Why is Compressor’s 720p-to-DVD downrezzing better than 1080p-to-DVD?

[Craig Seeman]

Because 1080 has higher resolution it tends to be harder to downrez. Read these two articles. Generally you’ll want to soften the 1080 image a bit.

https://www.xdcam-user.com/2011/03/getting-good-sd-from-an-hd-camera/

https://www.xdcam-user.com/2009/11/getting-sd-from-hd-and-the-problems-of-oversampling/

[Mike Boedicker]

As an update: I’ve seen transcoded the original 1080 24pA DVCPRO-HD test clip to ProRes 422 HQ, then used Compressor to burn the DVD from that, and the results were dramatically better — no stair-stepping and only minor compression artifacts visible.

I also tried using Toast to make the DVD from the original DVCPRO-HD clip, and while the results were slightly better than Compressor’s (no noticeable stair-stepping), there were a lot of other artifacts.

[Craig Seeman]

It’s not just the format but the resolution. Some HD cameras resolve higher than others. Sony EX cameras tend to be higher then HVX camera for example. The higher the camera resolves the greater the chance of line twitter.

[Mark Minshall]

Hi there , was having problems with pixelation when exporting from fcp 7 for sd DVD,I got this reply from Philip blooms forum, and found it very helpful.

mark minshall said: ↑
I have another question, is it possible to put h264 in an sddvd as it always looks better in the compressor viewer,basically im getting a slightly pixelated look in the mpeg2 settings of compressor even with 2 pass vbr
No – DVDs use MPEG2 only.

But again, I’m hearing little alarm bells when you say ‘slightly pixelated’.

That can be
Due to strong colours (blocky artifacts especially around stage or disco lighting – DVDs’ colour is ‘roughly speaking’ a quarter of the resolution of the image) – reduce saturation slightly
Due to poor down-scaling (but usually you see ‘twinkling’ detail and aliased lines)
Poor scaling up of smaller-than-SD sources
Now, I’m out on a limb here with no reference to what you’re actually seeing, except that I know you’re shooting interlace on one of your cameras, both of them are HD and you’re making DVDs from them. So that’s my way of saying I could be completely wrong here…

But I could easily see a situation where your HD project is coming out at half-rez and being made into SD by being scaled up. That would be a bit naff, quality wise. Also, if you’re deinterlacing the interlaced stuff at the compression stage, you’re deinterlacing the progressive stuff too – which will lower that resolution as well.

I’d prefer to see you making one progressive-scan master movie from your Sequence (no interlace or mix thereof) and using the anti-aliasing control in Compressor. The point is that there’s TOO MUCH info in 1080 to make a nice DVD. High end software does a slight blur on 1080 material before scaling down to SD, which removes some of the information and stops aliasing.

If your clients are asking for ‘DVD’ and ‘Nice Web Movie’ or something they can watch at home as a computer file, remember the best SD comes from 720p50 – the reason being that each frame gets used to make a unique field for the DVD – looks great – and you’re left with a 720p50 H.264 movie (no stutter from ENG camera styles) for YouTube/Vimeo/Home use.

Can you think of a display device other than the old fashioned CRT TV set that uses interlacing? Computer screens and almost all domestic TV sets are now natively progressive – and the cheap ones will deinterlace ‘interlaced’ 1080i badly. Best feed it nice progressive from the get-go, or go the 720p50 route.