Panasonic unveiled G1-HD...

Status
Not open for further replies.
Car3o said:
you're right, the hv20 does have "jello" but you're comparing it in a helicopter.
Click to expand...
again it's just a different in scale. you said skew was ok but not jello, and it simply doesn't work like that. if something is one inch long it's also 2.54mm, that's how they are connected, they're *the same thing*. it's not meaningful to discuss whether one has jello or "just" skew, or vice versa. what can be discussed, and must be, is how much they have compared to each other. the d90 has alot, many cell phones even more, the hv20 and red one much less. simple as that.

/matt
 
Eddy Robinson said:
Because it's about as good as it's going to get on a camera of this size.

First, the noise on the clip you mention is due to excessive compression. I'd bet money that it was recorded from the built-in mic rather than the external one; not sure whether it has the 'wind rumble' turned off or not (a low pass filter) but I would never expect anything other than tinny transistor noise from an onboard mic like that.

Second, the option to plug in an external microphone is a good compromise. The little stereo one with the hot-shoe adapter offered by Panasonic is, dare I say it, cute. On the downside, it requires external microphones to be 'Dolby compatible - so no feeding it from a mixer. You can get decent and affordable mics besides the Panasonic one, but I'm fairly sure the point of this decision was to drive accessory sales. I'm sure the Panasonic mic isn't fantastic, but it's probably average, which these days means quite good. I bet someone will offer a box to convert incoming line signals (ie from a mixer) before long.

Third, OK, so we have to consider external recording solutions, oh noes. Well, I do have a DAT recorder which is a giant brick that's bigger than 3 of these cameras...but I'd suggest a ~$300 Zoom H4/H4n which will do a great job, including allowing you to plug in expensive condenser mics. You can of course spend a lot more, but the H4 is very good and very compact. The new one records BWF, which means a timestamp in the audio metadata.

Fourth, If you're doing anything serious, you should be using an external mixer of some kind anyway. Nobody in their right mind trusts the camera for onboard mixing, any camera. This is why I'm not too worried about whether it has automatic gain control or not, though I suspect not. I mean seriously, you plan adjust focus manually AND manage two dials to adjust the audio level in real time? No, I didn't think so.

Fifth, but you'll have to sync it all up with an external recorder! Oh noes! Well, put on your big boy/girl pants and sync it up then. IT AIN'T THAT HARD. Frankly, I am constantly amazed at how video shooters paint this as some sort of epic struggle. Did you know that film cameras don't record sound at all? Yes, everything ever shot on film has been synced up in post! That's one of the main reasons to use a slate ($50-$60 for a basic one at any film equipment retailer). If you have a lot of money, you can shoot with a timecode slate and pay lab about $150/hour extra to automagically match up the audio in telecine, but most indie films do it by hand.

(Protip: you can also just point the camera at the display of the audio recorder or hold the recorder up in front of the camera before each shot. Bingo - you've got a frame-accurate timestamp reference for picture and sound. Don't worry so much about timecode issues - modern digital clocks and NLE software are so accurate that sync is a non-issue as long as you plan and do a few tests in pre-production.)

I've done it for a few features and more shorts than I can count. Sure, it's kinda boring. It can take a whole week or longer to do for a feature. Fortunately, nowadays you don't have to listen to/watch it all play back in real time as you capture from DAT or miniDV tape, youc an just hook up the USB port and copy the files, saving many hours of work.

Really. Double-system is a lot simpler than most people realize, and you'll get better results in most cases. As an audio guy myself, I view the onboard audio of this camera as a handy bonus - to be frank, I wouldn't care that much if it were just mono or had no external input, as I'd just go back to using a slate or similar.
Click to expand...
And with no use of slate? Does the Zoom H4/H4n have the same possibility as a tascam p2 for example?

http://www.bhphotovideo.com/c/product/405341-REG/Tascam_HD_P2_Portable_CF_Stereo.html

"The HD-P2 is equipped with comprehensive SMPTE and Video synchronization."

http://www.dvinfo.net/conf/969645-post7.html

"You don't even technically need to use a slate, you could record a wild track through the camera and your professional sound through the p2."
 
Eddy Robinson said:
Totally - it remains to be seen how the acessory mic performs on GH1 (or more specifically, what happens to the audio once it's in the camera). One upside is that I noticed it records at 48khz so there are gounds for optimism.

I was thinking more of narrative work where you have a floating boom and wireless mics coming in, and the expectations are a lot higher than for documentary, where you can set a safe level and leave it there. For narrative I ride the gain on the mixer a lot because you might have an actor go from whispering to shouting n one scene but it all has to sound smooth.
Click to expand...
Why? Is that so bad to have 44kHz such as happens with the 5D MKII?
 
