HD to SD Best Quality workflow

[Jason Livingston]

Ditto on what Jon and Ken said. I was thinking of posting something like this myself and I’m surprised it doesn’t come up more often.

I’ve seen a lot of suggested HD-to-SD workflows for FCP and Compressor, but anyone who knows what good SD looks like will know that most of them look terrible. (I’ve actually done empirical tests using res charts and the loss of quality is obvious.) Some people say “well that’s just what SD looks like” or “you have to watch it on a SD TV to see what it really looks like,” but I’ve seen what a good hardware downconvert looks like and there is no comparison.

Sure you can print to SD tape and capture back in via a video card with hardware downconversion, but who has time for that?

[Michael Gissing]

[Jason Livingston] “Sure you can print to SD tape and capture back in via a video card with hardware downconversion, but who has time for that?”

Two real time passes versus software rendering. I wonder who would win that race?

[Jason Livingston]

You make a good point, although not everyone has a DigiBeta or similar high-end SD deck and tape stock lying around. 🙂

[Ross Byron boughton]

hi Jon i am going to try your PC method out latter today and tomorrow the way i have been down converting is going into compressor and choosing DV50-NTCS that knocks it down to size and makes it 29.97
i’ll try and post some 30sec clips for everyone best Ross b
p.s will your method do the 23.98 to 29.97 frames? tia

Ross Byron Boughton
Editor / Webmaster / Sound Designer
https://www.inch.com/~bellevue/

[Jon Geddes]

[ross byron] “p.s will your method do the 23.98 to 29.97 frames?”

Not quite sure what you mean by that. Just about every mpeg2 encoder has the option of enabling the progressive flag.

Jon Geddes
http://www.precomposed.com

[Ross Byron boughton]

hi guys my PC wont see the QTProres444 codac at the moment so i will come back to it,
got a bit closer in the mac environment compressor to 10bit uncompressed SD with all the frames set to best settings then a 2 pass mpeg in compressor at best settings
looking to try HC encoder for the .M2V files

Ross Byron Boughton
Editor / Webmaster / Sound Designer
https://www.inch.com/~bellevue/

[Jon Geddes]

Do you have the latest Quicktime version on your PC? Also, you may need to install the Prores codec as well.

Jon Geddes
http://www.precomposed.com

[Alessandro Capitani]

Jon

I have read the entire post and since I am facing the same issue these days, I am completely with you.

First of all, I have not read in the thread of improvements provided by FCP7 (I am still using version 6) so I assume Apple has not worked on improving scaling/interpolation algorithms in FCP/Compressor/Quicktime in latest Final Cut Studio, that’s a bit sad.

Second, since I am in an almost completely file based enviroment, the “pass through baseband” solution (i.e. via Kona/hardware downscaling) sounds a bit ridicolous in my case.

I have taken some time to test some solutions and I have find interesting results. My workflow is implying editing in ProresHQ 1080/50i and creating file masters at PAL resolution (720×576) both “anamorphic” and letterboxed.

First, the “bonsai method”.
This is basically based on disabling the (weird) interpolation built in FCP by setting the “Motion filtering quality” setting in “Fastest (linear)” (not in Better Quality as I have read above, this is exactly where FCP is faulty in the way it interpolates during picture scaling! Apple, are you reading?).

Complete detail of the bonsai method here:
https://www.produxion.net/2008/04/08/hd-to-sd-conversion-the-holy-grail/

The solution is simple and free but it has its limits. It performs realtively well on “real life” pictures but, being antialiasing completely disabled, can create problems on graphics elements or sharp picture details. By the way it deserves at least a try, it is free and damn simple!

I have tested other solutions, I have got some interesting results with MPEG Streamclip (be sure you set “better quality scaling” in the Export window), but it is not perfect. Whilst 2D scaling is much better that FCP’s one, it seems it uses a 4:1:1 codec as intermediate during processing, so I notice some visible sub sampling artifacts on border of sharp and heavily saturated colours, especially red.

Adobe Media Encoder provided some weird results, maybe I just need more time to go more in depth in testing this solution.

The best results so far are from a brand new conversion solution from Miraizon.
This (new to me) software company has just released version 1 of their format conversion software: Reframe (https://www.miraizon.com/products/products.html). I have tested Reframe (for downconversion to SD only, no standard conversion tests for the momement) and I have got some promising results.

If you will test Reframe yourself be sure to set approriate intermediate codec (I ended up on using Apple Uncompressed).
Known downsides for the moment:
– it seems it processes in progressive despite source is interlaced. My source material is psf, so in my case it is not an issue.
– if you have multi-channel Quicktime, take into account output is a stereo pair at a time for the moment.

For 129 bucks I think it also deserves a try. Money back guarantee in case your are not satisfied 😉

Hope my findings will help you on getting a solution for your workflow. I will appreciate your opinion on this.

Best,
Ale.

“Editing is 80% diplomacy and 20% actual skill”

[Alan Langdon]

Vince-

I am interested in more details on your preference for AE downconverting HD footage. Could you describe the workflow? I am an extensive user of AE, and would tend to agree with you.

Currently I have a master HD file that is Apple Intermediate Codec 1920×1280 and would like to put in on a SD DVD. Do I stretch in the AE Render/Output Settings, or place the HD clip in a SD anamorphic comp? Should I introduce any fields or is this just adding noise/degradation?
My first intention is to have a SD master. Then, put it on a SD anamorphic DVD. I will use Toast, COmpressor and/or MPEG Streamclip for the encoding.

I was analysing a VOB from a anamorphic SD DVD somebody made, they originaly shot in 2K, but I dont know how they downconverted. But it looks pretty good on DVD and when I open the VOB in MPEG Streamclip, it claims the file is Upper Field dominant, although the image is definitelly not interlaced, but progressive 30p. Any ideas about this?

Sorry to through so many questions, I hope you can help with any of them. Thanks!

[Vince Becquiot]

I usually just create a new comp, either square 854×480 (that will show you the reduction quality without pixel aspect ratio conversion), or DV widescreen, and resize inside it. Not much more to it.

Again, that’s the way we’ve been doing it for years and there maybe (probably are) better ways to do this, especially with hardware conversion, but I’m pretty happy with the results, and I think that’s about as much time as I can bare spending on that process.

Vince Becquiot

Kaptis Studios
San Francisco – Bay Area