Desparate – Two problems using FCP to Compressor to DVDSP

[Jeremy Garchow]

2pass doesn’t mean better. It’s just more efficient as bits get assigned to where it’s needed most. That scene you speak of sounds particularly difficult for MPEG2 due to the GOP structure. Since one pass VBR is doing less analyzing, the file is looking better. Have you tried using compression markers? Those force compressor to take a good look at the footage (basically force more I frames). You might try that.

As far as the graphics, didn’t you say it was created at 1280×1080? What was the pixel aspect ratio? If created at square, that’s your problem. Also, once rendered to QT, it’s not in vector land anymore.

And I’ll ask again, how did you capture this footage? Please do tell. Tape or P2 based?

[David Heidelberger]

To add to what Jeremy said, I think you mentioned somewhere that it was rendered from Motion in the native format of the timeline, which would make it DVCPRO-HD, right? And even more than that, it gets heavily stretched from 1280 to the full 1080. That means that no matter how you render your timeline from then on (for example nested in a ProRes timeline), the graphics will not benefit because it’s baked in as DVPRO-HD. Can you see if you can get the titles re-rendered from Motion as ProRes 4:4:4:4, 1920×1080? That way, you’ll have them at very high quality no matter what format you render the timeline and might see better results from a ProRes timeline or using Send to Compressor. Are the titles red, by any chance? I’ve found DVCPRO-HD gets very blocky when dealing with reds.

– David

[Michael Sacci]

2 Pass VBR is not better than 1 Pass, it is more exact with the final bitrate average. 1 pass VBR can be way off from the set bitrate average. So just like the CBR in the pulsing spots the 1 pass is encoding at a higher bitrate. With 1 pass think of your average bitrate as a hopeful suggestion to Compressor. Unless you show us the problem titles we are all just guessing.

[Timothy Anderson]

Hi again,

Thank you very much to everyone first of all for the help – it is much appreciated:

Jeremy – the footage was P2 card based from Panasonic. The HD titles created in Illustrator were created in a ‘Film’ artboard at the DVCPRO HD 1080 setting. This is rectangle pixels. The pixels in the Illustrator docs. are native 1920×1080. Those were then sent out as .PNGs into Illustrator where they were placed and rendered out as QTs to match the sequence settings in FCP. I have tried compression markers on both troublesome areas.

David – that ProRes 4:4:4:4 re-render is a good idea – I will give it a try. You are correct in that they are DVCPRO HDs out of Motion and that they are predominantly RED – haha. However, they do appear solid and fine when an HD QT is rendered out of FCP so would the ProRes solve anything if the problem is Compressor’s ability to downscale the titles well? I guess I will see…. Would it be better just to throw the RAW Illustrator files in over slug in FCP in place of the QTs that were rendered out of motion?

On the pulsing: So, would it be agreed to go with a One Pass VBR best setting in compressor given the pulsing problems with the two pass VBR and that a solid CBR will not fit onto the DVD at 90 mins. setting? The one pass does eliminate the problem, but what would your recommended bitrates be for the one pass?

THANKS!
Tim

[Neil Sadwelkar]

My experience with HD to DVD, particularly non-square sizes like 1280×1080 (faux HD?) for 1920×1080 is…

Export sequence at its native HD settings.
Check how the movie looks in QT player. It will look fine.
Run this HD QT through Compressor and make an SD quicktime, same frame rate.
It will be a 720×576 or 720×486 QT which will be anamorphic. That’s normal.
Use this SD QT to compress to m2v in Compressor.

It will look better than an HD QT to SD m2v conversion.
Conversion time even the sum of the two stage HD-SD and SD-m2v will be less than the straight HD QT to SD m2v.

Experiment with the frame controls in Motion especially in the HD to SD conversion.
Do trials on small 1-3 min segments till you hit a good setting. Then do a longer 10 min compress, and then go for the whole thing.

———————————–
Neil Sadwelkar
neilsadwelkar.blogspot.com
twitter: fcpguru
FCP Editor, Edit systems consultant
Mumbai India

[Timothy Anderson]

Hi David,

Just an update – so I was able to re-render the Illustrator titles in Motion in ProRes 4:4:4:4 as you suggest above and in the raw .mov QTs, I did notice a small improvement of the titles. HOWEVER, when encoded to MPEG-2 via Compressor, the same terrible results persist with no visable improvement.

I have literally tried dozens of different workflows, settings, etc.. and am beginning to become convinced it is impossible to achieve clean, crisp titles from Illustrator on a SD DVD made by DVDSP and encoded for said DVD by Compressor. Compressor just can’t seem to do it correctly – neither able to scale down HD titles appropriately nor able to produce good results even when the Illustrator titles are created at native NTSC DVD 720:480 (anamorphic or not).

Out of ideas – downtrodden…

Tim

[Mark Petereit]

Just a thought. If your end product is DV resolution, wouldn’t it make more sense to render your title graphics out of Motion at DV resolution? If you render your vectors in Motion to HD, then resize to DV in compressor, you’re losing the advantage of your vector art, since you’re resizing rasterized HD video.

[Jeremy Garchow]

So let me get this straight. According to this post, you have a 1280×1080 ai document at the DVCPro HD PAR.

You then exported out of illustrator as a PNG, then brought that back into illustrator? That last post is confusing.

Also, on the red titles, take a look at your vectorscope in FCP. Are the levels out of bounds?
Confusing.

[Ken Jones]

[Timothy Anderson] “when the scene is encoded by itself with a CBR encode of 7ish, the pulse disappears – my problem is obviously I cannot encode the whole film (89 mins.) with a CBR and fit it on a DVD.”

That’s true – you will never fit and 89 minute “7ish” CBR on a single-layer DVD.

Have you tried burning to a dual-layer DVD? Try a CBR in your “7ish” range. The Compressor Inspector/Summary will tell you how big the completed file is going to be. Adjust the CBR setting so the final size will around or not much more than 8.0 GB. Also make sure your “Resize Filter” is set to “Best”. This is very important to deal the weirdness of going from HD-SD.

[Michael Sacci]

CBR at 6.4Mbps is what is needed for that length with the use of AC3 audio at .192Mbps.

Because of the advice to use DV resolution, if he is just talking about the resolution that is one thing. You never want this workflow to touch DV codec.