can FCP properly interlace progressive material?

I wish to convert a 50p clip to 50i. The proper conversion takes the first field from the first frame, the second field from the second frame, etc. I.e., it simulates what a 50i camera would have shot instead of the 50p camera.

When the 50p clip is dropped into an empty 50i sequence FCP asks whether it should change sequence settings to match the clip settings. I say "No". The sequence can then be rendered "Full" or not, but this doesn't affect the result. The result is a 50i sequence in which the first field comes from the first frame and the second field also comes from the first frame. The third field comes from the third frame and the fourth field also comes from the third frame. Etc. FCP does an improper conversion which completely ignores every other frame of the original 50p.

Compressor can do the proper conversion of the 50p clip to 50i. In Video Settings make Frame Rate 25 fps; turn on Frame Controls and make Output Fields top first or bottom first.

(Maybe I'm doing something wrong today in FCP because I recall succeeding with the sequence settings method in the past.)

Dennis Couzin
Berlin, Germany

Nope, sorry D., in my experience FCP itself has never done this correctly (what it does is exactly as you describe, drops every other frame before interlacing the remaining frames) ... so you either have to preprocess the double frame rate progressive clips first (i.e. convert to interlaced in Compressor or similar) or edit the whole thing progressive first and do the interlace conversion in Compressor on export.

Cheers
Andy

Andy, thanks for confirming.

Dennis Couzin
Berlin, Germany

Andy, I have a slightly different situation, but very similar question...I have recently started shooting with Canon DSLRs, and this is my first experience with progressive material (1080p30).

I have done a bit of research and reading, and I think I'm starting to grasp these concepts, but am still a little uncertain.

To make progressive material look properly on a TV (LCD or CRT) when played from a DVD, it needs to be converted to interlaced, correct?

I can transcode to ProRes and edit the project and export to a self-contained QuickTime file, run Compressor's Best Quality for DVD preset, burn the DVD and it still is (or at least appears to be) progressive...it has a bit of a jittery look to it.

You had mentioned doing the interlace conversion in Compressor. If my hypothesis is correct, if I did this, everything would look much better, right?

Can you explain more about how to properly do the interlace conversion in compressor?

Thanks!
Casey

Casey, you have nothing to interlace since you have only taken 30 images per second. If you do a 50% speed conversion with e.g. Motion you'd have 60 images per second and then you could make 60i using Compressor as described.

Otherwise you can achieve no more than what is called "bogus" interlacing in another strand. Please note that in that strand Apple terminology was used, calling 60i '30i'.

Dennis Couzin
Berlin, Germany

You do not need to convert progressive footage to interlaced to display on a TV.

Hmmm...okay....

So is my workflow of transcoding my DSLR footage (1080p30) to ProRes, editing the project and exporting to a self-contained QuickTime file, running Compressor's Best Quality for DVD preset, then burning the DVD...is that pretty typical? I know some people are shooting at 24...I've tried that with the same workflow I described, and it looks worse.

I keep coming back to the season finale of House a while back, and watching that on both HD and SD TV sets...my footage looks nothing like that...it looks like "movie mode" when I had my old Canon GL1...just an artificial effect, not smooth and sharp video like I saw on House. What are they doing that I'm not? I'm assuming they're using better software than Compressor?

It just seems like I'm missing a step somewhere, but I don't know where.

Thanks for your continued help!
Casey

Well for one thing the DP of House did a ton of tests before they ever committed to doing the episode. For another, I'm guessing it was all hardware compression that was used to convert the Canon footage to something useable for edit.

1080p30 is going to have issue when converted for interlacing if you use something like Compressor, which is a decent tool but not the best. So many people forget that it takes 60 FIELDS of information to create a 30 frame per second interlaced image. So your 30 FRAMES of progressive material has to be split to create the 60 FIELDS of information required for interlacing.

So in effect, Compressor has to create the second field of each frame interpolating the information from the first field. It doesn't do a great job of this. Telestream Episode is much better and the AJA Kona boards are better yet.

This is why, and I get hammered for the all time for being "incorrect for narrative," I love and shoot 720p / 60 as much as possible. Makes an absolutely seamless conversion to interlaced even using Compressor because there are truly 60 FRAMES of information there. Each progressive frame becomes a single field, nothing has to be interpolated or created from nothing. So those 60 frames of information are simply converted to 60 fields which make up the 30 frames needed for interlaced video playback.

FYI, when that episode of House originally aired we had multiple episodes on our DVR to compare that episode with. While it was extremely good on its own, there was a clear difference in the overall look and quality of that episode compared to the regular episodes. One thing that really stuck out what how dark the Canon episode was. Looked like they picked that particular storyline because it took place at night with a lot of shadows where the blacks could be completely crushed. The colors were much more muted in that episode as well. Overall a very very good looking show, but the shows they shoot normally look much better.

Walter Biscardi, Jr.
Biscardi Creative Media
biscardicreative.com

Was House aired interlaced, or interlaced with a pulldown? My guess is that they shot 23.976 and added a pulldown.

