DV Info Net

DV Info Net (https://www.dvinfo.net/forum/)
-   JVC GY-HD Series Camera Systems (https://www.dvinfo.net/forum/jvc-gy-hd-series-camera-systems/)
-   -   HD101 24p capture (https://www.dvinfo.net/forum/jvc-gy-hd-series-camera-systems/72184-hd101-24p-capture.html)

Nima Taheri July 24th, 2006 02:04 PM

HD101 24p capture
 
I shot two videos that I'm trying to capture and hopefully edit in AVXPro.

Video is shot in 720/24p.

I tried Liquid 7 logging tool, but I keep getting a "!" sign on the clip. So I tried to capture with CapDVHS, convert it with MPEG Streamclip (to uncompressed dv QT) and import it into AVXPro. This is my first trying to edit 24p, I usually edit 25/50, so this is new to me. When capturing the footage with DVHS, the footage is 59.93 (why?), and when I import it into a 720/24p project in Avid, the clip runs at about double speed (which probably has to do with the frame rate, I just can't figure out why or how to fix it).

So my questions:

1. Why does the captured/logged clip in Liquid get a "!" sign?

2. How can I get the footage in AvidXpressPro with the correct frame rate, and best possible quality, when capturing with CapDVHS and converting with MpegStreamclip.

3. (in AvidXpressPro) What is the difference of editing in 720/24p and exporting to SD, and downconverting the footage to SD and then editing in 24p SD in Avid? By that I mean the end result, will there be a change in quality?

The video is shot in HDV 720/24p with HD101, with color correction and some effects planned in post.

Thanks

Steve Benner July 24th, 2006 02:25 PM

I have Xpress Pro 5.5.2 and this is what I do:

1) Import the footage (.m2t) into MPEG STREAMCLIP. I use a DR-HD100, but DVHSCap works fine. The file is 59.94 because that is what is written over firewire. The true 24 frames are hidden in it due to pulldown. HDV 720P flags repeat frames.

2) In MPEG STREAMCLIP, enter the frame rate, 23.976 in the emtpy box. Uncheck the "De-interlace" boxes since the footage is progressive.

3) I use the DNxHD Codec to edit 720/24P, but you can pick any codec you want to. Uncompressed SD will be fine, but don't edit in a 720P project if you are using SD footage. I prefer to Edit in HD and Downconvert with Compressor personally.

P.S. If you are using a Non-Drop Frame Rate, use 24 instead of 23.976.

Nima Taheri July 24th, 2006 03:46 PM

Quote:

Originally Posted by Steve Benner

3) I prefer to Edit in HD and Downconvert with Compressor personally.

Thanks Steve.

Wouldn't it be better to edit in SD, due to processor usage when editing HDV ? (if you are planning to have the finished product in SD, that is)

Also, is Compressor stand alone software?

Quote:

Originally Posted by Steve Benner

P.S. If you are using a Non-Drop Frame Rate, use 24 instead of 23.976.

I'm just capturing straight from CapDVHS, how do I know if I used a non-drop frame rate?

Steve Benner July 24th, 2006 04:33 PM

Drop Frame is used in NTSC countries. It is why 24P comes up 23.976 and 30P comes up 29.97. In Non-Drop mode, the frame rate is 24 or 30 exactly.

Compressor is bundled with Final Cut Studio, and I think you are working on a PC so that is out. Whatever DVD Authoring Program you use can do what Compressor does. Likely you have Sorenson Squeeze 4 because of Avid. That will work fine.

It is easier to edit SD than HDV, but Uncompressed SD is not as easily edited as DV footage though. The choice comes upon what program is going to be doing the downconversion, and if it does it well. I use MPEG STREAMCLIP really just to convert to a different editing codec, and I don't change the frame size. I am unsure as to how well it downconverts, but I do trust the program. The only real way to know if for you to do a test on the same clip.

Steve Benner July 24th, 2006 04:33 PM

Double Post.

Stephen L. Noe July 24th, 2006 05:02 PM

Quote:

Originally Posted by Nima Taheri
1. Why does the captured/logged clip in Liquid get a "!" sign?

In Liquid, right click on Start and select "about" and take a look at the splash screen. Are you on 7.0 or 7.1? You need to be on 7.1. If you're on the demo version then you're on 7.0 and you'll need to service pack the install to 7.1 build 3494 SP1 (3581) VTX.

The download is available on Avid's support site Start by clicking here

Accept the agreement on the bottom of the page and then open up the Liquid folder and go into the 7.0 folder and download the update.

have fun cutting 24p....

Nima Taheri July 24th, 2006 06:05 PM

Quote:

Originally Posted by Steve Benner
Drop Frame is used in NTSC countries. It is why 24P comes up 23.976 and 30P comes up 29.97. In Non-Drop mode, the frame rate is 24 or 30 exactly.

I'm a little confused, do you mean drop frames when recording or broadcasting? I used the HD101 (European) in 24p mode, editing it to be broadcasted primarly in the US, but also Europe if possible. Should the project be a 720/23.976 or 720/24 ?

