I did a mulit-camera camera shoot using 2 - 100s and 2 - 100a cameras all set at 24P. Unfortunately I got the DVX book after I got home from the shoot. According to the book, the SMPTE time code does not work on the 100s when you shoot in 24P. I set free run timecode on every camera and we all pushed the set buttons at the same time, so we started off with all the timecodes within a second of each other. By the end of the day one camera was 10 seconds slower than the others - on the timecode display. After several days they all were displaying quite different timecode numbers. Fortunately I used a lot of slates. Shouldn't the free run timecode on the 100a models have stayed roughly in sync even after several days? Other than using a slate, is there a better way to sync up cameras on a multi-camera shoot using 24P?
mulit-camera time code drift
- Thread starter dcragoe
- Start date
Re: mulit-camera time code drift
If videocams are not "genlocked" they will drift. They are quartz stabilised (to be compared with a digital watch). 1 second/hour (10 seconds /day) is quite a lot. Maybe the cam's internal oscillator can be adjusted to a more precise value so that the drift will be lower.
If videocams are not "genlocked" they will drift. They are quartz stabilised (to be compared with a digital watch). 1 second/hour (10 seconds /day) is quite a lot. Maybe the cam's internal oscillator can be adjusted to a more precise value so that the drift will be lower.
Re: mulit-camera time code drift
Sounds to me like something is off in that one camera. Maybe the internal battery is weak/needs charging???? (I don't think that would change the running rate, but who knows). They all drift some, but ten minutes is a lot.
I push the buttons first thing in the morning. I let them warm up for a couple of minutes, try to match color as best I can, set the clocks and go. 12 hours later they are very close.
HTH, Chris
Sounds to me like something is off in that one camera. Maybe the internal battery is weak/needs charging???? (I don't think that would change the running rate, but who knows). They all drift some, but ten minutes is a lot.
I push the buttons first thing in the morning. I let them warm up for a couple of minutes, try to match color as best I can, set the clocks and go. 12 hours later they are very close.
HTH, Chris
FlacaProductions
Well-known member
Re: mulit-camera time code drift
check that all cameras are either drop-frame or non-drop-frame.
check that all cameras are either drop-frame or non-drop-frame.
Re: mulit-camera time code drift
24p always defaults to non-drop frame, so that isn't a likely cause here. The difference between DF and NDF is about 3.6 seconds/hour so if the cameras ran all day (~8 hours) there would be about a 29 second difference between DF and NDF TC.
It sounds like the quartz timing crystals in the cameras are oscillating at slightly different frequencies. The crystal accuracy is affected by the manufacturing specs, so a slightly different cut to the crystal or contamination on the crystal surface can affect the oscillation frequency.
24p always defaults to non-drop frame, so that isn't a likely cause here. The difference between DF and NDF is about 3.6 seconds/hour so if the cameras ran all day (~8 hours) there would be about a 29 second difference between DF and NDF TC.
It sounds like the quartz timing crystals in the cameras are oscillating at slightly different frequencies. The crystal accuracy is affected by the manufacturing specs, so a slightly different cut to the crystal or contamination on the crystal surface can affect the oscillation frequency.
JJ_Shantell
Active member
Re: mulit-camera time code drift
Im going to be doing this
so theretically, i can set each cam (in 30p) with the remote to freerun and they would be close over several days or would it be better to do it for each days shoot?
what about changing tapes? if I change tapes the tc will just pick up where it left off on the previos tape?
Im going to be doing this
so theretically, i can set each cam (in 30p) with the remote to freerun and they would be close over several days or would it be better to do it for each days shoot?
what about changing tapes? if I change tapes the tc will just pick up where it left off on the previos tape?
