JVC GY-HD100 720 True 24P or Sony HVR-Z1U 1080 60i converted to 24P?

[Geoff Murillo]

Hi Everyone,

I've been holding off from purchasing an HDV camera for a while, but I'm getting close to buying one. Even though they both make me drool, I don't think I will be able to afford a Canon XL H1 (cost of camera) or a Panasonic HVX-200 (cost of camera and P2 storage), so I'm trying to decide between the Sony HVR-Z1U or the JVC GY-HD100.

I want the camera to shoot short films and a low-budget feature (potential film-out). So, I've been researching both 720 24P (from the JVC) and converting 1080 60i (from the Sony) to 24P via DVFilm Maker.

My question is, if I'm going to film-out a project, which camera would be better?

The JVC footage is true 24P, however it's recorded at 1280X720.

The Sony records 1440X1080 with anamorphic pixels, but if I converted this footage to 24P, would the interpolation within DVFilm Maker soften the image, negating the advantage of shooting with the higher resolution?

Has anyone seen side by side comparissons? (720 True 24P vs. 1080 60i converted to 24P)

Any information would be greatly appreciated.

-Geoff Murillo

[Dylan Pank]

Without doing a side by side comparison, I would vote for the JVC in theory as

• shorter GOP and the Lower res means fewer artefacts.
• the Z1 is 60i, which risks adding more compression artefacts motion on detail that's essentially wasted.
• it's a much simpler workflow, if you're really set on 24p you're going directly there, no interlacing/slowing down footage.
• for the HD100 you could rent much better HD lenses than the stock lens and really get the best out of the 1280*720p resolution.
• a true manual lens, which I feel is always important for theatrical work.

The best solution would be to get your hands on the cams yourself and shoot tests, where you can tailor the conditions to your own situation (eg are you shooting in low key/high key light, fast action, or trying to get a particular look?) Talk to the lab that will do the blow up and get their opinion (DVfilm might be willing to answer some queries, they have a FAQ list on their site but it's a little out of date.)

[Jon Fordham]

* the Z1 is 60i, which risks adding more compression artefacts motion on detail that's essentially wasted.*

The Z1 can be switched to 50i or CF25 which would eliminate the need for pulldown software. And depending on the route chosen, preserve resolution and/or eliminate interlacing artifacts.

* it's a much simpler workflow, if you're really set on 24p you're going directly there, no interlacing/slowing down footage.*

I'm not an editor, but I am not aware of any plug and play NLE solutions that can handle native 720/24P HDV. To the best of my knowledge, 720/30P is the only native HDV setup on most, if not all, desktop NLE's. I've been shooting a demo with the HD100 for the last few days and know for a fact that you can't cut natively with FCP5.

* for the HD100 you could rent much better HD lenses than the stock lens and really get the best out of the 1280*720p resolution.*

Don't be so quick to think you have many HD lens options. I'm not aware of any other 1/3" or 1/2" HD lenses other than what Fuji is offering.

* a true manual lens*

Always preferable.

If you're not going to use the 50i switchability of the Z1, then the FX1 gives you the exact same camera and lens for alot less. But then you'd have to go the 60i to 24fps software pulldown route.

[Douglas Spotted Eagle]

Originally Posted by Jon Fordham

*

I'm not an editor, but I am not aware of any plug and play NLE solutions that can handle native 720/24P HDV.
.

Sony Vegas would be the one....

[Barry Green]

Verified, Vegas does indeed play 720/24p HDV files. It also can play the Canon 1080/24F files.

[Thomas Smet]

The only real advantage you get from down converting 1080i to 720p is being able to actually create a slightly lower resolution 720p video at a true 50 or 60 fps.

Current 720p HDV equipment can only record up to 30p which is great but for some people used to interlaced video they find 30p to jerky and not as smooth.

