1080 60i vs 30p

[Jim Stamos]

loving my ex1r . ive been shooting mainly 60i. ive read of some shooting mainly 30p. what are the main visual differences as far as which to go with.
i shoot pretty much every type of event.
weddings/corporate/documentary/stage productions.
a good xplanation would appreciated
jim

[Jack Zhang]

30p works better with web based destination content since it's progressive scanning. If you shoot in 60i for the web, you either have to de-interlace by discarding 540 lines or blend the 2 fields together and lower the resolution (There are good de-interlacers out there, but that's not the point.)

30p is technically 1080p, you actually improve chroma resolution slightly thanks to using progressive scan. It's also convenient for web upload since there is no special de-interlacing you have to do.

I would only shoot 30p for content headed for the web. If it's going to DVD or Broadcast, 60i would be best. Cinematic content should be shot in 24p.

Really important: Never edit progressive scan footage in a interlace project. Especially 24p. Final Cut likes to convert that to 30p and the frame rate then jutters instead of applying a telecine. You also lose chroma resolution on 30p footage since it's Progressive Segmented Frames.

[Doug Jensen]

I disagree with a lot of what Jack has said.

30P is not technically 1080P. 30P can be any number of screen resolutions including 1080, 720, and even SD.

There is almost zero content being produced for broadcast TV anymore (outside of sports) at 60i. 30P is the norm for broadcast HD television in the US. 60i acquisition is outdated and, in my opinion, looks terrible.

Editing progressive footage in a interlaced project is no big deal. 30P cuts into a 60i project just fine an even retains the look of progressive. Just look at broadcast TV. The signal is sent out as 60i from the broadcaster/network/cable channel, but the content is shot, and usually edited, as 30P -- and it still looks like progressive when it arrives at your TV.

Vortex Media: VIDEO & PHOTO Tools and Training

[Jack Zhang]

The original poster did say 1080. Yes, the framerate can be any resolution, but the original poster specifically said 1080 in the thread title.

Yes, 30p cuts fine in 60i, but once it is turned into Progressive Segmented Frames, it turns chroma from progressive chroma to interlaced chroma. This amplifies the 4:2:0 phenomenon and the reds look blockier as a result. See what I mean on Wikipedia: (the bottom 3 images of the section) Chroma subsampling - Wikipedia, the free encyclopedia

My rationale behind my choices are that you don't want to mix framerates in your edits. 30p on a 60i timeline would mean any animation, transition, or keyframed material will be in 60i. This causes a framerate discrepancy that I see WAY too often in reality television editing. Telecined 24p on a 60i timeline causes the same effect. Even worse, 24p without telecine on a 60i timeline converts 24p to 30p by duplicating frames, causing the image motion to stutter.

For 30p, you can solve this by editing 30p in 30p, rendering a intermediate 30p file of the final edit, then dropping that in a 60i timeline so any transitions or keyframed material will stay within the same framerate, 30p. Same with 24p, only you take the 24p intermediate and apply a telecine to render an interlaced intermediate to drop into a 60i timeline. (unless you're rendering for HDCAM 24p or HDCAM SR 24p, and for that you have to drop it onto a 24p timeline to print to tape)

Obviously, this will cause generation loss, so choose a high quality intermediate codec with minimal generation loss.

I also agree 30p is a better way to go, it's just how you work with your edit workflow that's critical.

[Luc De Wandel]

It's all a matter of taste, of course, but I don't like the 'choppy' aspect of progressive footage. Probably I'm just old fashioned, because the trend seems to be in favour of progressive.

[Andy Taplin]

It's only 'choppy' when viewed on computer displays that were never intended to display video.

TVs have variety of motion compensation controls to smooth out progressive material.

If you shoot interlaced you have to de-interlace it for any computer-based playback loosing resolution (unless you want jagged edges on all motion) so 30P is the way to go, and it can be easily converted to interlaced for broadcast.

Shoot 720 60P if you want the smoother motion of interlaced video.

