30p vs. 60i

[ksmprod]

I was wondering what is the better format to shoot in. I know 30p does better for web, but if I am shooting a wedding, and am just going to put it on DVD, should I shoot in 60i instead to get better quality? Is there a difference when it is time to edit (besides sequence settings?)

Thanks!

 

[David Jimerson]

Purely aesthetic. Choose the one you -- and your client -- like the look of better.

 

[BobDiaz]

I may need to put this into the FAQ....

I admit that I'm no fan of interlaced and always chose progressive. However, there are solid reasons for my position.

While 1080/60i has a high field rate then 1080/30p, each field is only 1920h x 540v. Notice that the vertical resolution is 1/2 the 1080 lines. This says that a freeze frame and any slow motion is going to have (pick one): (a) Poor vertical resolution (b) Better vertical resolution, but with misaligned fields due to movement.

AT least with 1080/30p, both freeze frames and slow motion does NOT have those issues. Now, there is a price, you lower the temporal (time) resolution to 1/2 of the 1080/60i. In the case of fast moving sports action, we really should have the high temporal resolution. However, 720/60p would work better for sports action.

Now unless the bride is going to do back flips while slam dunking the ball, weddings aren't about fast action. Thus, the higher temporal resolution isn't necessary.



Bob Diaz

 

[David Jimerson]

No, but she may still prefer the "live" look. Clients don't care about technical issues. They just care about if they like the footage or not.

 

[GrahamH]

I dont really agree that 1080i has "poor vertical resolution" because in practice, through persistence of vision, the viewer is still seeing 1080-high frames, not 540-high fields.

 

[ggrantly]

Hey David J.

Congrats on 19,999 posts!

Best Regards,
Grant

 

[v3rlon]

With weddings, its not really the vertical resolution that matters so much as the horizontal. If you can film it so the bride looks 10 pounds lighter without impacting the rest of the guests or bridesmaids, you'll have a licence to print money.

(sorry, couldn't resist)

Really, its what Jimerson said. Give the clients what they are looking for. Normally, 30p should be fine, and will give you access to longer shutter speeds which may be helpful in some shooting situations.

I guess the other factor is whether or not your NLE does interpolated slow motion and that is an effect you intend to employ. With interpolated slo-mo, it doesn't matter as much. Without it, 60i would give you more to work with. If you don't use it, ignore this whole part.

 

[BobDiaz]

I dont really agree that 1080i has "poor vertical resolution" because in practice, through persistence of vision, the viewer is still seeing 1080-high frames, not 540-high fields.

The research has been very clear about this one. If an object is moving, it's called dynamic resolution, if you look that the photo of the bike rider, the lines do not match between frames. Persistence of vision doesn't help put the fields together.

I understand it's easy to assume that with a sill image persistence of vision should put both fields together in our minds, but the research says, it doesn't work that way.

Over the past century a number of factors have been observed to improve the quality of sampled images. Three of these phenomena are reviewed. The reasons for improvement based on perception studies are given, the improvements are quantified, and how they can be used in a practical television system is shown. Interlaced scanning reduces vertical resolution for several reasons involving spatial and temporal masking effects in visual perception. A progressive scan avoids these limitations. Because of the oblique effect in vision and the statistical orientation of lines in scenes, diagonal sampling reduces the required number of pixels in an image. Quantitatively, our measurements show that the number of pixels is reduced by a factor of 1.4 for the same perceived sharpness. In addition, by taking advantage of the octave-wide tuning bands in visual perception, our measurements show that the perceived resolution in the vertical direction for a progressive scan can be double that of an interlaced scan. Using diagonal sampling, a 1920×1080 image with progressive scan at 60 frames/s requires the same transmission bit rate as a 1920×1080 Cartesian sampled image scanned interlaced at 30 frames/s. This results in an image that appears to be much sharper than the 1080 line interlaced format without the interlace artifacts.

http://spiedl.aip.org/getabs/servle...0003000597000001&idtype=cvips&gifs=yes&ref=no

List of other studies: http://scholar.google.com/scholar?q...esolution&hl=en&as_sdt=0&as_vis=1&oi=scholart


Now there are 1080/60p and 1080/120p ... sets that "de-interlace" the video. The only problem is how do the sets do it? A simple low cost system would generate intermediate lines to fill in the gap between the blank scan lines. This is simple and cheap, BUT does not restore the missing pixels.

