much longer when RAM is used up too quick. The bad is that with lots of ram you have lots of ram to get clogged. The memory purge has gotten much better over the years but Adobe has never been known for their elegant memory management. Even though we have a “clear all caches” button now, I find that even if I use it, then quit AE, I still can’t run some other intensive program right after. There’s still a hairball in the drain somewhere.

Finally, more RAM means more frames can be cached and played back in real time without rendering out to a file.

But it is a case of diminishing returns. Doubling the ram (after 16gb) doesn’t make AE twice as marvelous, and the difference gets harder to detect the closer you get to 256GB. That being said, more RAM means more left over for something else running at the same time – say switching back and forth to C4D to a project that’s linked and you’re tweaking both AE and C4D.

[Steve Bentley]

Hey Max.
Here’s how I edit. (I think there is another way as well)
Hit reply to your own post, then when the text entry area comes up, instead of tying, scroll up and just below the last item in the thread list (your last post) will be an edit button. That will let you edit your last post. Correct and then hit submit.
Sadly, far too many posts are completed before coffee is completed.

[Steve Bentley]

Speaking of region of interest, that too used to be a good trick to get real time playback out of a section of a comp that was too big to do so when viewing the whole thing. But as of AE2017, I”ve noticed it doesn’t seem to make a difference. And if you monitor ram usage, it’s nearly identical, whether you are using ROI or not. Did something change?
I’ve experienced little difference in both, comps that have lots of little elements all over the place (only the elements within the ROI should be heading into ram) and in comps that have large comp-spanning elements that will get loaded into ram no matter what, since at least part of them is always in the ROI.

I’ve also always wanted to know the difference in the memory management between AE and platforms like Nuke, Fusion and Flame/Inferno, as these are able to play much larger uncompressed comps back in real time. Back when, I always assumed that it was the SGI magic pixie dust that made this possible, but now, everything is on PC’s so that playing field is leveled. And if there is a big difference in software only why hasn’t AE made better strides in this area?
Even back in the I’m-too-embarrassed-to-say-how-long-ago, Jaleo was playing 6 streams of uncompressed HD (6x186mB/s) overlaid and composited in real time (no rendering required) while AE was still struggling to play/cache 1 stream of SD (without special hardware) – about 30mB/s. (to be fair that was partially an SGI aided accomplishment, but still, the difference is staggering)

[Steve Bentley]

I am not seeing that here but it could be due to your settings in AE or AI. What you can do though is go into the interpretation settings of the file you have just imported (File/interpret footage) , go to the color management tab and force it to the setting you want. Then tell AE to remember that interpretation which it will do for subsequent imports (File/Interpret Footage/Remember Interpretation).
You can also do this for the kind of alpha channel you want and frame rate. (it doesn’t change the number of frames available it just changes the rate at which AE will use them up – so a 2 second clip originally at 30fps will be 4 seconds long in a 15fps timeline if interpreted at 15fps – AE will normally look after this anyway when it compares clip.fps to timeline.fps, but sometimes you want a little more control)

[Steve Bentley]

You can use the crosses in the corners to drag the image around (via the PowerPin) to match image to the movement of the tablet as well as distort it. If you do both (tracking the tablet AND tracking the 4 corners) you might find the two systems fighting each other or adding onto each other.
As far as the image that goes in the screen, tracking the tablet is irrelevant – each corner is going to track differently, so tracking the tablet as a whole becomes a question of “where on the tablet do I track?” For a single tracking point, there is no good answer for that. if you were putting a dot on the tablet, a single track would work, but as you rotate the tablet in the tablet’s “Y” (or world Z) all the crosses will get closer to each other: the left ones will move right and the right ones will move left, even if the tablet center doesn’t actually move. So tracking the tablet itself doesn’t really help.

[Steve Bentley]

You can just drag the project into the message box (where you type) and the COW will look after the upload for you.