Audio sync answers - very long and sort of off topic!

Audio sync answers - very long and sort of off topic!

Moderators - I'm happy for this to be moved if necessary. I wrote it out here because I feel it will be particularly helpful for people shooting with the newer cameras like the d90 or GH1 where sound may need to be captured separately.

Gordon Prince said:
And with no use of slate? Does the Zoom H4/H4n have the same possibility as a tascam p2 for example?

[...]

"The HD-P2 is equipped with comprehensive SMPTE and Video synchronization."
Click to expand...

Not exactly. The HD-P2 outputs a SMPTE timecode signal which you could use to drive a smart slate (ie one that displays timecode) or a Lockit box going to the Genlock input on a pro camera. The HD-P2 is pretty nice, although I have had reliability and signal path problems. It was new then, it might have been an early model problem.

Zoom gear will not output timecode like this - that's one of the things you pay a good bit more money for. See below on the workaround.

"You don't even technically need to use a slate, you could record a wild track through the camera and your professional sound through the p2."
Click to expand...

Yeah...but you'd still need to synchronize some way. Let me break post sound tech down very simpl(istical)ly.

A: BASIC SLATE
The slate serves two purposes. One is that you write on it with a dry erase marker the scene, shot and take you are shooting and display that info to the camera before every take (and speak it for the sound guy). For obvious reasons, having this is absolutely invaluable for the editor.

The other is the clacker, the stick on top of the slate which makes a bang when it's closed. It's easy to see on what frame the slate closes (this is why the stripes are on it) and the sound of the clacker being closed makes a very thin high peak in the audio waveform. After you've done a lot of audio editing, You get used to seeing the distinctive shape and can easily pick out the head of each take by eye.

Syncing up video and picture is a simple matter of locating the video frame on which the slate closes, checking the scene-shot-take against the scene-shot-take called out for the audio recording before the camera rolls, and then dragging the audio clip line up with the video frame.

It sounds elaborate, but becomes very quick with experience. You could do 3-400 shots per day (realistically).

B. TIMECODE
As you know, audio used to be recorded on analog tape, with big reel-to-reel devices. Unfortunately, Tape is subject to tiny variations in speed, giving rise to phenomena called 'wow' and 'flutter' where the pitch warbles ever-so-slightly. Over time, these minor variations can cause sound to drift out of sync with picture.

Timecode was invented to help overcome this: by recording a low-frequency square wave on the second track of the tape, the wow and flutter could be measured and corrected for in the studio. Additionally, it can be used to drive a slate over a cable or even wirelessly: so if you shoot film, in telecine you just pause the film at the start of the take, read the #s on the slate from the screen, type those into a Studio DAT deck (or straight into the computer) and have it cue up the relevant audio for you. Then you press the play button, and the film strip plus the audio play back in sync, including any pulldown you are doing in the USA.

C. Shooting Video
...is almost the same. You run a cable into the camera and derive timecode directly from the audio recorder (which by tradition rather than necessity, is always the master clock source). Or you set the same frame rates on camera and audio recorder and use the timecode slate method outlined above.

If audio and picture are recorded at different frame rates, then you run into big problems, because - most often - the audio is running 0.1% slower than the video and the two drift out of sync over the course of a long take. Actually it happens on every take but it's not as noticeable on short ones, leading to the mistaken belief that 'some of the audio is fine, some is out of sync'. The Producer and Editor will always blame the sound person regardless of who is actually at fault.

D. The POINT, FINALLY
No matter what you do and what recorder you use, you still need some common point of reference between the audio and video recordings, or all your precious timecode is useless.

If you shot without a slate, you also won't have audio or video synchronization OR a written/verbal reference of scene, shot and take #. This means that 20-40% of your material will be mis-labelled when it's brought into the NLE. You'll end up trying and failing to line up audio from one take with video from a different take on a regular basis, and you will cry. A lot. It's fixable, but slow and frustrating.

If you shot wild sound in the camera, you have a rough point of reference - the audio waveform for the camera track will resemble that of the separate audio track and you can line them upfairly quickly. This will still be subject to error, and tedious, but it will be manageable. NOTE: a similar problem exists when you shoot with two cameras, both recording sound, but don't use a slate on every shot. Matching the two video streams on the timeline will drive you nuts. Different takes of the same shot look very similar and there will be a lot of mistakes. Ooooops.