If you are shooting progressive for computer display you need to think a little about how you shoot - I would avoid medium speed pans for example which can show the progressive judder. And make sure you choose a shutter angle of 180º ( shutter speed double your frame rate i.e 1/60th) for 'natural' motion blur.

[Alister Chapman]

Originally Posted by Jack Zhang

30p on a 60i timeline would mean any animation, transition, or keyframed material will be in 60i. This causes a framerate discrepancy that I see WAY too often in reality television editing. Telecined 24p on a 60i timeline causes the same effect. Even worse, 24p without telecine on a 60i timeline converts 24p to 30p by duplicating frames, causing the image motion to stutter.

For 30p, you can solve this by editing 30p in 30p, rendering a intermediate 30p file of the final edit, then dropping that in a 60i timeline so any transitions or keyframed material will stay within the same framerate, 30p. Same with 24p, only you take the 24p intermediate and apply a telecine to render an interlaced intermediate to drop into a 60i timeline. (unless you're rendering for HDCAM 24p or HDCAM SR 24p, and for that you have to drop it onto a 24p timeline to print to tape)

Obviously, this will cause generation loss, so choose a high quality intermediate codec with minimal generation loss.

30p in a 60i signal presents no frame rate discrepancy. the frame rate for both 60i and 30p is 30fps.

I don't understand how Jack believes that editing 30p in a 30p timeline, then rendering out a 30p file, then dropping that into a 60i project will help with either colour space issues or the non-existent frame rate issue. All I can see is a significant quality reduction through extra generations. After all once you drop the second generation 30p into the 60i timeline you will get the same colour space issue anyway.

[Tom Roper]

I think I understand some of Jack's points.

1.) 24p -->30p, whether you encode repeat flags or additional frames, there is added motion judder.
2.) 30p -->60i, to put it on a 60i timeline means two consecutive progressive frames will get segmented into fields and blended, introducing a temporal offset that wasn't in the original frames. Even though you still end up with the same overall framerate, the blended fields have been reconstructed with a temporal offset artifact that wasn't in the original frames. Sony Vegas has a "reduce interlace flicker" filter that applies a gaussian filter to reduce the stairstepping on diagonals caused by the temporal blending but this causes softening.

The better solution, is to put everything on a 60p timeline. 24p. 30p and 60i will all look good together. 24p will benefit from 3:2 pullup, 30p gets 2:2 and 60i fields get interpolated or blended, your choice. Render it all out to Avid DNxHD 220 mbps 4:2:2 10 bit, or Lagarith YUV uncompressed, then pass it to x264 with the following parameters:

-- framerate 30000/1000 -- force-CFR -- fake-interlaced

x264 will output a true 30p file (29.97) but flag it as interlaced. That allows 60i playback for Blu-ray of natively progressive encoded 30p frames that haven't been field blended (except for the 60i). The principle benefit is to not muck with the native 30p frames in a way that Jack (and I have noticed) can cause temporal artifacts to appear that do not exist in the original source. Correspondingly, 24p will take on an acceptable 3:2 cadence of the type we are accustomed, and 60i fields will be pre-combined into blended 30p frames twice repeated, just like any HDTV deinterlacer chip would do for us.

Again, the problem to avoid with 30p is the one that's caused when you throw away half of each 30p frame to make a field, and then combine the fields to make a frame. Half the new combined frame would be from a point in time 1/60 second apart. You want to avoid that. If you do, 30p should be indistinguishable when played back as 60i through the deinterlace chip on a HDTV set.

[Doug Jensen]

Originally Posted by Tom Roper

2.) 30p -->60i, to put it on a 60i timeline means two consecutive progressive frames will get segmented into fields and blended, introducing a temporal offset that wasn't in the original frames. Even though you still end up with the same overall framerate, the blended fields have been reconstructed with a temporal offset artifact that wasn't in the original frames..

That is not the way it works. When you transfer 30P to 60i you get two identical fields that are each on the screen for 1/60th of a second. There is no temporal offset because both fields are EXACTLY the same except one is odd lines and the other is even.

