[Steve Bentley]

Hi Tero,
So are you saying that the interface is different in QT with fewer controls, OR, are you saying that Apple is using a non standard codec?
And I’m talking QT Pro here (with the latest version and dot upgrade to the codec files) just so we’re apples to apples (Ha!) And I’m doing this on a PC in case this matters, with some sort of porting issue of the original code from the original Mac platform, from before PC could speak QT.

I will grant that the controls in QTpro are missing a few items that might make things better, but (at least in my experience) Adobe Encoder’s output (size/quailty) has never been able to touch QTpro’s version of the same file. And I find QT PRo faster. With one caveat:
After you render in AE there seems to be a memory management issue (I know, shocking right? it is adobe after all) that prevents QT from running at full tilt when it’s compressing. If we have a big file to crunch (and we need to use QT pro to do it) we will restart the machine to clear the baffles after AE has done it’s thing. Only then will it compress at a reasonable speed. The quality will always be the same it’s just the time it takes to do the job that changes.

If indeed there can be differences, I wonder if different encoders (not codecs) have a sweet spot for bandwidth. Most of the stuff we do gets pushed out at 3,500kb/s to 29,000 kb/s. So maybe I’m not seeing any QT sins that are happening down in the sub 1000kbs range, or conversely a stellar job Adobe Encoder might be doing in the same low kb/s range.
But given the algorithms needed to get to H264 (that shouldn’t be changing between encoders) I would have thought using the same settings would garner the same results, regardless of platform or encoding engine – with the exception of the time it takes to do the job.

[Steve Bentley]

Good point Eric.,
You can just check the Link panel in Ai and if there are any entries in it that means there are linked raster files that will gum up the works when you bring the file into AI (as far as “continuously rasterize” that is).

I should point out one other thing – and perhaps this is what Remus was referring to.
When you get an overly complex true vector illustration – like one of those stunning portraits that look like a photograph, made of a gazillion vector paths and grads – often it is better to make a raster image out of that in AI or Pshop and bring in pixels to AE because AE will be hard pressed to rasterize all of those vectors and may even run out memory.

But again the problem with Shutterstock vectors is they often just way way way too small on the art board.

[Steve Bentley]

Output from AE in an uncompressed format first – then encode in another peice of software. I can’t say I can recommend Jpg as Remus suggested because JPG is a compressed format and when you go to compress to H264 you are already a step down on the beginning quality.

[Steve Bentley]

??????

H.264 is a codec and doesn’t care whether its a .mov or a an mp4. Qucktime takes longer to process H264s than other programs (I find Adobe Encoder slower than QT) but neither are as fast as FFmpg. And all of these will be faster than AE. Many Mp4s out there are just .mov’s in an mp4 wrapper (as are FLV’s)

The output quality has to do with the settings you choose in your compression software not the compressor you used or the format it ends up in. H264 is a standard, not some variable thing that is reliant on what program made it.
Remus has a habit of saying “stay away from this or that” without having in-depth knowledge of the thing he is warning you about. If you are getting poor quality compresses, it just means your settings in your encoding software need to be tweaked. Its also possible you are asking too much of the codec. A most mostly black movie will be very small in mb, but a mostly noisy movie (lots of action and color) will never be super small if you want a clean, small mb movie in H264.

The issue is that AE can’t wind up through the content and look at all the frames before it begins encoding (Ok it can, but you are basically rendering twice). Any good encoding software can do a dual pass and analyze the content to get the best out of it.

.Mov is a perfectly valid clean format to work with and if you are on the Mac, one of the best choices if you want to keep file sizes reasonable. .Mov is just a receptacle for codecs. I can make a .mov in BlackMagic or Animation/best or Cineon – it’s still a .mov.

But your pipeline should be be: render out of AE in an uncompressed or Lossless format, and then put that final file through and encoder or QT or FFmpg or MS Expression or whatever to compress it.

Besides, what if there is an error in your movie that you have to correct in AE (wrong color, popping keyframe etc) do you really want to waste time encoding before you have a chance to inspect it?

[Steve Bentley]

Ok, that’s just plain wrong.
I feel badly for you Remus, because under that false assumption you are missing out on some of AE’s best features.
Masks, AI files, some of the plug ins, paint strokes are all vector based. While the final output is raster, internally it can most certainly take advantage and cleanly scale and rotate vector based content.
There are limits of course but then there are limits to spline resolution even in AI and a high quality RIP.

Actually I’ll correct that: you can output vector when you export as a Flash file.

You are never better off importing a really big PNG – imagine a circle made of 1000 x1000 pixels – now scale that down to 10×10 pixels. It’s not going to be a circle any more. It will get soft, and square or stepped. But if you use a vector circle scale that, it will get subpixel sampled when it gets down to a small size and still be as close to a circle as the resolution of your screen will allow.

[Steve Bentley]

No I was just using the wrong words. You dont’ have all the fade settings in the visibility tab? There’s even a special pair for z depth.

[Steve Bentley]

More often than not the shutterstock eps files have to be massaged in AI to get them to perform correctly in AE. IN addition many of the shutterstock files are insanely small on the pasteboard in AI. So first thing scale them up to a reasonable size in AI – make sure to include strokes in your scale (use the scale tool and don’t just drag the bounding box). If they are too small the “continuously rasterize” button can’t do it’s job.

[Steve Bentley]

In the version I have I can turn off min clip depth and max clip depth. But it might also have to do with your gfx card. I know when I’ve made my environment very large it exceeds the memory space of the card and odd things start to happen. But you can turn that off (open gl) in the AE prefs and it fixes the issue.

I also didn’t know what you meant by “looking outside the camera”. What version of Ae and Particular are you running? It sounds like a very old version of trapcode where there was a particular camera that wasn’t the world camera.

And with a size of that magnitude you might have reached memory or decimal place limit. Why so large?

[Steve Bentley]

I think what you are referring too is the multi layer scale that is going on there within the flower. With elements that are slightly transparent (or screened or multiplied or both) you can overlap many layers all scaling at slightly different rates or at the same rate but shifted in time and they can take on magical transforming qualities. There is a screen saver for android that is nebula and asteroid fields and they use the exact same technique – many stills scaled and screened over each other and it give some incredible depth. (much better than this flower).

[Steve Bentley]

Ah ok (i didn’t get an email with your response so sorry if I’m late to the party).
Under the main cracked plane we put another plane with a static dynamic tag on it, so the sphere pushes up but the static plane underneath keeps everything from falling (or you could just turn off gravity in the Project settings)