GH1 - Frame Rate Conversion Tests - 720 60p & 30p to 24p vs. HPX 170 1080 24p

Yeah the way around it is through method 1. Compressor alone, or 2. Cinema tools and then compressor or 3. Cinema tools and FCP. You get glass smooth 60p to 24p any of those three separate ways. Only dropping right into the timeline with 60p give you stutter.

Still a little confused by this workflow, probably because I don't have my GH1 yet and the test footage I have is difficult to make out differences because there is not much movement other than pans. I personally don't mind the stutter, and I think it will be unnoticable on most shots, except for ones like you did in your test, where a large object moves completely horizonitally, and we see that frames are being played in a weird pattern (3:2?)

EDIT: I'm talking about 60p to 24p here.
 
Jack, thanks for all your hard work. I'm a bit new at this 1080p 24 stuff, so I'm a bit confused. You mention "This problem is significantly reduced in 60p AVCHD making it suitable for most applications where 1080 24p might fail, and non existent in 720 30p MJPEG an intra frame codec".

I guess where I'm confused is your reference to 60p. I know that the GH1 doesn't have a 1080 60p option, so are you talking about dropping a 1080 24p clip into an editing project set up as 1080 60p? If that's the case, are you saying that this procedure doesn't require reverse pulldown (which I still can't find in my Edius Pro program)?
You're correct about no 1080 60p.
The modes are:
1080 24p AVCHD
720 60p AVCHD
720 30p AVCHD

Your goal is a 24p project. Whether that project is 1080 or 720 is up to you. There's an argument to be made for each:

1080 24p TIMELINE ARGUMENT:
I've shot some 1080 material and I don't want to downrez that and I'm going out to blue ray or film or even down converting to SD I want to start with as much possible resolution. Besides I need to be able to tell my clients, sales agents, distributors that I'm delivering 1080 HD.

720 24p TIMELINE ARGUMENT: Why would I want to uprez my 720 footage to 1080 only to downrez it again. That has to create some kind of artifacts with all that digital manipulation. Besides I'm only going out to standard def DVD (blue ray burners are expensive) and 720 is plenty of resolution. I've seen stuff projeced from an SD DVD player in a movie theater and it looked great.

Now to the timeline setting question:
You want a 24p timeline. So any drag and drop approaches are dragging onto a 24p timeline.

60p and 30p are not really devised for reverse telecine but you can put them through reverse telecine and it will delete frames. 60i is.

If you drop them on a 24p timeline it does the exact same thing - delets frames to get to 24p. The result i somewhat choppy with 60p and very choppy with 30p.

So if you do reverse telecine with 60p or 30p material your kind of applying a process that was not designed to work.

The "Conform" then "Retime" method gets you 24p without dropping frames, but by changing the duration of the frames. First you conform, now all our 60p frames have grown longer and only 24 of them occur once a second. Now you have slow motion. So now you retime. You speed things back up and the software reinterprets the frames 24fps to a faster 24fps.

Make sense?
 
Still a little confused by this workflow, probably because I don't have my GH1 yet and the test footage I have is difficult to make out differences because there is not much movement other than pans. I personally don't mind the stutter, and I think it will be unnoticable on most shots, except for ones like you did in your test, where a large object moves completely horizonitally, and we see that frames are being played in a weird pattern (3:2?)

EDIT: I'm talking about 60p to 24p here.
You may be right about it not being hat noticeable. To me it is not noticeable in the GH1 Action Short / Test in my signature banner but other's claim it is. And that's where I just threw everthing on a 24 timeline with no conversion.

The point here is to know the options and that's why the test was shot the way it was, almost the Yin to the GH1's Action short's Yang. This one was set up as a controlled experiment to really know what was going on. The Action short was about what you could get away with. So again, it's about knowing the options.

Using Cinema tools and compressor is pretty perfect. So if you want to have 24p that would look like you shot it with real 24p go that route. If the workflow seems daunting and you never notice a difference, just drop the 60p in your timeline and get on with being creative. But when you come to that tracking shot you want to be oh so smooth, then you will probably want the options of the "conform then retime" methods rather than drag and drop telecine. If you were editing Bourne Identity, drag and drop might be fine. If you have steady cam footage following Danny on his trike in the SHINING, then you would probably want the "conform and retime" over "drag and drop", or the glass smooth effect might be lost.

I don't want to make it sound like just dropping 60p footage into a 24p timeline is totally the way to go, but for some it may be fine.
 
It stays with it. Though you make a copy of the original that you can swap out the old audio for the new. Not sure why, just that I've seen this step recommended in some other workflows. Maybe there's some voodoo or subtle artifacting that may occur from slowing down your audio and speeding it up again? Dunno. But it stays in sync and with with it.
 