If you didn't record any sound in the camera, you'll be trying to match up an audio dialog recording with a silent video image of someone speaking. You won't just cry, you'll roll around on the floor pulling out your hair. People will yell at each other and friendships will end. A good audio post person (who may or may not be the picture editor) can straighten this out, but it will take a lot of patience, and you should be willing to bribe them with money and whatever else it takes. You'll be lucky to get through 100 shots per day if you have to do this - it becomes very, VERY repetitive and tedious.

E. THE SOLUTIONS
As ever, all of this could have been avoided if it had been addressed during pre-production and certain rules followed - religiously - during the shooting. Whether it's separate audio and camera or multiple cameras, you MUST use a consistent system to both identify every single shot and wild recording, and to keep audio and video (or second video camera) in sync.

1. Write everything down
Just writing down the time and shot details every time someone pulls the trigger will eliminate many problems.

2. Use a slate
A timecode slate is great but not necessary unless you have the budget and expertise to fully exploit it. A regular slate will save you a vast amount of time. If you can't use a slate (because you left it at home, or you don't want people to see you're shooting a movie...)

3. Announce each take and clap hands
In front of the camera - then you can see the handclap and hear it on the audio recording. This is not really ideal, though. A cellphone that beeps (and does something on screen) each time you hit a button is better. The basic point: use something that makes a visual and audible signal which you can see in a single frame. Use it consistently.

4. Timecode I
Usually people set 'Rec Run' timecode, so that you get continuous TC on the video tape, from 0-60 minutes. Consider using 'Time of Day' instead. Set your A camera and your audio recorder (or B, C, D...cameras) to the same time of day. Then shoot for the rest of the day. Have your editor be very careful to keep the footage from every shoot day in a separate folder/bin. then put it on a timeline using the timecode information you have to create the sequence.

Your multiple timecode sources will not be synchronized properly. Even though you set all devices to the same time of day, they'll be out of sync because it's impossible to hit the 'time set' buttons perfectly accurately on different devices. BUT: this offset will be consistent throughout the day. Meaning you pull all of the day's footage into the NLE, put it all on a timeline, then line up the video and audio (or 2nd video stream) ALL AT ONCE.

Actually - it won't be perfectly consistent - different clocks have jitter and will drift fractionally throughout the day, even clocks on the same make and model of camera. But fixing that is quick. You can do it in a few seconds per clip whenever you notice it (since you'll be on the lookout for this possibility), or you can jump to the end of the timeline with each day's footage on it and check how bad the drift it on your final clip. It';s unlikely to be more than a frame or three. To fix that, just check every 10th clip or so on the timeline, and when you find one that feels 'off', resync that and everything to the right of it on the that day's timeline.

5. Timecode II: clocks

This isn't true timecode, but a workaround involving an external clock. As noted elsewhere, do not be a slave to the timecode generated on your A camera - over-reliance on it can be misleading and cause costly mistakes.

Even a cheap audio recorder such as the Zoom will have a time display - it'll say something like 'Clip 5 - 0:11:43.374' ie, clip #,h:m:s:ms. If you start recording on both devices and then point the camera (when handheld) at the display of the audio recorder, or alternatively hold the audio recorder in front of the camera lens (when on a tripod) on every take, you'll have a few seconds of video telling you exactly what the runtime is on the audio recorder.

If the recorder saves audio as .bwf, this time will be in the file metadata and recognized by most NLEs or audio editing packages. If it records .wav or .aiff, then the time information will be either included in the filename of the audio clip or if not, hour and minute will be on the file's timestamp.

Bring the audio into your NLE first and put it on a timeline. Now you have a timeline with all the audio you recorded in sequence from beginning to end. Exactly how depends on the recorder and how it saves time info per clip and how the clip is named. Now you take your video, and for each video clip look at the footage which displays the audio recorder's clock, and it tells you which clip # to match with, and where in time to match it.

Example: You have a cheap recorder. You start recording audio at 2:17pm on Jan 1. You put it in front of the camera and start shooting and the audio display says 'clip 1, 0:0:3.532...0:0:5.453' (the clock is running from the 1st to the 2nd #s). You move it out of the way and shoot your scene.

You import the audio to your NLE as clip 1. You look at the time on the file and set up your timeline to start at 14:00:00:00 (2pm = 1400 hours). You move the clip 1 to 14:17:00:00. You bring in clip 1 from the camera.

Find the frame of video showing 'clip 1, 0:0,3:532...' You slide the video clip so that that frame is 3.532 seconds in from the beginning of the audio clip. Hit play: you should have sync. Lock the audio and video together. Do this for every clip you shot that day. Save the project often.