At the other end a complex motion analysis can be performed to align things that are moving. This takes some real processing power and most likely is not cost effective.

A compromise is to analyze if there movement or not. If a section has no movement, put the two fields together, BUT if there's movement, just generate intermediate lines to fill in the gap between the blank scan lines. This system is possible and on still images does have the higher resolution, but moving objects are still only 540 lines vertical.

Sadly, no TV manufacturer says how the de-interlace the video in the 1080/60p. 1080/120p sets.



Bob Diaz

 

[John_in_NY]

In my opinion, it is better to shoot 60i than 30p. The way I look at it, you only have one take and 60i has been the standard for many decades. I prefer it myself, everyone is used to it and it is not an issue when viewing in real time. You can realistically convert a 60i image to a 30p image in post, but not realistically convert the other way around. If you were instead talking about 1080 60i vs 720 60p, that would be another matter all together.

 

[BobDiaz]

It's a free country, so what ever looks or feels good for you that works, go for it. :thumbsup:



Bob Diaz

 

[Alejo_NIN]

It's a free country, so what ever looks or feels good for you that works, go for it. :thumbsup:



Bob Diaz

quote of the day!

 

[ksmprod]

Thanks for all the input. I have been shooting 30p for years and was just wondering how much of a difference it would make.

Thanks!

 

[hazydave]

The research has been very clear about this one. If an object is moving, it's called dynamic resolution, if you look that the photo of the bike rider, the lines do not match between frames. Persistence of vision doesn't help put the fields together.

Of course, much of the research on interlacing is based on the days of analog television, too. There latter day televisions added to the problems by doing all sorts of clever tricks. For example, comb filtering. The comb filter subtracts 1/2 of the previous and next scan line from the current, based on the assumption that there's not all that much change from line to line. For NTSC, this is very effective at separating chroma and luma without over-filtering the luma (I wrote a software NSTC decoder, back in 2008... don't ask

I understand it's easy to assume that with a sill image persistence of vision should put both fields together in our minds, but the research says, it doesn't work that way.

The original point of interlace was to allow for a greater preceived resolution without straining the best of 1930's technology, bandwidth-wise. Also consider that it's not just your visual system providing the presistance, but the phosphor of the televison screen.

This has been rendered rather moot these days, anyway, since every digital television display is inherently progressive, anyway. So while I may feed 1080/60i into my 71" Samsung, the display only runs at 1080/60p.. the software's de-interlacing, inverse-telecining, and all kinds of mean, nasty things.

With that said, you're certainly better off, all things considered, with progressive video in most cases.

Now there are 1080/60p and 1080/120p ... sets that "de-interlace" the video. The only problem is how do the sets do it? A simple low cost system would generate intermediate lines to fill in the gap between the blank scan lines. This is simple and cheap, BUT does not restore the missing pixels.

There aren't really any gaps.. that's just presentation on an analog display. There are any number of simple de-interlacing algorithms... we all know these, because they're generally options available in DAWs: field blending, smart blending, bob and weave, etc.

But the algorithms used in modern HDTVs are very sophisticated. They're using motion tracking with correlation and other magic tricks, to really create effective intermediate frames at 60p or even 120p (I'm not convinced any of the 240p TVs do real 240p interpolations, but I'm not sure I'd see it anyway). They're looking at multiple frames at a time... given the power of video DSPs and the cheapness of RAM, this is cost effective these days. And they have to sell thing like 120Hz as being better. Without magic tricks, there's really no difference from 60Hz on any video you actually have and a modern digital display (displays being refreshed all at once, not scanned like a CRT).

