Shot 1080 30p -> 24p or 60i for Blu-Ray?

[Greg Voevodsky]

Ok - Blu-Ray sucks in that it does NOT SUPPORT 30p. I have shot beautiful "Hawaii Beaches" and waves with my EX1 in 1080p 30fps as recommended from my stockfootage house and am soon to be making both DVDs and BDs.

I was simply going to go with 60i for BD - being the most info and best picture quality, however, from a marketing perspective and ignorance on Amazon.com - I'm considering converting it to 24p 1080p (supported by BD) rather than 60i 1080i - purely from a marketing perspective. There are a bunch of dumb techies wanting 1080p over 1080i for their new HD TVs and think it is better no matter what... which it may or may not be... fine for 24p film but for 30p HD -> 60i is probably better technically... So - here are the questions:

1. Has anyone done this test 24p vs 60i on BD from 1080p 30fps?

2. How much of a look difference is there on BD? (I assume very hard to to tell 60i from the 30p original) However, Im worried about jerky motion in 24p from 30p??

3. If 24p does NOT look much different than 60i from my 30p.... then what is the best way to convert 30p to 24p in FCP? Do I just open a new sequence and drag and drop? I read this is the best way to downcovert to SD for DVD... and then encode?

4. Side Question - What setting do you use for SD - obviously DV (which is what I use to shoot) is not the format to transcode to... do I use uncompressed? or pro-RES SD? Right now - I'm in pro-rez HD and loving it. I assume I drag and drop into pro-rez SD - and then encode - with my favorite Bit-Vice?

Any other work flows, examples - warnings etc... are much appreciated. Thanks!

[basspig]

I have dealt with this same dilemma. Here's my observations of various conversions:

Converting 30p footage to a 24p project: Preserves spacial color subsampling, but has significant motion artifacts, ie, a frame dropped every 6 frames to accomplish 3:2 pulldown.

Converting 30p to 60i and burn as a 60i BD: Preserves temporal resolution, but reduces color subsampling significantly.

The test material I used was some parade footage that had intense colors, such as red fire engines, floats with 'day glo' paint colors, etc. These intense colors appeared blocky on the 60i version of the footage. Information is simply lost.

In each case, the producer must evaluate the nature of the material and determine whether the spacial or the temporal resolution is more important. Low motion scenes with high color content would favor preserving spacial resolution over temporal. High motion, subdued or muted color scenes would favor temporal over spacial resolution. Program material must be evaluated on an individual basis.

WRT SD output, I speak from an Adobe point of view, but I'll state that I do all editing work in the native resolution of the source material and downconvert at render time. The results are superb--often superior to DVD products mass produced by Hollywood. The two things one has to be acutely aware of are MTF (Modulation Transfer Function) and Aliasing. If you know your software system and rendering engine well, you can straddle the fine line between Mr. Nyquist and Mr. Dull and render out spectacular SD product that rivals 720p broadcast HD.

[moldcad]

Converting 30p to 60i and burn as a 60i BD: Preserves temporal resolution, but reduces color subsampling significantly.

Hi Bass; I'm in PAL area and have a similar dilemma: I'm shooting almost exclusively in 25p, and - when delivering on BD - simply render it out as 50i from Vegas.

I'm not sure how it works, but I'm not getting any combing, or other interlace artifacts (my 25p material being probably interpreted as 25PsF - but that's just my guess).

One thing (see quote) got me interested, though: am I really loosing colour resolution? I haven't noticed that, but then - I wasn't looking for it.

Could you please elaborate on this statement - why should it happen?

[Tom Roper]

For Blu-ray, 24p.

Whether there's a marketing perception attributed to 24p or not, what's not in question is that this is an XDCam EX topic.

Accordingly with the EX (at least), it shoots with more vertical resolution in 24p/30p modes than 60i. I've measured the MTF50 with Imatest, and while I forget the exact numbers, I've got it posted somewhere, but I'll venture it's 15-20% more vertical resolution, maybe more.

Adam Wilt attributes the loss in 60i vertical to dual row summation as a noise/artifact reduction mechanism. In 60i, the EX1 has about the same vertical and horizontal resolution as the Canon XH-A1/XL-H1, which is quite good. But the EX1 otherwise vaults into a higher standard at 24p/25p/30p.

So that doesn't really help you. Blu-ray is a problem for 30p, and really 60i is uncommon for it as well.

24p is beautiful when done properly. It's entirely capable of smooth motion through skill and technique.

Would I rather Blu-ray support 30p? Absolutely! But it doesn't. And as the other poster has mentioned, converting 30p to 24p is problematic.

To make the best of the EX1 and Blu-ray, I shoot 24p.

Also, I agree with everything basspig said, very well put.

[moldcad]

Also, I agree with everything basspig said, very well put.

Tom,

Piotr here - cheers:)

So, could you emlightem me, why

"Converting 30p to 60i reduces color subsampling significantly"?

Tia,

Piotr

[Tom Roper]

Hi Piotr,

Maybe I doth agree too loudly because of basspig's convincing style! I really can't comment on the 30p to 60i subsampling observation.

[Greg Voevodsky]

Thanks everyone... Im not sure I want to shoot in 24p since I do like the look and more info of 30p... and further discussion is encouraged as well as work flows. Special thanks Bass Pig.

[basspig]

Converting 30p to 60i in Adobe Premiere, as I did, produced a very definate and measureable drop in color subsampling resolution. No interlacing artifacts--just a loss of color information. One may speculate on whether this is a factor of 4:2:0 color sampling, or something else going on with the conversion.

You're welcome, Greg. Glad to have helped.

[moldcad]

Basspig, any theory behind loosing the color information at conversion? I don't know about Premiere, but using Vegas I simply render from my 1080/25p project to a 1080/50i BD format; when I re-open the m2v clip in Vegas it looks identical (only is flagged upper field first now)?

[basspig]

Yes, in 4:2:0 interlaced images, vertical resolution of the chroma is roughly halved since the chroma samples effectively describe an area 2 samples wide by 4 samples tall instead of 2X2. In addition, the spatial displacement between both fields can result in the appearance of comb-like chroma artifacts.
As is the case in the 30p to 60i conversion, this effectively cuts the chroma resolution in the vertical plane by half. Noticeable on some material, but not all. The more intense the primary colors and the sharper the edges of those colors, the more likely it will make a visible difference. From across a room though, it would be insignificant to the average viewer.

[Tom Roper]

Rather than re-encoding the video, there are some utilities for adding repeat flags into the stream. In other words, the video remains progressive frames as it was originally encoded, but through the application of repeat flags, the frames are repeated in a 3:2 or 2:2 pullup cadence inside a 50i or 60i stream.

Would that still cause the halving of the chroma samples? I would not think so.

I have actually done this for Blu-ray, mixing 24p and 60i on the same timeline. It's done by adding the repeat flags (no rendering involved) to the 24p stream, and dropping it onto a 60i timeline.

It plays back from a Blu-ray authored disk with 24 frame motion for the 24p clips, and 60fps motion for the 60i clips. Other than going from 24p to 60i as the clips change, I didn't see any negative consequence to adding these 3:2 repeat flags to the 24p stream, it looked great played back from a 1080p Pioneer plasma at 60i.

But I do vaguely recall a problem trying to make that concept work with 30p, can't recollect. May give it a whirl again...

Edit:
The utility I use for this, dgpulldown.exe regards 29.97p as already being 60i, so it won't add the repeat flags to 30p. It's really just limited to adding 3:2 pullup flags 24p->60i.