Also, I don't know about Vegas, but with other NLE's the flicker reduce filter is for going from interlaced to progressive -- not the other way around. There is no flicker to reduce when going from progressive to interlaced because there's no temporal offset.

Originally Posted by Tom Roper

Again, the problem to avoid with 30p is the one that's caused when you throw away half of each 30p frame to make a field, and then combine the fields to make a frame. Half the new combined frame would be from a point in time 1/60 second apart. You want to avoid that.

Really? That is quite impossible. You can't extract two fields from a 30P frame and say that each field occured at a different point in time -- 1/60th apart. Why? because the whole entire 30P frame was captured at once (igoring rolling shutter for CMOS). So you can divide a picturre into two fields, three fields, a hundred fields, or whatever you want, but all the fields would still just be slicing up the same root image that was captured at one moment in time -- thus no temporal offset and no flickering.

Unless someone is doing something terribly wrong, there is no problem with using 30P nor does it require any special processing or convoluted workflow. My advice to stay away from 60i interlaced acquisition because it has no place in today's world.

[Tom Roper]

Doug I agree with what you say, wrong on the 30p to 60i. I must have been on crack when I wrote that, incredibly stupid of me, sorry. The anti-flicker filter in vegas is for progressive to interlace, it's in the documentation. I'll post the exact wording later.

Would you try something and then explain it for me? Start with your nanoflash. Check the box for "record psf as progressive." Make a 30p recording of something with lots of fine, linear horizontal detail recorded to your nanoflash. Move the camera around a little bit to add some motion. Now hook up the nanoflash to an HDTV monitor, with the HDMI. Play the file back twice, once with the checkbox for "play progressive as psf" checked, and then again with the box unchecked.

Do you not see added moire and line twitter in the displayed image for the psf output? Why? It's the same file.

[Jack Zhang]

My explanation is not for signals or straight cuts, this is if you do animation edits or keyframed edits in a timeline that is different from the source framerate. For instance, if you speed a clip up 200%, a 60i project would make a frame a field, creating framerate discrepancy when displayed at 60 fields. Same if you did an animation or a transition, a framerate discrepancy in the 60i edit would mean there are more frames/fields in the animation as opposed to the source footage, causing a somewhat jarring effect. 30p in 30p with edits applied would sustain 30p without going higher in framerate if such edits were applied. It's basically a framerate cap.

Vegas with "Disable Resample" tends to solve some of the ghosting issues involved with what Tom has experienced. Another thing to check is if the frames are Quantized, improperly quantized frames usually suffer from inter-field intricacies. It also may be down to the encoding codec, since I've already brought up that 4:2:0 interlaced is lower resolution chroma-wise than 4:2:0 progressive.

I would not mix more than one framerate in a single edit. If possible, I would choose one framerate and stick to it, then conform the rest of the footage to that framerate if there was other framerates. The only ones I would never do are conforming 24p to 30p and editing 30p in 60i. If there is 24p content, I always make the project 24p and conform 30p and 60i to 24p. If there is primarily 30p footage going to a 60i timeline/destination, I would make the timeline 30p and put the rendered 30p intermediate in the 60i timeline.

Psf may just be a 1080i signal in some cases. I would check the nanoFlash file on a computer to be absolutely certain it isn't the TV's de-interlacing that's causing the odd temporal motion artifacts. I see that each time I feed a 1080i signal to an HDTV that has 30p footage in origin.

[Tom Roper]

Jack,

Smart resample in Vegas interpolates frames. It's only applicable when up-conforming to the project frame rate. For example, if you had a 60p project timeline, and you dropped a 24p clip onto the timeline, smart resample will create new frames. If you disable resample, it will instead add pullup flags and just repeat the existing frames to conform to the project frame rate.

The observation I'm making about 30p, as fond as I am of it, you lose some of the cleanliness of the progressive images somewhere along the path to playback, most likely as you stated in the hdtv deinterlace chip. Theoretically yes, 30 psf should get reconstructed to perfect 30p but in practice it always seems to acquire interlace artifacts, losing one of the reasons for doing it.