I don't think you can create a proper 60i signal (perhaps a 30 psF signal) from a 30p signal off any capture card. You can add a pulldown on a 23.976 signal, or you can run it through Compressor/Twixtor or the Alchemist to re-interlace the signal using optical flow. Optical Flow, when properly implemented looks smoother than adding pulldown.



www.strypesinpost.com

> What are they doing that I'm not? I'm assuming they're using better software
> than Compressor?

Unlike most indie-level DSLR videographers, the DP of House, M.D., Gale Tattersall, made the decision on the camera not based on the fact that he owned one, but based on the needs of the script, well ahead of production. They researched the camera, did tests, noted the compression issues, and lit and shot the scenes based on what they knew. They probably also engaged the services of numerous experts in production and post-production well before shooting. If they'd waited until the software conversion level to address the shortcomings, they probably would have had similar problems.

I'm also pretty sure they didn't shoot on 30p.

> Looked like they picked that particular storyline because it took place at night with a
> lot of shadows where the blacks could be completely crushed. The colors were much more
> muted in that episode as well.

There was a Popular Photography article about this. They didn't pick the 5D for its imagery, but for its ability to get into that cramped space. The chaos of the situation also lent itself to a grungier look.

Check out early (Seasons 1-2) episodes of House, M.D. sometime. The look was actually far more distinctive than it is now. The show's cinematography used to be super-bright, super-saturated, and extremely distinctive especially for a medical show (except the pilot, which had a gave all the characters a bizarre "boiled-egg jaundice" look). They got more show-offy starting in Season 3 (heavy filtering and playing with saturation and contrast), but far more generic...and now sometimes it looks like you're watching six different shows as you go through a season. There's no unity anymore.

> This is why, and I get hammered for the all time for being "incorrect for narrative," I love
> and shoot 720p / 60 as much as possible.

In my experience, it's a lot easier and better-looking to simulate 24p from 60i source than the other way around. Extracting extra existing frames rather than trying to add in ones that don't exist. So when in doubt, I'd be inclined to go 60i.


www.derekmok.com

Thanks guys...this is starting to make sense to me!

So basically what I'm hearing is that you can't get a good 60i signal from 30p, but you can if you shoot 24p and use pulldown, correct? I have never used Twixtor or Alchemist...and again there, you are talking about using the 24p, not 30p, right?

All I'm just trying to do is get my DSLR footage to look as good as possible on a SD DVD

Thanks!!
Casey

A pulldown isn't 60i. It is just 24p with a pulldown. Looks great if the cam op knows what he is doing.



www.strypesinpost.com

Ok....so a better way to understand it....24p with pulldown simulates 60i?

Casey

It just repeats fields and frames in a certain pattern so you can float a 23.98 fps picture in a 29.97 interlaced stream. It does not mimic 60 temporal samples a second. But it looks pretty close to 24p.



www.strypesinpost.com

Okay...getting foggy again...so in layman's terms, the purpose of doing pulldown from footage from a Canon 7D using 1080p24, is simply to make it look better when played at 29.97, say on a DVD, right?

> the purpose of doing pulldown from footage from a Canon 7D using 1080p24

"Adding" pulldown. Not "doing" pulldown.

NTSC videotape only ran at 29.97fps, nothing else. Pulldown is the way you put source material that had been shot at 24fps ("true 24", originally from a film source) onto a medium that runs at 29.97fps (NTSC tape).

[www.zerocut.com]

30p is a new monster made possible by digital camcorders, which are not confined to the regular frame rates used in the film industry. That's why it causes us so many headaches.

If you shot at 24p (23.98fps, unless you're dealing with a format like 35mm or RED, which shoot "true 24", there is no reason at any point in your post-production chain to add pulldown to make 29.97fps, unless you need to go out to NTSC videotape. And even then, you don't convert to 29.97fps until the very end. I don't know where you got the idea that 24p converted to 29.97fps "looks better" than 24p footage burned directly to video DVD, but it's not true. Are you sure you're not just unfamiliar with the way 24p motion looks?


www.derekmok.com

Yea, what Derek said. Don't add a pulldown unless you are going out to broadcast and it is part of the specs. Also, don't use 24p to refer to 23.98. It's confusing as hell.



www.strypesinpost.com

I gotta get hip to the lingo!

>Are you sure you're not just unfamiliar with the way 24p motion looks?
Well, that's entirely possible

I think I need to regroup my thoughts and think about what it is I am trying to figure out!

I wish there was someone local (Minneapolis) I could simply invite over to my studio and show them exactly what I am seeing.

Thanks for your feedback and patience!!!!
Casey

These guys are some of the best and they're in your neck of the woods.

[splice.tv]

Walter Biscardi, Jr.
Biscardi Creative Media
biscardicreative.com

Thanks!!

> Also, don't use 24p to refer to 23.98. It's confusing as hell.

Everybody does it...which is why when you talk tech with other people -- camera guys, assistant editors, producers -- always clarify which one you mean. And just to be even more fun, if you want to be literal and accurate, 23.98fps is actually 23.976fps. The more accurate number comes into play when, for example, you're exporting image sequences.


www.derekmok.com

Casey, I can't really add much more to the expert opinions here, these guys all know their stuff.