When they're all done, you have, say, 73 clips. Save the project again, trim the start and end of each clip to be the same length. Export all the clips to a new folder (or better, to a separate hard drive) called 'sync - day 1'. Close the project. Open a new one, and import those clips. Now, start editing. The import-export stage is to print the newly-established timecode on them, and also for sanity: you get one set of video clips with audio instead of 2 sets of separate video and audio clips, meaning you only have to keep track of half as many files while editing.

If you're shooting with two or more cameras (each of which are recording sound onboard), it's almost the same. Keep a decent-size, decent-quality digital clock on set. Battery powered is better so it doesn't lose the time if you unplug it. At the beginning or end of EVERY take, EVERY camera gets a shot of the clock (IN the take - it doesn't work if you pause the camera before or after shooting the clock). You MUST shoot the clock long enough to see the seconds changing (or the minute, if you got a really cheap clock). Now every clip from every came includes a visual reference to the same external time source.

Import all your video, put it on a timeline, each camera on a separate track - the frames which show the clock tell you where to put it on the timeline. Your timeline will be as long as your shooting day was. You get frame accuracy by lining up the frames where the second (or the minute) changes over with that point in the timeline. Now, all the footage on the different tracks should be in sync and you can cut seamlessly. Add sync markers or whatever you do sync clips in your NLE.

F. FINALLY, DAMN

The above won't solve all your potential a/v sync problems, but it will at least tell you how to approach them. Like many other things, start thinking about this in preproduction. Don't rely on your own preconceptions, or reassurances from your camera person, editor, or sound person. Plan and test. It's a sad fact that a lot of camera and editor and VFX people know a bit about timecode, but not enough to solve problems. Sound people know more, not all of them know all about it. And in a less technical way, camera and picture editing people often don't know much about audio editing.

Audio editing has more in common with VFX compositing than anything else, and there are just as many ways to screw it up. I also think production sound has a lot in common with lighting. The traditional model is to do the visuals first and consider the audio second, and much time (and money) is lost from this approach. Just as it's good to involve an editor in pre-production or think ahead to the editing if you do it yourself, plan your creative and technical sound stuff before you begin shooting rather than afterwards. Personally, I think that editing picture and sound are best done by the same person, though such people are relatively few and far between. 80-90% of post audio problems are best fixed in the NLE rather than rushing to get a pretty picture locked and then tackling the audio issues separately.

There are other issues I've not addressed here - drift, what happens when you have multiple microphones on telephoto shots, reasons when you should break sync and how to do it without being noticed, reasons TV and theater sync are different, and the many many ways of cheating your audio in post (but please don't assume you can 'fix it in post' unless you know specifically how).


In summary, if you have a full-size camera that can record sound on set you're off to a great start, but if not then don't worry - these issues were around before video and were solved for film. And while sync is the most obvious technical issue, it's not the only one - sound is a many-splendored thing, and the more you plan for it the more fun it will be in post.
 
i just synced up 40 hours of material for a feature, no timecode and no slates, not even voice slating, but with the aid of d90 built in sound for reference. i simply used the timestamp on the files to get the right clip (usually there was one audio clip per video clip so i didn't even have to do that), then set markers at the wonderful sound of my own voice yelling "action" on both tracks, then aligned them. it took an entire day, but not more than an entire day, which isn't bad.

/matt
 
Good video piece on the camera you found.

Notice how they emphasized full manual control in video mode, and that it was a bridge between consumer and pro video....

Jeez, what was Canon thinking; or not thinking when it came to not having 24fps, manual control and more on the 5D. Funny is Panasonic did this with the DVX100. Before that there was this lack of 24fps and good robust small form pro camera and then they knocked out a great camera the DVX100. I'm seeing a similar parallel here too!

June time frame release as well.....


Michael Pappas
http://www.pbase.com/Arrfilms
http://www.Myspace.com/PappasArts
http://www.facebook.com/people/Michael-Pappas/573417404
Arrfilms@hotmail.com
http://www.PappasArts.com
CONTACT VIA AOL INSTANT MESSENGER
AT { PAPPASARTS2 }
 
Yeah, I'm pretty pumped about this camera. Panny really might have a winner with this.

I don't think it's significant, but with Canons MkII you get FF sensor. Better light, less noise and so on. So I guess there are trade offs. But the GH1 looks like a clear winner especially if the rumor is that it's going to be under $1500.00.
 
Nice read and workaround.

Eddy Robinson said:
5. Timecode II: clocks

This isn't true timecode, but a workaround involving an external clock. As noted elsewhere, do not be a slave to the timecode generated on your A camera - over-reliance on it can be misleading and cause costly mistakes.

