|
|||||||||
|
Thread Tools | Search this Thread |
November 19th, 2014, 07:46 PM | #1 |
Trustee
Join Date: Oct 2012
Location: Sydney, Australia
Posts: 1,149
|
One point of failure vs many?
Three situations:
-- Shooting the entire day on one 64gb card -- Shooting it on two 32gb cards -- Shooting it on four 16gb cards Which is best? Or are they all equally good/bad? In the case of just one card, if you have card failure, the consequences are catastrophic. But in the case of four cards, presumably you've now quadrupled your risk of card failure, all else being equal. |
November 19th, 2014, 08:31 PM | #2 |
Inner Circle
Join Date: Jul 2009
Location: Perth, Western Australia
Posts: 8,441
|
Re: One point of failure vs many?
Hey Adrian
You are obviously shooting with more than one camera but really, if you have 4 cards fail as opposed to one then you are buying bad cards surely?? I simply use a card that will give me enough space for the amount of footage I'm likely to be recording ... my main cam that does the ceremony and speeches has a 32GB card (in case speeches go on and on (I had 10 of the buggers on Friday's wedding!!) but my other two cams do fine on a 16GB card. You obviously don't want to change cards mid event either!! Chris |
November 19th, 2014, 10:01 PM | #3 | ||
Major Player
Join Date: May 2008
Location: Dayton, OH
Posts: 307
|
Re: One point of failure vs many?
Quote:
In your case, if you have the discipline to not lose things, I would say 4-16gb cards, under the assumption that if one of them fails, you still have 3/4th of your footage. Quote:
BUT... Once you have one card fail, the odds of another failing are 1:1000, so its still best to have dual recording |
||
November 19th, 2014, 10:14 PM | #4 |
Trustee
Join Date: Oct 2012
Location: Sydney, Australia
Posts: 1,149
|
Re: One point of failure vs many?
Yeah, dual card recording is obviously best. But, Vince, I've got a feeling you've done the wrong equation.
If the chance of a card failing is 1/1000, and if you use four cards, then the chance that at least one will fail is 1/250, isn't it? So the likelihood that something will fail increases the more cards you use rather than decreases. (Disclaimer: last time I did any sort of probability thinking was longer ago than I'd care to remember. So I might be completely wrong!) Last edited by Adrian Tan; November 19th, 2014 at 11:05 PM. Reason: Fixed dodge italics attempt |
November 19th, 2014, 10:24 PM | #5 | |
Major Player
Join Date: May 2008
Location: Dayton, OH
Posts: 307
|
Re: One point of failure vs many?
Quote:
As I recall, you need to be careful if they are independent of each other. Look at 4 sets of dice, and assume rolling a "1" means failure. If you roll dice#1, the odds of getting a "1" is 1:6. If you then roll dice #2, the odds of getting a "1" is still 1:6, etc. Now, lets roll 2 dice at once. The odds of getting "1" and "1" is 1:36 (two failed cards) |
|
November 20th, 2014, 01:00 AM | #6 |
Major Player
Join Date: Nov 2014
Location: Charlottesville
Posts: 255
|
Re: One point of failure vs many?
I think the best way to deal with potential card failure is to have some sort of backup system (if possible). If your camera has a clean HDMI out then an Atomos Ninja Star or Ninja 2 will offer higher quality ProRes or DNxHD codec while also allowing the camera's internal recording to serve as a backup. The higher bitrate 4:2:2 codec will help a LOT when color grading and this way you've got a backup to give you peace of mind. If your camera doesn't have a clean HDMI output then I would say go with a good brand of digital media like Sandisk. I've read about many people saving a few dollars on brands that are less expensive and then they regretted it when they lost footage.
I'm a big fan of using a backup/safety shot as well. I bought a Canon G20 which I just place on a tripod near the back of the ceremony, set auto iris and auto focus, and then hit record. I usually pick a fairly wide shot that I can cut to when our manned cameras don't have a good shot, but it also acts as a backup in case our B Camera (Canon 70D) has an issue. The 70D does not have a clean HDMI out and only can record to a single SDHC card so if the 70D had a card issue then we could still get away with the G20 as our second shot instead of just having our A Camera to work with. Our A Camera (C100) has dual SDHC slots so it simultaneously records to both cards so that I always have a back up. This is the main camera with the groom's audio going to it so it's more important than the 70D in many ways. I don't think there's any way to guarantee you won't lose data, but I think that there are measures that can be taken to make it much less likely. |
November 20th, 2014, 01:59 AM | #7 |
Major Player
Join Date: Sep 2012
Location: Belfast
Posts: 823
|
Re: One point of failure vs many?
For ceremony and speeches I have two much cheaper video cameras on light stands rolling continuously on wide angles. This way a B&G can't complain that I never provided their wedding. This is vital in my book.
Even if my card on my Sony A7S failed, the B&G still have their vows and speeches. So I wouldn't get my ass sued quite so badly... But yea, I regularly see togs backing up cards over the meal. Myself however, I go home that night, set the back down, and copy the cards to the hard drives whenever I get a chance. ...I feel I should take some more care... perhaps take the laptop and start backing up at downtime on the day. As for the probability question. If one card failing is 1 in a 1000, the probability of any of your four cards failing is 1 in 1000. The probability of all four failing is very very slight. I get the natural logic that the more you have, the higher the risk of incident - but I'm not sure if it is true... |
November 20th, 2014, 02:17 AM | #8 |
Inner Circle
Join Date: Jun 2007
Location: Belgium
Posts: 9,510
|
Re: One point of failure vs many?