Stephen:

Thanks for the tip, I'm downloading the update now.

1. As I mentioned, I'm primarly outputting to tape for US Broadcast, but might also need output to tape in PAL. How would I output the timeline to tape if I needed PAL output (if that is even possible in Liquid) when editing in 24p?

2. Any way of changing the timeline from HD to SD when exporting to tape?

Steve Benner July 24th, 2006 06:41 PM

If the area of distribution is primarly the US, then just go with the 23.976. It has to do with the configuartion of the NTSC Format. Direct 24 is used when cutting telecined film. If you go to PAL, then you will have to convert to 25P anyway.

Nima Taheri July 24th, 2006 06:48 PM

Thanks Steve, I got ya:)

Steve Oakley July 24th, 2006 07:42 PM

Quote:

Originally Posted by Steve Benner
Drop Frame is used in NTSC countries. It is why 24P comes up 23.976 and 30P comes up 29.97. In Non-Drop mode, the frame rate is 24 or 30 exactly.

NTSC is 29.97 period. Drop vs non-drop has to do with the TC count rate, not the frame rate the video is running at. Way back in the days of B&W TV, NTSC ran at 30.00, but due to introducing color, a cheat was devised on how to fit the color signal into the limitations of the broadcast signal of the day ( I am greatly simplifying here & skipping some details ) The end result of this cheat was 29.97. TV's of the day had sync circuts that could still lock the picture up without rolling, and B&W TV's could display the color picture so this misery was born.

So, once this happened, the problem was, if you ran TC at 30, but your real frame rate was 29.97, you'd be off one frame every 999 frames, 7.5 seconds per hour. If you had to cut a program to be 1:00:00, there needed to be a solution. that was drop frame TC. You simply skip COUNTING a frame every minute, except every 10 minutes, or something to that effect as I recall. yes its silly, but TC only runs on whole number frame rates, hence this horrible hack.

so that aside, most of the time 30 is used as shorthand for 29.97. if you have film shot at 24.000, its telecined to 23.976, and 3:2 pull down added for NTSC to come out at 29.97. in PAL, its sped up 4% to 25.000fps. the audio is speed adjusted as well for either scenario. Avid and FCP have audio Pull Up/Down.

the HD100 shoots 23.976, but records it to a NTSC video stream at 29.97. If you want to count in fields, its 59.94. When recording 24P ( Shorthand for 23.976) the camera adds 3:2 pull down using I beleive 2:2:2:4 pattern, and thats whats on tape. Your NLE then extracts the duplicated frames/fields from the sequence and produces a native 23.976 clip, which can be edited in 23.976. the clip will be reported as being either 29.97 or 59.94, but as long as your NLE can read the field metadata on duplicate fields, you can edit native 23.976.

Quote:

Originally Posted by Steve Benner
It is easier to edit SD than HDV, but Uncompressed SD is not as easily edited as DV footage though.

first, unless you are trying to cut with a 1ghz machine or something very underpowered, its just not true. If you have a modern computer made in the last two years or so, HDV edits just as easily as SD. I cut HDV with two+ streams of HDV in RT on my 1.5Ghz G4 laptop using the internal drive. HDV 720P has a data rate of 2.5mb/sec, DV is 3.5mb/sec, so HDV takes less spaces and has a lower speed requirements.

as far as uncompressed. FCP does more streams of RT with uncompressed than compressed. While the data rate is much higher, the CPU doesn't have to decompress the data to have a frame. Uncompressed is LESS work for the CPU. Data is transfered to RAM directly by the disc controller, this is known as DMA - direct memory access - something that has been in desktop computers for a very long time. The CPU doesn't have to handle the data, it just has to tell the DMA controller what to do, and a address in RAM to place it. very little overhead on the CPU while data flows from HD to RAM. This means the CPU and GPU can do other things, like calculate color correction, scale, position, crop, ect. on as many streams as it can handle.

Steve Oaklet

Nima Taheri July 24th, 2006 08:01 PM

Quote:

Originally Posted by Steve Oakley

the HD100 shoots 23.976, but records it to a NTSC video stream at 29.97. If you want to count in fields, its 59.94. When recording 24P ( Shorthand for 23.976) the camera adds 3:2 pull down using I beleive 2:2:2:4 pattern, and thats whats on tape. Your NLE then extracts the duplicated frames/fields from the sequence and produces a native 23.976 clip, which can be edited in 23.976. the clip will be reported as being either 29.97 or 59.94, but as long as your NLE can read the field metadata on duplicate fields, you can edit native 23.976.

If I edit on a 23.976 timeline, then export it to tape for NTSC broadcast, will the video then work?

Steve Oakley July 24th, 2006 08:17 PM

yes, because your NLE will add normal 3:2 pulldown on output turning your 23.976 to 29.97. FCP has a number of tools/features to support various 24/25/30 workflows combined with cinema tools. I don't have it all in my head right now, but you can RTM because apple does provide good docs on this. of course you'll need a VTR that can accept NTSC, which most DV units made in the last two years or so do, again check the specs of the deck you plan to use for output.

