Downconverting a sharper SD image

[Greg Boston]

Originally Posted by Doug Jensen

No, I would never stretch the footage. I cannot stand that when I see it on TV.
I'd either put black pillar boxes on the sides of the screen or add graphics to fill the empty space. Watch any of the nightly network news programs (CBS, NBC, ABC) and notice how they mix 4x3 footage by adding motion graphics to fill the space on the sides. I'd say it works pretty good and is much better than stretching stuff.

Another alternative would just be to produce your whole video in 4x3 and crop all the 16x9 footage. I nice thing about working with HD clips in an SD timeline is that you can change the Motion tab settings to a setting between 50% and 100%, and then slide the clip around within the canvas to get the framing that works best.

Personally, I am a fan of putting 4:3 into 16:9 using the same video on a lower track enlarged and blurred to fill the side pillars. It gives a great effect and IMO, is the least visually jarring. You'll see this on some of the networks also, especially in sports retrospective pieces where they have to mix older material with the new.

-gb-

[Doug Jensen]

Greg, yes that is another good solution I forgot to mention. Good idea.


Anton, people have been shooting 24P and 25P for decades. If you're judging your footage on the camera's LCD or on a computer screen you are probably not getting a true picture of what the footage really looks like. Also, shutter speed is important.

[Anton Strauss]

I would never judge anything on a PC monitor

when I edit, the component out of Edius SP hardware is connected to 42" full HD Plasma

I think the real problem is that the DVD spec is interlaced, I noticed that newer players perform better, especially Blu-ray players because they upscale the DVD and somehow manage to hide the artifacts

[Kevin Walsh]

Originally Posted by Doug Jensen

Hi Dan,

Yes, I edit exclusively with FCP. I also have an F350, EX1, EX1R, and F800. I use the exact same editing workflow for all of them.

Personally, I don't think there is any reason to use ProRes in a typical XDCAM workflow. Okay, if you want to render effects and stuff like that to ProRes, that's one thing, but other than that ProRes doesn't povide any benefits and just wastes time and hard drive space.

Although I don't consider myself to be a Final Cut Pro expert, I have found editing XDCAM to be very easy and simple with no special steps or unusual issues to deal with. I strongly believe it is very important to edit with Sequence settings that are as close as you can come to matching your final output. In other words, if you aren't going to author a Blu-ray disk or some other HD final product -- then don't edit in HD. Edit in SD. Choose sequence settings that match your output -- not the source footage.

Here are the steps I follow for Final Cut Pro for a project that will be released on DVD or the web.

1) I shoot most footage with the HQ1080/30P Video Format.

2) I open a new Sequence in FCP and use the "DV NTSC 48Khz Anamorphic" preset.

3) I change the Field Dominance to "NONE"

4) I edit the entire program within that Sequence.

5) When I'm done editing, I then Export a QuickTime movie of the Sequence. I choose "Current Settings" and I do NOT choose to "Make Movie Self-Contained".

6) I then take that QuickTime movie and bring it into Compressor.

7) I then choose the Compressor preset for DVD Best Quality and modify a few of the settings (such as bitrate), but nothing major.

8) After that file is finished rendering, I bring it into DVD Studio Pro and author the DVD normally

I found that this workflow is fast and easy, requires no extra software or plugins, and produces very nice results. Plus, I have not given up all the huge benefits of working tapeless.

This is what I do too. It works like a charm. Doug what do you do if you have to deliver both Blu-ray and DVD?

[Doug Jensen]

Hi Kevin,

I've never had to deliver on Blu-ray. I thought I was going to have to once, so I bought Blu-ray burner in December 2008 and the box is still sitting on a shelf with the shrink wrap intact. By the time I ever use it, it will probably be as outdated as a floppy disc drive.

However, getting back to the intent of your question, when I have to deliver on other HD formats, I edit with an XDCAM HD 422 timeline that matches the majority of my raw footage. I would then downconvert the final edited video if a DVD version was also needed. That workflow also gets great results, but I still feel there's no reason to edit in HD if I'll never need any HD output. There are many advantages to editing on an SD time line if that's the only output I'll ever need.