Using different smaller cards throughout the day instead of one big card is an equal risk, if any of those cards would fail you won't be able to provide the client a finished product, maybe the client told you their church is THE most important part of the day so you change cards before the ceremony and when that card dies...There goes your backup strategy.
The safest way is still recording to 2 cards simultaneously, in that way you don't need to swap out cards and use 2 big cards that will cover you all day, the risc that both cards would die at the exact same moment is like winning the lottery. |
November 20th, 2014, 02:52 AM | #9 |
Trustee
Join Date: Oct 2012
Location: Sydney, Australia
Posts: 1,149
|
Re: One point of failure vs many?
Re probability question, I had a think over dinner and dimly recalled this: to work out the probability of at least one failure, that's 1 - (probability of no failure). Does that seem right? In this case: 1 - (999/1000) to the power of 4.
I think some people are suggesting that, given a number of dice throws, the chance that at least one of those is a 6 is always 1/6. But surely that's incorrect? What if you have a million throws? Still only 1/6 chance that at least one is a 6? |
November 20th, 2014, 02:58 AM | #10 |
Major Player
Join Date: May 2008
Location: Northampton, UK
Posts: 915
|
Re: One point of failure vs many?
More cards increases your points of failure. However, it reduces your loss should a card fail.
A Single card with everything is just suicide. If that card fails... you loose EVERYTHING. We shoot dual slot and a single card for all and then change cards in slot B. In a worst case scenario we only loose whatever is stored on that one card and as we change between each major part of the day we loose just one part of the day (but have multiple cameras anyway). If you shoot single card and you are lucky enough to get a card which NEVER fails then top stuff. You save money, you make capture easier. But the thing with cards is... you don't know if you have that magic card. Its a lottery. My advice is break the day up into many smaller cards. Each card we buy gets a full test before we use it by recording non stop until its full. Format and go again. We did this with our last batch of cards and found the 64GB cards were faulty and would crap out when half way full. Luckily we have dual slot but what if that was my only source? You also have to remember. As careful as you are the camera itself can cause damage or they can get stolen. By breaking it onto smaller cards again you only loose whats not in the camera.
__________________
mintyslippers.com |
November 20th, 2014, 03:27 AM | #11 | |
Inner Circle
Join Date: Jun 2007
Location: Belgium
Posts: 9,510
|
Re: One point of failure vs many?
Quote:
I have been thinking though about getting a atomos recorder for one camera but mainly to use it when I shoot for companies, then I only use one camera but for weddings it's a too expensive upgrade path. Like I said, even if you swap out several small cards throughout the day, if one card from you main cam dies you still we be able to deliver but your client won"t be happy at all if that one card contained critical footage. |
|
November 20th, 2014, 04:33 AM | #12 |
Inner Circle
Join Date: Feb 2009
Location: Norwich, Norfolk, UK
Posts: 3,531
|
Re: One point of failure vs many?
Still 1 in 6 as previous throws do not influence the outcome of subsequent throws. The dice doesn't know that it's throw 1,000,001 & that 6 hasn't come up in a long time.
|
November 20th, 2014, 04:51 AM | #13 |
Trustee
Join Date: Oct 2012
Location: Sydney, Australia
Posts: 1,149
|
Re: One point of failure vs many?
Yep, sure! That's the "gambler's fallacy" -- "I've had a lot of blacks in a row, so there's a higher chance that the next spin will be red." Previous outcomes don't influence the next spin of the wheel.
But asking about the next throw is a different question, isn't it? I mean, different to asking, "Given a random sample size of a million throws, what is the probability that at least one of those is a 6?" Surely the probability here is higher than 1/6? |
November 20th, 2014, 04:56 AM | #14 |
Inner Circle
Join Date: Feb 2009
Location: Norwich, Norfolk, UK
Posts: 3,531
|
Re: One point of failure vs many?
You need o balance the equation that the more cards you use the higher the chance that one is going to fail but the fewer cards you use the more critical it is if one does actually fail.
No DSLRs or CSCs have the option of dual card recording so with multiple cameras we always preferred to have one card in each camera & use it all day. Swapping cards leads to the possibility of loss. At least you know that the card is in the camera (until the camera gets lost:-). If you only shooting with a couple of cameras then an Atomos recorder on each would be worthwhile as transfer & logging after the shoot would be easier too as the cards could be ignored & just regarded as backups in case the recorder failed. |
November 20th, 2014, 06:02 AM | #15 |
Inner Circle
Join Date: Feb 2013
Location: LIncolnshire, UK
Posts: 2,213
|
Re: One point of failure vs many?
The trouble with these sort of threads is that you can get paranoid about everything. What happens if you have twin card cameras for everything, or change cards several times during the day, then your car fails between the church and the reception, or on the way to the church. Does anybody have a spare car and driver available in case of such a failure? What do you do if your main camera has two cards, but the camera itself fails in the middle of the ceremony?
I have been filming weddings for 30 years and have never lost a wedding yet through tape or card failure. It may happen one day, but I check recordings periodically during the day to make sure everything is normal. If a card failed at some point during the day, it is quite likely that data would be recoverable as it is should only be corrupted from the file that it failed on. I also carry spare cards so that of a card fails, I can immediately change it, losing only a few moments. I use more than one camera for the main parts of the day, so I won't be losing any sleep over failure. I also have a clause in my contract that says that I cannot be responsible for events beyond my control and I think that sudden card failure would be such an event. Roger |
| ||||||
|
|