I agree about conforming to one project frame rate, that's how I do it, and the best way I've tried is to conform everything to a 1080/60p project timeline, and render to a low loss intermediate like DNxHD, for final render as h.264 AVC, 30p flagged as interlaced using x264. The 60p timeline plays nice with 60i/30p/24p, all three at the same time. A 30p timeline as you noted does not play nice with 24p. Whether you smart resample (interpolate frames) or disable resample (add repeat flags for pullup) is up to you. I use both methods situationally. One method is not best for every situation. I think this benefit from using a 1080/60p project will carry over to animations and transitions between the key frames.

Again, 1080/60p project timeline, render to DNxHD, feed that to x264 and ask it to render 30p flagged as interlace. The bottom line, is you get a 1080/60i file output that's Blu-ray and broadcast compatible, yet was progressively encoded end to end.

[Jack Zhang]

I would still think if everything you shoot is in 30p, edit in 30p. However, if you have an even mix of all frame rates, I would certainly consider the 60p route. My leaning is that the framerate of the majority of your source footage should be your edit framerate.

Examples of shows I can think of that need a workflow makeover:

American Chopper: all 30p, should be edited in 30p, but edited in 60i.
Auction Kings: all 1/24 shutter 24p, Should be in a 24p timeline, but the progressive footage is telecined to 60i to be edited in 60i.
Anthony Bourdain: No Reservations: V1U shot in 24p, graphics are in 30p, edit is in 60i. I would make the graphics 24p and the whole project 24p.
Man vs. Food: Same as above with No Reservations, except the V1U part.

Example of a show that nailed it when it came to native framerate editing:

Departures (a Canadian international travel show on Citytv and OLN): HVX200, shot in 24p with a lot of overcranked 24p shot at 60 frames, graphics/animation in 24p, edited in 24p. (Google it if you don't know it)

Mismatching framerates when it comes to animated keyframed effects/transitions really bugs me to no end.

[Doug Jensen]

Originally Posted by Tom Roper

Would you try something and then explain it for me? Start with your nanoflash. Check the box for "record psf as progressive." Make a 30p recording of something with lots of fine, linear horizontal detail recorded to your nanoflash. Move the camera around a little bit to add some motion. Now hook up the nanoflash to an HDTV monitor, with the HDMI. Play the file back twice, once with the checkbox for "play progressive as psf" checked, and then again with the box unchecked. Do you not see added moire and line twitter in the displayed image for the psf output? Why? It's the same file.

Tom, I don't have access to a NanoFlash right now, but even if I did, I wouldn't need to do your test.

1) I see no reason to shoot psf in the first place, so there's no reason to experiment with it.

2) I never have any reason to playback footage from a Nano, so how that looks direct to an HDTV via HDMI is irrelevant to me no matter what the playback settings are. A more important test is to put the footage through your actual workflow (whatever that is) and then judge the output at the other end. In my workflow, it looks fantastic.

3) 99.99% of everything I have shot since the spring of 2006 has been 30P. If there was a problem with twitter, moire, or anything else in post or final delivery, I would have seen it by now.

4) The original poster was asking for opinions about 30P vs. 60i. There is no question in my mind that 30P is superior for many reasons.

http://www.vortexmedia.com/DVD1.html

[Alister Chapman]

Here are my reasons why I no longer shoot interlace for my own projects, I primarily shoot 25p:

P standard converts without messing up the field order.

Most TV's are using progressive LCD panels.

Computers are progressive.

Web and Internet streaming is normally progressive.

You can't accidentally screw up the field order in the edit.

Most editing is done on computers, with progressive monitors, so if you do screw up the field order you won't see it.

Frame grabs look better.

Accidentally recording interlace on a NanoFlash with the PsF in box ticked is a nightmare to correct.

P keys so much better

4:2:0 Progressive has symmetrical colour space which is almost impossible to tell from 4:2:2. 4:2:0 Interlace colour can look terrible.