I thought it was the other way around. The sky looks blue in the GH-1 footage where as the HPX seemed white and blown out a little.

Then again...in the GH-1 clips when you look at the pickup truck...or anything that is supposed to be white...they have some sort of blueish tint to it.

Thanks for this test JDS.


Yeah I was Pxl-peeping and the GH1 is considerably sharper; it handles the highlight edges like a DSLR instead of a camcorder. The big dense sensor playing the role here. The house in the center frame a few hundred yards away is soft in the hpx; the Gh1 still has detail and has way less edge softness like the HPX.


.
 
Last edited:
Need some advice. I don't want to upgrade my older FCP at this moment since im not using it for business curently. I have the latest imovie, however never used it before cause I have FCP. Whenever my GH1 arrives, I would like to shoot in the 1080/24 mode however what can I use that is the best for extraction of the 1080/24 avchd. Then after extraction I need to make high quality 1080/24 M-jpeg files of those clips. Neoscene has been mentioned, I just don't want to assume that is the best option/ best quality.

So to summarize my planned process.

GH1 1080/24-> imovie capture/Neoscene or? -> Conversion of those files to M-Jpeg at 1080/24..

Am I missing any of the needed parts

BTW this is a bithchin thread, very informative and enlightening Jack!

.
 
Need some advice. I don't want to upgrade my older FCP at this moment since im not using it for business curently. I have the latest imovie, however never used it before cause I have FCP. Whenever my GH1 arrives, I would like to shoot in the 1080/24 mode however what can I use that is the best for extraction of the 1080/24 avchd. Then after extraction I need to make high quality 1080/24 M-jpeg files of those clips. Neoscene has been mentioned, I just don't want to assume that is the best option/ best quality.

So to summarize my planned process.

GH1 1080/24-> imovie capture/Neoscene or? -> Conversion of those files to M-Jpeg at 1080/24..

Am I missing any of the needed parts

BTW this is a bithchin thread, very informative and enlightening Jack!

.

Voltaic or Neoscene are necessary for removing pulldown from 108024p AND properly reconstructing the interlaced chroma back to progresive.

Neoscene does one of 2 things.
1) remove pulldown and reconstruct chroma from your 108024p footage and convert the MTS files to either ProRes or Cineform.
2) convert your 60p AVCHD to Pro Res or Cineform (leaving it at 60p and and native resolution.

It doesn't do anything to the MJPEGs

The MJPEGS already come in a nice little QT wrapper from the camera. I convert them in compressor to ProRes or Quicktime.

So you need Voltaic or Neoscene for the 108024 pulldown and chroma reconstruction (can't remeber where JES deinterlacer rates, I think you can do the same but need other FCP Suite software working with it)

EDIT: I see you want to make MJPEG's not convert them. NeoScene won't do that. Only Pro Res or NeoScene QT's.

Voltaic will do it.
 
EDIT: I see you want to make MJPEG's not convert them. NeoScene won't do that. Only Pro Res or NeoScene QT's.

Voltaic will do it.

I could export the cineframe file to M-Jpeg after?

I have never heard of Voltaic. How does it compare to NeoScene. I like that Neoscene does chroma correction, that sounds cool..



BTW, just DL'ed the file. It's crazy how the 720 gh1 is better than the HPX 170 @ 1080. It's almost evil wrong that a 1499 now gets you a core image jump like this and with removable endless optical choices. So evil that it taste sweet....... And this is only V.1

.
 
I could export the cineframe file to M-Jpeg after?

I have never heard of Voltaic. How does it compare to NeoScene. I like that Neoscene does chroma correction, that sounds cool..



BTW, just DL'ed the file. It's crazy how the 720 gh1 is better than the HPX 170 @ 1080. It's almost evil wrong that a 1499 now gets you a core image jump like this and with removable endless optical choices. So evil that it taste sweet....... And this is only V.1

.

Yes if only we could get B frames now :). If they put B frames in via firmware or in the GH1a next year, then it will be evil indeed. We're guessing that because encoding AVCHD is intensive, they got rid of B frames and made it easier on the processor by having only an I frames followed by Predictive frames.

From an email from Barry Green "So a normal long-GoP structure might look like IBBBPBBBPBBB, whereas with the GH1 there's no "B" frames so it's all IPPPPPPPPPP."

For Voltaic look at Isaac Brody's stickies in this section. It repairs the chroma as well and costs $39 vs. Cineform's $139 (I think). Cineform is a codec and so is MJPEG, so you can't have a Cineform MJPEG, you could output Cineform and then convert that to an MJPEG. But Cineform itsef only outputs two flavors. ProRes or Cineform. It's advanatage over Voltaic is that you get an awesome 422 codec that (I think) is better than ProRes and has smaller file sizes. Voltaic's advantage is more versatility. You can output to whatever you want.
 