Once again, ProRes would not be needed at all in my workflow. I don't think there's any benefit to using ProRes for people who are already working with XDCAM. ProRes is a great option for the "other guys", but I don't have any use for it.

[Uli Mors]

Hi,

I really dont want to open the "p vs i" shooting discussion again, but here its part of the question.

My experience: If you shoot progressive (25p, 30p) , downconversion in NLEs is MUCH BETTER than shooting interlaced.
In both worlds, NLEs do a (deinterlacing if needed and) downscaling, of course.
Its a question of filtering "too high" detail frequencies to avoid aliasing and other artefacts - simply spoken: SD cant take as high frequencies as HD can.
Since most NLE Plugins do some simple downscaling, the results are different.

In case of deinterlacing, the process is even more complex. But there are software workflows that produce a good quality even with Interlaced footage.

What I found is that any hardware conversion (HD->SD, progressive or interlaced) is WAY better if done by hardware. I guess its a question of filtering high frequencies too.
I tried to copy my HD master back to camera (I used XDCAM F330, XDCAM 700 and HDV) and play the signal out in SD. IT´S FANTASTIC.
In case of F330 (aka 350, 355 etc.) you can benefit from 2 things:
a) use SD Detail to enhance your SD Signal to your taste
b) use Firewire AVC out (switch from FAM to AVC) to ingest the footage digitally. The SD Signal via Firewire is great (digital) quality with the same SD Detail enhancement

In case of PDW700 and others there is often no AVC stream out possible. In this case I ingest analog via component or composite outs.

Again, I found the hardware conversion to be the winner over most NLE conversions. Especially with interlaced material.

Have a try & compare!

Doug, I like your air show , but I am afraid this is not the problematic footage we are talking about.
Instead, have a try with slow pans or zooms over straight lined objects like brick-lined houses or walls, mega-detail city overviews etc. I assume these are the shots that get critical when converted to SD.

Regarding shooting P or I: Its right, dont jugde your P footage over an LCD screen or the viewfinder. If played correctly in the NLE or a BluRay /DVD Player the shutter impression is less.
For shooting: I dont like shooting (PAL...) 25p. I find it often too difficult to quick focus or follow a person.
To my opinion (an broadcast opinion in general) 1080i is a good compromise as long as you have a way to downconvert to SD (DVD etc.) in a high quality. I dont agree that 50i (60i) is an old fashioned method. Its a compromise for motion as long as we dont have 1080 50p.

I recently shot a comedian piano concert in 25p and my customer complained: "Whats wrong with the motion when the comedians jump an and run around?". And my 2nd cameramen complained: "it was so hard to be in focus". That was a test for me. And the reason why I will shoot most of my footage interlaced again (had an ice show yesterday - fantastic!).

ULI

[Alister Chapman]

Downconverting progressive material is the least problematic because you can simply divide the vertical and horizontal pixels by a near symmetrical amount to get the smaller frame size you want. Interlace is much harder as first you must separate out the fields and then you have to divide each field by an asymmetric amount. In the case of NTSC you have to drop every other line, plus an additional line every 4 lines, for PAL you drop every other line for 3 lines then keep a line. This un even number of dropped lines can lead to stair stepping and artefacts. The problem is even worse if you convert to DV as the DV field order is reversed and many NLE's will make a complete hash of this.

The next problem is the need to include some form of low pass filtering in the down conversion process to reduce the high frequency components in the HD image. Just as a decent video camera will have an optical low pass filter to eliminate aliasing, we need to do the same with any video downconversion. I posted some examples of this on my web site.
XDCAM-USER.com Getting SD from HD and the problems of oversampling.
One cure is to add a 3 pixel blur to the HD image before you down convert it. The difference in the final down converted image can be quite striking.

[Les Wilson]

This is now the location of Alister's article:
Getting SD from HD and the problems of oversampling. | XDCAM-USER.COM