Help with: Loooong renders in AE CS3

[Richard Mamary]

UPDATE

As a test, I deleted all but the Photofilter effect, just to see how this would affect render time.
With only this effect, the render time still says 6 hours.

This also seems to me to be a bit long. Am I wrong? Is 6 hours normal for a two hour piece, with Photofilter?

Is 16 hours for all the effects normal as well?

[Richard Mamary]

Dave, thanks for the response.

I imagine your :30 spots are filled with lots of effects, 3d renderings, animations, etc., and are not straight footage with color correction.

I hear you about more gigs.

Just curious, what kind of system + specs, are you using for your renders?

[Kevin Camp]

is multiprocessing enabled? preferences>mutiprocessing, check the render mutliple frames option. then read the description bellow to see if ‘1 additional processor’ is used.

if not go to the memory settings, lower the max memory to 110% (it should still say 3.0gb) and check the mulitprocessing to see if you gained the extra processor. if not, decrease the max ram cache a little until you can get that extra processor… it shouldn’t take much with 5.5gb of ram.

if that doesn’t speed things up enough for you and you like to live dangerously, you could try enabling opengl renders (i’m only recommending this for color adjustment work, with your older mac i believe you only have an opengl 1.5 support and there are limitations with that).

first you’ll want to make sure you see the previews in opengl. choose preferences>previews and enable opengl and choose accelerate effects using opengl. now, in the preview window, click the little lighting bolt. select use opengl – always. now you can see your adjustments with opengl processing (things may look different, so make adjustments accordingly).

now you need to set ae to render with opengl. choose composition>make movie, click the render settings and set to render using opengl. then render as normal, see if that speeds things up. several of the color adjustments you were using are opengl accelerated and this may help (although, i believe the photofilter is not). for a list of accelerated effects check the help menu for opengl effects…

Kevin Camp
Designer – KCPQ, KMYQ & KRCW

[Kevin Camp]

one other thing that could help is if you are having ae separate fields just to do color work and then reinterlacing the final render, then you are having ae do 2x the work.

set ae to not separate fields (select footage in the project window then choose file>interpret footage>main) and then don’t set the render setting to render fields (they are already there, since ae did not remove them).

now if you manipulated the scale or rotation, or added a blur or anything like that, you will need to separate fields, but if all you are doing is color adjustments, then not separating fields is fine.

yet another issue could be the codec used for the media files. if it uses interframe (or temporal) compression (like mpeg and many hd codecs) then ae will take a lot longer to render. exporting that media to another codec like animation or photojpeg, or an image sequence may help speed things up quite a bit.

Kevin Camp
Designer – KCPQ, KMYQ & KRCW

[Richard Mamary]

Kevin,

Thank you so much for your well though out response. I had to run out for a bit, then when I came back, run tests on your suggestions.

Yes, multiprocessing was enabled, and described “as being used” below on the prefs page. The memory setting was even at 120% and it still showed.

I already had openGL enabled for previews, but not for renders. When I enabled it for renders, it actually slowed things down about 10%, even when PhotoFilter is not being used.
Open GL seems to work fine for previews though.

When I checked Interpret Footage, in the file menu item, it did say lower field first, I changed it to none, this did seem to get me quite a bit of speed as well.

I will remember to set it back to lower field first, when I work with sources I’ll be scaling and rotating in sequences.

As far as codec, it is DV-NTSC source footage, and I am rendering it the same. This should make the render as fast as possible, right?

What is, “storage overflow,” in the render settings, and should that be checked or not?

Does “enable disk cache,” affect speed? What settings are best?

But for the biggest speed increase, I would be better suited to just use one type of correction, perhaps Levels only, if I want speed. The 16 hour render time for a 2 hour piece is cut down to a mere 4 hours, when I use only one filter, without separating fields. This may be as good as I can do.

[Kevin Camp]

[Richard Mamary] “As far as codec, it is DV-NTSC source footage, and I am rendering it the same.”

dv doesn’t use temporal compression, so it shouldn’t slow ae down too much. technically even intraframe compression slows things down a little, but i don’t know if converting all you footage to uncompressed would be worthwhile. however, i might recommend rendering to lossless or an uncompressed format, not for speed, but for higher quality (your final render will need to get compressed to mpeg2 for dvd, so you’d be better off not compressing to dv then to mpeg-2).