Re: mulit-camera time code drift
If you are shooting in 30P I think it's better if you have all 100a cameras and not use the older 100s. Supposedly, according to the DVX book, the SMPTE time code only works on 60i in the 100 model. The time code is supposed to work in all modes on the 100a. It may drift at 30p, I don't know. Also, the book tells how to use one remote control to start the time code on all of your cameras at one time. You just make sure it's controlling all of the cameras and then start the time code from 0. This way all the cameras should read exactly the same time code from the start. There apparently is an internal battery in the camera that keeps the time code running when you make battery changes, so you don't lose your place. If you have 2 cameras together you might try a test and see if the time code stays the same. for a day while you change batteries and run some tape. I'm curious myself if the time codes would stay together.
If you are shooting in 30P I think it's better if you have all 100a cameras and not use the older 100s. Supposedly, according to the DVX book, the SMPTE time code only works on 60i in the 100 model. The time code is supposed to work in all modes on the 100a. It may drift at 30p, I don't know. Also, the book tells how to use one remote control to start the time code on all of your cameras at one time. You just make sure it's controlling all of the cameras and then start the time code from 0. This way all the cameras should read exactly the same time code from the start. There apparently is an internal battery in the camera that keeps the time code running when you make battery changes, so you don't lose your place. If you have 2 cameras together you might try a test and see if the time code stays the same. for a day while you change batteries and run some tape. I'm curious myself if the time codes would stay together.
Re: mulit-camera time code drift
There will always be a TC shift, even with the most precise quartz cut tolerances. In best cases the shifts are comparable to digital watch accuracies (less than 1 sec/days). Temperature compensation which would yield better results is excluded because of the extra battery power needs. "Genlocking" is the only way...if the cams allow it.
There will always be a TC shift, even with the most precise quartz cut tolerances. In best cases the shifts are comparable to digital watch accuracies (less than 1 sec/days). Temperature compensation which would yield better results is excluded because of the extra battery power needs. "Genlocking" is the only way...if the cams allow it.
JJ_Shantell
Active member
Re: mulit-camera time code drift
[quote author=dcragoe link=board=Events;num=1101655316;start=0#8 date=12/02/04 at 14:14:06] If you are shooting in 30P I think it's better if you have all 100a cameras and not use the older 100s. Supposedly, according to the DVX book, the SMPTE time code only works on 60i in the 100 model. .[/quote]
????
what if i shoot 60i on dvx and 30p on dvxa? i dont care about matching the image-just sync
[quote author=dcragoe link=board=Events;num=1101655316;start=0#8 date=12/02/04 at 14:14:06] If you are shooting in 30P I think it's better if you have all 100a cameras and not use the older 100s. Supposedly, according to the DVX book, the SMPTE time code only works on 60i in the 100 model. .[/quote]
????
what if i shoot 60i on dvx and 30p on dvxa? i dont care about matching the image-just sync
JJ_Shantell
Active member
Re: mulit-camera time code drift
on Dec 2nd, 2004, 2:14pm, dcragoe wrote: If you are shooting in 30P I think it's better if you have all 100a cameras and not use the older 100s. Supposedly, according to the DVX book, the SMPTE time code only works on 60i in the 100 model. .
?
what if i shoot 60i on dvx and 30p on dvxa? i dont care about matching the image-just sync
is this really an issue or heresay??
on Dec 2nd, 2004, 2:14pm, dcragoe wrote: If you are shooting in 30P I think it's better if you have all 100a cameras and not use the older 100s. Supposedly, according to the DVX book, the SMPTE time code only works on 60i in the 100 model. .
?
what if i shoot 60i on dvx and 30p on dvxa? i dont care about matching the image-just sync
is this really an issue or heresay??
Bart_Boge
Well-known member
Re: mulit-camera time code drift
When I first started my video projects (multicam event videography) I heard a lot of ballyhoo about "losing synch" and "drift." I think it is much to do about essentially nothing.
If you are running a 3-camera shoot and you notice a slight timing discrepancy between your cameras at some point, then you just drop a frame from the faster camera or add one to the slower camera when it is not the selected shot. How hard is that? At 24p, who the heck will notice the difference?