Even a cheap audio recorder such as the Zoom will have a time display - it'll say something like 'Clip 5 - 0:11:43.374' ie, clip #,h:m:s:ms. If you start recording on both devices and then point the camera (when handheld) at the display of the audio recorder, or alternatively hold the audio recorder in front of the camera lens (when on a tripod) on every take, you'll have a few seconds of video telling you exactly what the runtime is on the audio recorder.

If the recorder saves audio as .bwf, this time will be in the file metadata and recognized by most NLEs or audio editing packages. If it records .wav or .aiff, then the time information will be either included in the filename of the audio clip or if not, hour and minute will be on the file's timestamp.

Bring the audio into your NLE first and put it on a timeline. Now you have a timeline with all the audio you recorded in sequence from beginning to end. Exactly how depends on the recorder and how it saves time info per clip and how the clip is named. Now you take your video, and for each video clip look at the footage which displays the audio recorder's clock, and it tells you which clip # to match with, and where in time to match it.

Example: You have a cheap recorder. You start recording audio at 2:17pm on Jan 1. You put it in front of the camera and start shooting and the audio display says 'clip 1, 0:0:3.532...0:0:5.453' (the clock is running from the 1st to the 2nd #s). You move it out of the way and shoot your scene.

You import the audio to your NLE as clip 1. You look at the time on the file and set up your timeline to start at 14:00:00:00 (2pm = 1400 hours). You move the clip 1 to 14:17:00:00. You bring in clip 1 from the camera.

Find the frame of video showing 'clip 1, 0:0,3:532...' You slide the video clip so that that frame is 3.532 seconds in from the beginning of the audio clip. Hit play: you should have sync. Lock the audio and video together. Do this for every clip you shot that day. Save the project often.

When they're all done, you have, say, 73 clips. Save the project again, trim the start and end of each clip to be the same length. Export all the clips to a new folder (or better, to a separate hard drive) called 'sync - day 1'. Close the project. Open a new one, and import those clips. Now, start editing. The import-export stage is to print the newly-established timecode on them, and also for sanity: you get one set of video clips with audio instead of 2 sets of separate video and audio clips, meaning you only have to keep track of half as many files while editing.

If you're shooting with two or more cameras (each of which are recording sound onboard), it's almost the same. Keep a decent-size, decent-quality digital clock on set. Battery powered is better so it doesn't lose the time if you unplug it. At the beginning or end of EVERY take, EVERY camera gets a shot of the clock (IN the take - it doesn't work if you pause the camera before or after shooting the clock). You MUST shoot the clock long enough to see the seconds changing (or the minute, if you got a really cheap clock). Now every clip from every came includes a visual reference to the same external time source.

Import all your video, put it on a timeline, each camera on a separate track - the frames which show the clock tell you where to put it on the timeline. Your timeline will be as long as your shooting day was. You get frame accuracy by lining up the frames where the second (or the minute) changes over with that point in the timeline. Now, all the footage on the different tracks should be in sync and you can cut seamlessly. Add sync markers or whatever you do sync clips in your NLE.

There are other issues I've not addressed here - drift
Click to expand...

Using your workaround, what do you have to say about it?

http://www.dvinfo.net/conf/canon-eos-5d-mk-ii-hd/138476-how-about-good-quality-audio-6.html
 
mattsand said:
plus that $2,000 olympus f/2.0 zoom. drool. :-)

/matt
Click to expand...

Haha..., how cheap it is. I think, the most important downsides of this cam are the lack of manual control the Iso values in combination with a kit lens and f4.0. Besides I don´t like this small teeny weeny body. Maybe it´s not very handy...
 
who said you can't control the iso? the kit lens is bullshit, i agree, and since it probably comes with all cameras there's no market for it either.

/matt
 
mattsand said:
who said you can't control the iso? ...
Click to expand...

In the specs and the YT-Clip you only find statements referred to manual control over shutter speed and aperture - no iso. You know anything more? And yes, you can of course use glasses of other brands with an adapter, but then there will be no AF-abilities...
 
Last edited:
Other than a few applications (sports, wildlife, point and shoot etc) I don't see the need for auto focus myself.
 
For you and the most guys here, maybe. But the main target group for this little thing are amateurish user. For them the absence of AF is an absolutely no go...
 
but the amateur will be happy with the stock lens, which has AF. The pros will be looking to add expensive glass (non AF).

All i'm saying is it is not a big loss for most of us to grab better glass, but lose AF.
 
Status
Not open for further replies.
Back
Top