Steve Oakley

Steve Benner July 24th, 2006 09:26 PM

Quote:

Originally Posted by Steve Oakley
first, unless you are trying to cut with a 1ghz machine or something very underpowered, its just not true. If you have a modern computer made in the last two years or so, HDV edits just as easily as SD. I cut HDV with two+ streams of HDV in RT on my 1.5Ghz G4 laptop using the internal drive. HDV 720P has a data rate of 2.5mb/sec, DV is 3.5mb/sec, so HDV takes less spaces and has a lower speed requirements.

Steve Oaklet

Even though the Bit Rate is lower than DV, HDV requires MUCH more processing power because of the MPEG-2 Analysis and Reconstruction than needs to occur for every frame outside of an "I" frame. I have a G4 (1.33 GHZ with 2 Gigs of RAM) and the Intra-Frame DNxHD Codec edits much better than raw HDV. The biggest problem is around transitions in reguards to HDV. In Final Cut Pro, HDV behaves similarly.

I have no problems cutting it either, but HDV is not a great editing codec. I use it because that is what the camera shoots in, and I don't feel like converting every file to another codec when I can live with it.

In reguard to Uncompressed, I do not use so I can't say from experience, but here is a graph reguarding Real Time Streams of playback for a G-Raid.

http://i9.photobucket.com/albums/a56...RAID-chart.gif

Steve Oakley July 24th, 2006 10:35 PM

Quote:

Originally Posted by Steve Benner
Even though the Bit Rate is lower than DV, HDV requires MUCH more processing power because of the MPEG-2 Analysis and Reconstruction than needs to occur for every frame outside of an "I" frame. I have a G4 (1.33 GHZ with 2 Gigs of RAM) and the Intra-Frame DNxHD Codec edits much better than raw HDV. The biggest problem is around transitions in reguards to HDV. In Final Cut Pro, HDV behaves similarly.

well what if you decode in blocks of GOPs, 6 or 15 frames at a time rather than deconding each frame at a time ? in the end, you still have to decode 15 frames to play 15 frames no matter what code you use. the question is how much load it takes. NLE's don't work with single frames, they cache a certain amount in ram to ensure playback in case of drive hiccups or other system resources interupt the flow of data. Thats why they all have some pause factor between when you hit play. That said, Mpeg2 probabley does take more overhead, but thats where SSE & alitvec comes into play. using those specialized instructions, you can process more per clock cycle than just using the standard CPU instructions. Many GPU's also have Mpeg2 decoders built in, but I can not say for a fact if FCP takes advantage of this or not.

on my machine in FCP5.0.4, as a base, I can run two clips with the 3way color corrector through a 1 second or longer dissolve all day long. I can also get at least one static graphic. I've also run dual PiP on the laptop was well.I've gotten three layers to in RT, so for basic editing, what more do you need ? if you need more RT streams, you should be on a dual/quad proc tower. I find HDV edits just as easily and quickly with about the same amount of RT as does DV25. I can't speak for avid, I don't use it.

on my G5 2.5G with 6800GT, I've done 4 layers of HDV at once. certainly the G5 have a lot more HP, but HDV is editing just as well and as easily as DV25.


Quote:

Originally Posted by Steve Benner
I have no problems cutting it either, but HDV is not a great editing codec. I use it because that is what the camera shoots in, and I don't feel like converting every file to another codec when I can live with it.

huh ? if it edits ok, why change the codec ? its a great codec for laptop editing because of its low data rate. unless you're going to uncompressed for a really good reason, going to another compressed format is another compression hit without any gain, and certainly some loss (1280X720 HDV->960X720 DVCpro is a big loss). its not a format for those aren't good shooters because you have to be on the money with exposure.

Quote:

Originally Posted by Steve Benner
In reguard to Uncompressed, I do not use so I can't say from experience, but here is a graph reguarding Real Time Streams of playback for a G-Raid.

FW800 isn't up to the task. you need SCSI320 or FC. basic math is @ 8bit= 21mb/sec per stream or 7 streams X 21mb/sec = 147mb/sec. I've done this with a internal SATA raid. That said, its a bit of a moot point because how often do you ever do something like 7 PiP's ? I'd expect at most about 4 streams for most edits. perhaps you're editing multicam might be the only time you'd get into need that many streams to play at once.

http://i9.photobucket.com/albums/a56...RAID-chart.gif

Steve Benner July 25th, 2006 05:15 AM

Quote:

Originally Posted by Steve Oakley
how often do you ever do something like 7 PiP's ?

If I was an editor on 24, I would have an answer for you...

Thanks for informing me on all this stuff by the way.


All times are GMT -6. The time now is 08:08 AM.

DV Info Net -- Real Names, Real People, Real Info!
1998-2024 The Digital Video Information Network