I just to want to reiterate though, since I'm also a Canon HDSLR shooter (7D) that the "look" you're going for is probably 24p (actually 23.98 or 23.976 fps, but I'll say 24p for now), which has a filmic quality that mimics the frame rate that actual film is shot and projected at (a true 24 fps). So try shooting at 1080p24 on the Canon and see how you like it. And be sure to set your shutter speed to 1/50 so you get smooth motion without any high-speed shutter artifacts.

And again, there's no need to do any pulldowns, conversions or other craziness if you're just burning to DVD. DVDs can playback 24p (23.98) material just fine, there's no need to convert it to 29.97.

_______________________________________
SCQT! Self-contained QuickTime ? pass it on!

>Everybody does it...which is why when you talk tech with other people -- camera guys,
>assistant editors, producers -- always clarify which one you mean.

Maybe in the NTSC world. Both 24 and 23.976 are formats on their own. Hdcam SR does both 24 and 23.98. And I get file formats that are in 24, which I have to slowdown to 23.98 and add a pulldown so I can cut it next to 60i stuff. If I tell the tape op to black a tape to 24 when I mean 23.98, he'll give me 24fps and I will have problems going to tape.



www.strypesinpost.com

> Maybe in the NTSC world. Both 24 and 23.976 are formats on their own. Hdcam SR does
> both 24 and 23.98. And I get file formats that are in 24, which I have to slowdown to 23.98
> and add a pulldown so I can cut it next to 60i stuff.

Tech personnel are usually a little more precise. The problem comes when we get a project direct from a producer or director who didn't check this stuff or is loose with the terminology, and just tells us "24p". It's still staggering to me how many directors and producers don't lock down on a spec as important as the frame rate before shooting; it shouldn't even be a question to any key personnel by the time the first shot is taken, let alone by beginning of post-production. And then there's those "prosumer" cameras that are even looser with the terminology, for example the Canon Vixia camcorders like the HF S10 grafixjoe and I own. It claims to do 24p, has the setting right in the camera and on the documentation, but turns out to be a consumer-grade 24p simulation that yields 29.97fps footage. I've had a director/DP team try to mix that with an A camera that was a Panasonic HVX200, 1280x720 footage shot at 23.98fps.


www.derekmok.com

And that is what camera tests are for. Kinda shocking that it is never highly regarded. Without a test, nobody should even start a shoot.

With some cameras, they refer to 24fps as 24pN. And 24p means 24fps with standard pulldown. Then there is 24pA, which is 24fps with "advanced pulldown", which means you need to remove the pulldown on capture/ingest.



www.strypesinpost.com

> With some cameras, they refer to 24fps as 24pN.

That's a Panasonic thing. If I remember correctly, Panasonic refers to 24p as 23.98fps footage, but with 2:3 pulldown frames added. 24pN is "actually" 23.98fps, and so you also get more recording time.


www.derekmok.com

Ok guys.. how's it going.. new to the forum but this topic right here is exactly what I was searching for... Recently shot a project on AF100 some 7D that was edited in FCP using ProRes 23.98. The project is going to broadcast so I need to dump to HDCAM and Digibeta for two separate networks. I need to create broadcast master files that are 29.97 interlaced in both HD and SD..in short whats the best way to accomplish this through compressor. What settings should I be adjusting to get the best possible output. I know that going out through a card such as AJA/Kona/Blackmagic and allowing the hardware to do the conversion is best.. but thats not currently an option so compressor is it for me... any advice would be appreciated thanks...

Also is anyone familiar with closed captioning.. I have questions regarding that as well.. but willing to discuss in a separate forum.

Thanks in advance

Oh yeah.. whats the story with 59.94? is that even an option for broadcast delivery?

Not really. When you are adding/removing pulldown, there is no difference between doing it via software or hardware, unless you are working with compresses formats, then you may encounter a generation loss on whatever codec you are using. Of course, I mean you add a standard pulldown, not the default 2:2:2:4 pulldown in FCP. But truth is, you don't have to worry about it. The IO device you will use to interface with the decks should be able to add a pulldown for you on the way out to tape. I'm not sure how deck control works when you edit to tape with different timecode, but I believe that is handled by the card. Others who are in NTSC land will chime in.

Regarding CC, again, my knowledge is theoretical, as I don't work in north America. You would either send your stuff to the captioning house and have them do the CC, or invest in a CC encoding device such as maccaption:

[www.cpcweb.com]


59.94? Sure. 1080i59.94, or 720p60 (aka 59.94 fps), depending on the network.



www.strypesinpost.com

Hey strypes... thanks for the response... but more specifically I was wondering what the best settings are for compressor to create a broadcast quality master 1080i 29.97 file from the original 1080 24p file. My understanding is that the re-interlacing of the file can introduce jitters, etc. So I'm just trying to figure out the best way to export a clean conversion.

Thanks for clarifying 59.94.. it's the progressive frame much like 23.98.. thanks