Converting 1080i to 720p with the right tools will give you 1440 x 540 x 60. You then scale that to 1280x720x60. So you will have a slightly softer 720p due to starting with 540 pixels instead of the full 720p. So really you will have 720p video at 60p with about the same amount of detail as the SONY Cineframe modes since only only use one field worth of detail. Of course it is actually less when you take into consideration interlace filtering. You may only really have 410 lines of detail with this method but the horizontal resolution would be a full 1280 and you would have 60 frames per second.

For all the mess however I would prefer to deal with the 30p of a real 720p.

[Jon Fordham]

Barry & Spot,

I've been shooting with the HD100 for a few days now, and I'm still at a loss for how it is recording the 24P to tape when recording 720 HDV. The manual states that a 3:2 pulldown is used to covert the 24P to "60 frames". But it doesn't specify whether or not the recording is interlaced or Progressive. Any ideas?

Vegas...

Here's a some questions:

Can Vegas detect whether or not the recording is i or P?

Is the preset in Vegas actually 720/24P or 720/60?

When the footage comes into Vegas what needs to be done to work with the footage on a 23.98 timeline?

Does the Vegas 720 preset keep the footage in Native HDV when coming in through Firewire, or convert it to/from something else? In other words, is the preset actually an HDV preset, or just a generic 720/24P preset?

More importantly, have either of you actually captured 720/24P footage from the HD100 via Firewire into Vegas and cut it? How was the workflow? Any eccentricities? How did you get it back out of Vegas and onto what format?

[Douglas Spotted Eagle]

Originally Posted by Jon Fordham

Here's a some questions:

Can Vegas detect whether or not the recording is i or P?
No. It just sees it as a stream. But you set Vegas to a Progressive project, it handles the media as just that. However, interlaced footage dropped into a progressive project simply has either blending or interpolation applied to create a progressive stream.

Is the preset in Vegas actually 720/24P or 720/60?
720/24p, but you have to create that preset on your own, as the JVC wasn't available when Vegas 6c was built. However, creating a preset takes about 4 seconds.

When the footage comes into Vegas what needs to be done to work with the footage on a 23.98 timeline?
Either have "Remove Pulldown" checked in the Prefs, or right click the footage and in the properties instruct it to remove.

Does the Vegas 720 preset keep the footage in Native HDV when coming in through Firewire, or convert it to/from something else? In other words, is the preset actually an HDV preset, or just a generic 720/24P preset?
On capture, Vegas doesn't touch the stream. It comes into Vegas as an m2t file, unless you use the CineForm capture and/or conversion tool or use the VASST GearShift conversion tool

More importantly, have either of you actually captured 720/24P footage from the HD100 via Firewire into Vegas and cut it? How was the workflow? Any eccentricities? How did you get it back out of Vegas and onto what format?

My experience with capturing HD100 media is limited to 2 hours of media shot at both WEVA and at Sydney's Pittswater harbor. Working with it is no different than any other HDV or 24p media, IMO. I DID convert all the Sydney footage to CineForm, and of course gained a 100%, always solid stream. However, on my dual-dual core, I get 30 fps out of m2t, as well. Just not all the time. The 720p is considerably smaller than the 1080 stream.
Barry's had a lot more quality time with the cam, I think, so he'd be a better judge of that side of things, I'm sure.

[Douglas Spotted Eagle]

Originally Posted by Thomas Smet

Converting 1080i to 720p with the right tools will give you 1440 x 540 x 60. You then scale that to 1280x720x60. So you will have a slightly softer 720p due to starting with 540 pixels instead of the full 720p. So really you will have 720p video at 60p with about the same amount of detail as the SONY Cineframe modes since only only use one field worth of detail. Of course it is actually less when you take into consideration interlace filtering.
For all the mess however I would prefer to deal with the 30p of a real 720p.