[Richard Mamary] “What is, “storage overflow,””

it is what ae uses if a render is too big to fit where you told it to render too. so if your drive had 1gb of storage and your render ended up being 2gb, ae would use the designated storage overflow and split the render (hald where you told it go, and half where you set the storage overflow). it won’t help performance, it will just allow a render to finish.

[Richard Mamary] “Does “enable disk cache,” affect speed?”

yes, it will have a negative effect on performance, however it can reduce crashes that are related to memory limitations.

[Richard Mamary] “I would be better suited to just use one type of correction, perhaps Levels only, if I want speed.”

i haven’t tested it, but i would think that if you could make all the necessary adjustments within one effect, that would probably be faster than using 3 effects.

Kevin Camp
Designer – KCPQ, KMYQ & KRCW

[Richard Mamary]

Thanks again, very much.

However, i might recommend rendering to lossless or an uncompressed format, not for speed, but for higher quality (your final render will need to get compressed to mpeg2 for dvd, so you’d be better off not compressing to dv then to mpeg-2).

Yes, I am going to mpeg-2, which unfortunately AE does not support. I was hoping it would be included in CS3.

I often use the animation codec, (though the file is huge), for quality purposes when rendering animations.

But, I have also always thought that if the source is already DV-NTSC, then for AE to render it such, it would not require further compression. Have I been wrong about this all along?

Would I maintain more quality by rendering the original compressed DV-NTSC source as Animation, before compressing to mpeg-2?

Should I be capturing this stuff, (mostly analog 8mm tapes), as uncompressed, or DVCPro to begin with? Would it make that much of a difference?

And yes, I will only use one filter from here on!

[Darby Edelen]

[Richard Mamary] “Yes, I am going to mpeg-2, which unfortunately AE does not support. I was hoping it would be included in CS3.”

AE has had an MPEG-2 output module since version 7 on Windows, CS3 introduced an MPEG-2 output module for the Mac. The compression is probably not as good as a stand alone product, so you’d probably be better off rendering to something lossless/uncompressed and then using a quality MPEG-2 encoder to encode.

[Richard Mamary] “But, I have also always thought that if the source is already DV-NTSC, then for AE to render it such, it would not require further compression. Have I been wrong about this all along? “

Pretty much, AE needs to render everything it receives in order to output anything. So even if what you use the same codec for your input and output AE needs to decode what you gave it and then re-encode it. The choice of codec can be significant, however, because some codecs take longer to decode (usually anything that uses interframe compression, MPEG-2, H.264, etc. because AE needs to look at previous frames in order to build the current one), and some take longer to encode.

[Richard Mamary] “Would I maintain more quality by rendering the original compressed DV-NTSC source as Animation, before compressing to mpeg-2? “

Yes. When you render from DV to DV you can end up increasing compression artifacts.

[Richard Mamary] “Should I be capturing this stuff, (mostly analog 8mm tapes), as uncompressed, or DVCPro to begin with? Would it make that much of a difference? “

It probably wouldn’t make a huge difference from that source, it would eat up your hard drive space much faster as well!

Darby Edelen
Designer
Left Coast Digital
Santa Cruz, CA

[Kevin Camp]

i’m not sure, exactly how much of an improvement it would be if you were to capture old analog vhs/beta/hi8 to an uncompressed codec, color correct and render back to an uncompressed/lossless codec then compress for dvd.

it wouldn’t be much more trouble, but it would require a lot more drive space, and the quality may not be much better…

it wouldn’t be too hard to test if you had the time.

the highest quality workflow would probably be something like this:

1. capture/digitize to uncompressed format
2. take to ae for color corrections and cleanup
3. render back to uncompressed
4. take render to a compression utility like sorenson squeeze of compressor for mpeg-2 encoding*
5. take that to dvd authoring software

*generally speaking, using a compression utility will produce a better mpeg2.

that would give you about the cleanest transfer to dvd. the only real issue would be the connection from the analog source and the capture/digitizing device.

again, it’s hard for me to say how much better uncompressed will be over dv for old vhs media…. but generally the less compression and recompression the better.

Kevin Camp
Designer – KCPQ, KMYQ & KRCW

[Richard Mamary]

I just tried a test render from the DV-NTSC -> Animation, and it indicated that it would be a ten hour render, instead of 4 hours for the NTSC->NTSC render.
As I have hours and hours of footage, I can’t afford that much time for the potential quality.

Thanks for your input Darby, I’m learning a lot. I just hope my old brain can hold on to it!