You can find papers on these in places like the IEEE archives, but yeah, they don't likely tell you just what they're doing, other than through the various patents. This is a fairly competitive field among TV manufacturers, incorporating upscaling and temporal interpolation, too, these days.

The simple interpolation... you can have 30p at full resolution, or 60p at half resolution, those are relatively established algorithms. But they're really trying to deliver the visual impact of there being more frames than really exist. This is also being done in Blu-Ray players.

 

[BobDiaz]

As I had said,

At the other end a complex motion analysis can be performed to align things that are moving. This takes some real processing power and most likely is not cost effective.

Given the increase in RAM size and processing power, maybe I should have left the comment about cost effective off. The rules of the game are changing all the time. :happy:

In the case of editing software, it is possible to take advantage of some of the more complex motion analysis to help fill in the gaps for freeze frames and slow motion with 60i footage. There are going to be issues with color sampling with 4:2:0 on interlaced footage, but the right software could clean things up a bit. It all depends on what the software package has to offer.

On the other hand, one could shoot 30p and avoid all of that...


Bob Diaz

 

[David Jimerson]

Only if absolute resolution is more important than the visual characteristics of the motion. As much as you, the shooter, would prefer to do something because you, the shooter, know something technical about the footage, the client may disagree. And you're not shooting it to please yourself.

 

[hazydave]

As I had said,
Given the increase in RAM size and processing power, maybe I should have left the comment about cost effective off. The rules of the game are changing all the time. :happy:

Yup. For my first HDTV... wow, like over 10 years ago, I went shopping with a number of things in mind. One was upconversion. In those days, they really had all mastered it. Mitsubishi had some decent tech, Pioneer Elite even better, but out of my price range.

There seems to be a ton of work on the "magic" part of this. I found half a dozen articles from Samung alone on the IEEE server. I knew they were getting more sophisticated all the time, but they're really getting a little crazy. There are even folk working on "fake" 3D... where the video DSP analyzes the scene and splits stuff into foreground and background. I'm sure it'll be as good as fake stereo and fake surround were (eg, not at all), but it's kind of surpring they have kind of horsepower to play with. This is all processing on RAW HDTV, in realtime.

In the case of editing software, it is possible to take advantage of some of the more complex motion analysis to help fill in the gaps for freeze frames and slow motion with 60i footage. There are going to be issues with color sampling with 4:2:0 on interlaced footage, but the right software could clean things up a bit.

Yeah, I wasn't thinking on that.. if they're colorsubsampling after splitting frames into fields, it's going to cause distortion. And this would be the normal way to encode 4:2:0, since encoding a frame at once, given interlaced video, would yield even more distortions on high speed video. 4:1:1 solves this, of course, but introduces other issues. That would be a tricky one to fixe completely in software.

On the other hand, one could shoot 30p and avoid all of that...

If you get real 30p, and not 30p encoded as 60i, of course. And there's still the issue of delivery: 30p's not in DVD or BD, so you wind up having to ship it as 60i anyway. At least Blu-Ray supports native 24p.

 

[John_in_NY]

If you get real 30p, and not 30p encoded as 60i, of course. And there's still the issue of delivery: 30p's not in DVD or BD, so you wind up having to ship it as 60i anyway. At least Blu-Ray supports native 24p.

That is one reason I prefer 60i over those other frame rates. I think of 60i as a common denominator with wider compatibility and the best motion, next to 60p.

 

[hazydave]

That is one reason I prefer 60i over those other frame rates. I think of 60i as a common denominator with wider compatibility and the best motion, next to 60p.

Right. If I have to match the HMC40 to another camera, it's pretty much going to be 1080/60i. That's a perfectably acceptable thing if your main traget is DVD/BD, and you can get ok web video, with some de-interlacing evil, at 720/30p, which is usualy the practical limit of web video anyway.

If it's just the HMC40 and you don't need the frame rate, 24p is the other real option. But this is the only camera I have supporting it (Sony's fake 24p doesn't count).