Thomas, how are you converting this, and have you actually done it? Your math is obviously correct, but math and what the eye see are often two different things. There are means/ways of doing this without doing two transcodes, I've done a LOT of it lately, as has Brian Mercer in Toronto. (He's probably forgotten more about scaling than most of us ever will know) We've both got great upsample and downsample schemes, and none of them involve doing two transcodes.
I'd be happy to make comparison pix if you'd like to find a piece of footage that serves as a baseline. It might be you've got a better way. I'm judging my footage on a Luma monitor and on a 12' projection screen.

[Luis Otero]

Remember, during the downconversion from 1080i to 720p you lose resolution. Based on every person that has posted here their opinion, that is their subjective opinion. On the other hand, going from 720p to 1080i is easy and you do not lose any resolution.

[Douglas Spotted Eagle]

Originally Posted by Luis Otero

Remember, during the downconversion from 1080i to 720p you lose resolution. Based on every person that has posted here their opinion, that is their subjective opinion. On the other hand, going from 720p to 1080i is easy and you do not lose any resolution.

of course you lose resolution. You don't lose numbers; you're making things bigger. You're changing out horizontal lines of resolution, (720 down to 540 if you're going to interlaced 1080) and you're shifting vertical lines of resolution as well. 1280 to either 1440 or 1920.
Not to mention that 720p is a square pixel format and 1080 i or p is anamorphic/non-square pixel format.
In the overall process, downscaling is less difficult than upscaling.

[Barry Green]

Originally Posted by Jon Fordham

I've been shooting with the HD100 for a few days now, and I'm still at a loss for how it is recording the 24P to tape when recording 720 HDV. The manual states that a 3:2 pulldown is used to covert the 24P to "60 frames". But it doesn't specify whether or not the recording is interlaced or Progressive. Any ideas?

There is nothing interlaced going on -- it's all progressive, all the time. The pulldown isn't pulling frames into fields, it's duplicating frames. Same thing the VariCam does when shooting 24p: each odd frame is recorded twice, each even frame is recorded three times. The result is a 60p data stream with the look of 24p (because there are only 24 individual, distinct frames displayed each second).

In the MPEG-2 data stream, yes it's a 60p file. The duplicate frames are handled with a simple "repeat" flag, which takes up no bandwidth -- so you get the compression efficiency of only storing 24 frames' worth of data, but you get the compatibility of it being a 60p file. Remember, 24P is *not* one of the HDV standards; just like 24P is not a DV standard. In DV 24p was implemented by carrying it within a 60i data stream; in HDV it's been implemented by carrying it within a 60P data stream.

Now, here's where it gets a little weird -- in editing, Vegas sees the file as 24p. Not as 60p. I haven't captured any HDV in Vegas yet (just upgraded a couple of days ago) but I've captured it using Pixela HD Capture and using HDV Rack. Both write .m2t files to the disk, and when you drop those .m2t's into the Vegas media bin, the properties show "23.976 fps". I don't know quite how to explain that, but it's really cool, and it continually impresses me how forward-thinking Vegas is.

Can Vegas detect whether or not the recording is i or P?

Most definitely.

Is the preset in Vegas actually 720/24P or 720/60?

There isn't an existing 720/24p HDV preset, but there is one for "HD 720/24p", and that works just fine with HDV files. Don't use the 720/60p preset, that will unnecessarily cause the program to have to duplicate/synthesize unnecessary frames. If you want 24p, just use the "HD 720/24p" preset.

When the footage comes into Vegas what needs to be done to work with the footage on a 23.98 timeline?

Not a thing. Just use the 720/24p preset, drop the files on the timeline and go.

[quote]Does the Vegas 720 preset keep the footage in Native HDV when coming in through Firewire, or convert it to/from something else? In other words, is the preset actually an HDV preset, or just a generic 720/24P preset?[quote]
Haven't captured through Vegas yet, so I defer to others to answer that.

More importantly, have either of you actually captured 720/24P footage from the HD100 via Firewire into Vegas and cut it? How was the workflow? Any eccentricities? How did you get it back out of Vegas and onto what format?