On my last project, I did a 3-camera shoot of a wedding and recorded the audio completely separately via a MOTU 896HD interface and a Mac on location. When editing, I just looked for a camera flash from the stills guy to sync the video streams to. Walaah! Synched video streams. During the editing process I used the audio from just one camera to edit to. If one of the other cameras drifted away slightly from that audio, then I would trim it back into sync. This might occur perhaps ONCE in an entire live multicamera shoot.
Then I take that scratch audio, export it as a .aif, and import it to my audio mixing program where my ACTUAL audio mix will be coming from. I find a loud audio event and use it as a sync point between the scratch audio and the final audio so that they are identically timed, then I trim the recorded 896HD audio to fit the exact length of the scratch audio. Using Digital Performer, I can watch a frame-accurate mp4 of the video with the audio to be sure it matches. Then I export the mix, import it into A.Pack, and burn the DVD.
The result is great audio (5.1 surround, in this case) that is in sync with all of the video, and all of the camera footage is in sync as well. If this sounds like a goofy way to do things, remember that 35mm film does not record audio--every major motion picture records its location audio on a separate system and syncs them in post (or foley/loops new audio in post, or a combination of the two).
BTW, using two DVX100a's and one DVX100 on my last project, I did not have to trim a single frame due to drift (45 minute run time).
My $.02, FWIW.
Edit:
You can get all wrapped up in timecode issues if you want. I don't worry about it--if the images are synched together, I don't care what the numbers say or if they match. And I certainly don't rely on either external synchronization via timecode or any "record start" synching methods. I literally walk from camera to camera, hit record, then on to the next one. I might have 3 full MINUTES of record/TC discrepancy this way, but when I edit in post it takes me about 5 minutes to line up all the streams to frame-accurate to that one still-camera flash (or any other sudden movement/video event, like a clapper). No biggie.
When I first started my video projects (multicam event videography) I heard a lot of ballyhoo about "losing synch" and "drift." I think it is much to do about essentially nothing.
If you are running a 3-camera shoot and you notice a slight timing discrepancy between your cameras at some point, then you just drop a frame from the faster camera or add one to the slower camera when it is not the selected shot. How hard is that? At 24p, who the heck will notice the difference?
On my last project, I did a 3-camera shoot of a wedding and recorded the audio completely separately via a MOTU 896HD interface and a Mac on location. When editing, I just looked for a camera flash from the stills guy to sync the video streams to. Walaah! Synched video streams. During the editing process I used the audio from just one camera to edit to. If one of the other cameras drifted away slightly from that audio, then I would trim it back into sync. This might occur perhaps ONCE in an entire live multicamera shoot.
Then I take that scratch audio, export it as a .aif, and import it to my audio mixing program where my ACTUAL audio mix will be coming from. I find a loud audio event and use it as a sync point between the scratch audio and the final audio so that they are identically timed, then I trim the recorded 896HD audio to fit the exact length of the scratch audio. Using Digital Performer, I can watch a frame-accurate mp4 of the video with the audio to be sure it matches. Then I export the mix, import it into A.Pack, and burn the DVD.
The result is great audio (5.1 surround, in this case) that is in sync with all of the video, and all of the camera footage is in sync as well. If this sounds like a goofy way to do things, remember that 35mm film does not record audio--every major motion picture records its location audio on a separate system and syncs them in post (or foley/loops new audio in post, or a combination of the two).
BTW, using two DVX100a's and one DVX100 on my last project, I did not have to trim a single frame due to drift (45 minute run time).
My $.02, FWIW.
Edit:
You can get all wrapped up in timecode issues if you want. I don't worry about it--if the images are synched together, I don't care what the numbers say or if they match. And I certainly don't rely on either external synchronization via timecode or any "record start" synching methods. I literally walk from camera to camera, hit record, then on to the next one. I might have 3 full MINUTES of record/TC discrepancy this way, but when I edit in post it takes me about 5 minutes to line up all the streams to frame-accurate to that one still-camera flash (or any other sudden movement/video event, like a clapper). No biggie.