BTW, I'm going to go ahead an coin this right now...
Cam nickname = the "No 'B' H-1"
(you know for lack of B frames get it ... get it?)
Hey - I’m here all week. Try the veal. Don’t forget to tip your servers.
 
How about naming it the "No 'G' H-1" because none of us have our freaking cameras yet...*grouse grouse grouse*
 
I can semi-sympathize. Barry Green and I are sharing a camera and he has it right now for official lab testing stuff. I miss it.

6a00d8341c94c853ef0105362dd0a2970b-800wi
 
Yes if only we could get B frames now :). If they put B frames in via firmware or in the GH1a next year, then it will be evil indeed. We're guessing that because encoding AVCHD is intensive, they got rid of B frames and made it easier on the processor by having only an I frames followed by Predictive frames.

From an email from Barry Green "So a normal long-GoP structure might look like IBBBPBBBPBBB, whereas with the GH1 there's no "B" frames so it's all IPPPPPPPPPP."

For Voltaic look at Isaac Brody's stickies in this section. It repairs the chroma as well and costs $39 vs. Cineform's $139 (I think). Cineform is a codec and so is MJPEG, so you can't have a Cineform MJPEG, you could output Cineform and then convert that to an MJPEG. But Cineform itsef only outputs two flavors. ProRes or Cineform. It's advanatage over Voltaic is that you get an awesome 422 codec that (I think) is better than ProRes and has smaller file sizes. Voltaic's advantage is more versatility. You can output to whatever you want.



Ok not evil good, however good seems more applicable....... Voltaic sounds interesting ( also sounds like a Harry Potter character ) I will download the trial.

Oh no, not that crying native American..... He makes me want to go and pick up my trash! ;-/


.
 
Voltaic or Neoscene are necessary for removing pulldown from 108024p AND properly reconstructing the interlaced chroma back to progresive.

Neoscene does one of 2 things.
1) remove pulldown and reconstruct chroma from your 108024p footage and convert the MTS files to either ProRes or Cineform.
2) convert your 60p AVCHD to Pro Res or Cineform (leaving it at 60p and and native resolution.

It doesn't do anything to the MJPEGs

The MJPEGS already come in a nice little QT wrapper from the camera. I convert them in compressor to ProRes or Quicktime.

So you need Voltaic or Neoscene for the 108024 pulldown and chroma reconstruction (can't remeber where JES deinterlacer rates, I think you can do the same but need other FCP Suite software working with it)

EDIT: I see you want to make MJPEG's not convert them. NeoScene won't do that. Only Pro Res or NeoScene QT's.

Voltaic will do it.

First, Jack... WOW... thank you very much for taking the time to do these awesome tests. It's really going to help everyone with their workflow.

So ... this is where I was confused. Are you saying that for 720p footage from the GH1, to get to 24p, we can't use Neoscene? And that's why you're using Cinema Tools/Compressor/FCP?

And then for 1080p footage we can't use Cinema Tools/Compressor/FCP? This might be where I'm confused, because I don't understand why you can't do Log & Transfer with the 1080p media, then put that media (Pro Res) through Compressor or Cinema Tools to remove the pulldown. Or is this not necessary? How much of a difference would we see between ProRes and Cineform? Are there any examples anywhere?

So basically, there's not a one application solution if you're shooting in 720p and 1080p?

Between Methods 1 and 2... you say they are both very high quality for 60p -> 24p. Is one better than the other quality-wise?
 
WHEW! All tutorials up. It's late and I'm tired and there's likely some mistakes there. If I change anything major, I'll also make a new post pointing it out so don't take them as in stone today and subscribe to the thread so you'll know if I post correcting some crucial step a week, a month or a year down the line.


First, Jack... WOW... thank you very much for taking the time to do these awesome tests. It's really going to help everyone with their workflow.

So ... this is where I was confused. Are you saying that for 720p footage from the GH1, to get to 24p, we can't use Neoscene? And that's why you're using Cinema Tools/Compressor/FCP?

And then for 1080p footage we can't use Cinema Tools/Compressor/FCP? This might be where I'm confused, because I don't understand why you can't do Log & Transfer with the 1080p media, then put that media (Pro Res) through Compressor or Cinema Tools to remove the pulldown. Or is this not necessary? How much of a difference would we see between ProRes and Cineform? Are there any examples anywhere?

So basically, there's not a one application solution if you're shooting in 720p and 1080p?

Between Methods 1 and 2... you say they are both very high quality for 60p -> 24p. Is one better than the other quality-wise?

Mike I can answer some of that but Isaac would be better suited to answer most of it. Let me see if I can get his attention for you.
 
Very nice tutorials Jack ! even though I am PAL and dont need to do most of them, its so nice to see this step by step description you so tediously made!! :beer:
 
Back
Top