I've been cutting it for experimentation purposes, but none of it was captured into Vegas. I did find some oddness in Vegas 5; occasionally files would have black frames in them. Haven't noticed that in Vegas 6 yet, but I've literally only used 6 for like five minutes (just upgraded). The performance is pretty slow on my system, but then again, so's my system (2.66ghz single-CPU).

[Jon Fordham]

"720/24p, but you have to create that preset on your own, as the JVC wasn't available when Vegas 6c was built. However, creating a preset takes about 4 seconds. Either have "Remove Pulldown" checked in the Prefs, or right click the footage and in the properties instruct it to remove. "

Wait a minute... Now I'm confused. If Vegas doesn't touch the stream and can't detect whether or not the stream is Progressive or interlaced, then how does it know what to look for when it is capturing and why would you need to tell it to "remove pulldown" if it was told to capture a 24P stream?

Maybe this shows my ignorance in the world of post, or perhaps I'm too sheltered in my own little FCP universe, but it seems to me that you would need to tell Vegas that the HDV footage was 720/60 in order to capture it via Firewire in the first place. If you tell Vegas the footage is 720/24P, wouldn't it look to capture 24fps and not know that the 24P was pulled down to 60fps to record? I'm not at all talking about working with the files once it's captured. I can easily create my own custom presets in FCP too. But when it comes to capture, that's a different story.

So, does Vegas actually capture 720/24P natively via a Firewire connection, directly from the HD100's 6 pin 1394 port, or do you have to create some sort of unique setup to get the footage into the system?

[Douglas Spotted Eagle]

Originally Posted by Jon Fordham

"


Maybe this shows my ignorance in the world of post, or perhaps I'm too sheltered in my own little FCP universe, but it seems to me that you would need to tell Vegas that the HDV footage was 720/60 in order to capture it via Firewire in the first place. If you tell Vegas the footage is 720/24P, wouldn't it look to capture 24fps and not know that the 24P was pulled down to 60fps to record? I'm not at all talking about working with the files once it's captured.
So, does Vegas actually capture 720/24P natively via a Firewire connection, directly from the HD100's 6 pin 1394 port, or do you have to create some sort of unique setup to get the footage into the system?

Yes, Vegas captures video streams without any setup. Information about the stream is in the header file that accompanies the stream. Whether it's HDV, DVX/24p, etc...there is nothing to set up. It's merely a data transfer. Vegas currently doesn't capture the HD100, Barry is using DVRack, which does capture it correctly, as does the CineForm tools. Once captured, Vegas then allowed you to drop it on the timeline where it is correctly read. Barry is using the HD, not HDV preset, but overall it doesn't matter.
I'll be in NYC in a couple weeks...if you'd like, I'll be happy to take some time and show you the ins/outs of why many feel Vegas is a superior editor for HDV. :-)

[Thomas Smet]

Originally Posted by Douglas Spotted Eagle

Thomas, how are you converting this, and have you actually done it? Your math is obviously correct, but math and what the eye see are often two different things. There are means/ways of doing this without doing two transcodes, I've done a LOT of it lately, as has Brian Mercer in Toronto. (He's probably forgotten more about scaling than most of us ever will know) We've both got great upsample and downsample schemes, and none of them involve doing two transcodes.
I'd be happy to make comparison pix if you'd like to find a piece of footage that serves as a baseline. It might be you've got a better way. I'm judging my footage on a Luma monitor and on a 12' projection screen.

I have been writing a program to convert 1080i video into 720p 60p video. It works pretty well but it currently cannot write mpeg2 so you have to convert to another format. The sad thing right now is that for the one market that would want this the most (weddings) would have no logical purpose to use it right now because it is very slow. I managed to get the rendering faster but still this is almost useless right now for hours worth of footage. I'm also not even sure if I recomend doing this anymore anyways since you would end up encoding your video multiple times. Who knows maybe Cineform will add this feature in the future so this is done in the background inside your NLE.