I just recorded samples with a Hyperdeck shuttle @ 10 bit 422, vs the internal SXS 8 bit 420. Quick shots of a green screen and grey gradient background. I really can't see a difference. Under extreme color correction, they look the same, even zoomed in 500%. Am I missing something? I'm using Premiere 6, so is there a setting that I need to use, because I though there would be a dramatic difference.
8 bit vs 10 bit on ex1...what's the dif?
- Thread starter pulpwoody
- Start date
Barry_Green
Moderator
The benefit of 10-bit recording is really only evident if the signal is low-enough noise to show the difference. Any noise in the signal will smooth out the gradation and therefore mask any benefit you'd otherwise be getting from 10-bit. There was a review on, I think, DVInfo that tested the EX1 with 10-bit a few years ago and came to the conclusion that there really wasn't a point to bother recording 10-bit on it because the noise masked the difference.
Fantastic Barry, thanks for saving me a ton of wasted time and money.
starcentral
Veteran
There are lots of people who will tell you 10-bit is way better than 8-bit as it will prevent banding of gradients, hold up better in grading, and give you a more detailed image.
It's not to say that isn't true, but from shooting 8-bit 420 XDCAM EX on the F3 for the last 2 years I can't say I've ever noticed anything that really stood out or pissed me off with the image. Obviously if you press pause and zoom in 200-400% you can find differences in a side by side comparisons, or if you shoot some serious gradients perhaps then you can break the 8-bit codec trying to do its job - but for me I can't recall a time where reviewing footage or working with it in post I said, "Damn, I really wish I shot that in 10-bit, look at how bad that looks".
Some people shoot 50mbps or higher because some TV networks demand it as minimum spec and truth be told it will have smoother colours and hold up better in post when doing excessive grading, but how much of that becomes noticeable when everything is said and done? There is a difference, its just not night and day.
I have done some of my own tests putting 8-bit 420 up against 10-bit 422 ProRes and even uncompressed and the differences are subtle. Here are some tests done by other DVX members, take a look and see what you think:
http://www.dvxuser.com/V6/showthrea...-s-Comparison-(Matthew-Allard)&highlight=slog
http://www.dvxuser.com/V6/showthread.php?302433-8bit-XDCAM-(mpeg4)-vs-10bit-RGB-(ProRes444)
It's not to say that isn't true, but from shooting 8-bit 420 XDCAM EX on the F3 for the last 2 years I can't say I've ever noticed anything that really stood out or pissed me off with the image. Obviously if you press pause and zoom in 200-400% you can find differences in a side by side comparisons, or if you shoot some serious gradients perhaps then you can break the 8-bit codec trying to do its job - but for me I can't recall a time where reviewing footage or working with it in post I said, "Damn, I really wish I shot that in 10-bit, look at how bad that looks".
Some people shoot 50mbps or higher because some TV networks demand it as minimum spec and truth be told it will have smoother colours and hold up better in post when doing excessive grading, but how much of that becomes noticeable when everything is said and done? There is a difference, its just not night and day.
I have done some of my own tests putting 8-bit 420 up against 10-bit 422 ProRes and even uncompressed and the differences are subtle. Here are some tests done by other DVX members, take a look and see what you think:
http://www.dvxuser.com/V6/showthrea...-s-Comparison-(Matthew-Allard)&highlight=slog
http://www.dvxuser.com/V6/showthread.php?302433-8bit-XDCAM-(mpeg4)-vs-10bit-RGB-(ProRes444)
Barry_Green
Moderator
True 10-bit is indeed substantially better than 8-bit, if the camera head can deliver a signal clean enough to take advantage of the additional recording power.
BobbyMurcerFan
Well-known member
Another thing is that 10-bits really helps in the shadows. If the image is "properly" exposed, the difference is harder to notice.
Having used an EX3 for years now I can truly tell you that recording 10 bit 4:2:2 from the EX1 or EX3 is definitely, for sure and certainly much better than the 8 bit 4:2:0 XDCAM. Don't ever let anybody tell you otherwise.
If you were talking 10 bit 4.2:0 XDCAM vs 8 bit 4:2:0 XDCAM, then yes, I could see the "it's not much better" point. But keep in mind that you wont be recording XDCAM and you wont be recording 4:2:0. The whole package do make a lot of difference.
If you were talking 10 bit 4.2:0 XDCAM vs 8 bit 4:2:0 XDCAM, then yes, I could see the "it's not much better" point. But keep in mind that you wont be recording XDCAM and you wont be recording 4:2:0. The whole package do make a lot of difference.
starcentral
Veteran
So if its "much better" can you explain why the poster did not see much difference between it and 10-bit uncompressed in the case he described? It's not going to be "much better" in all situations and a lot will depend on what you are shooting and the actual detail within the scene.
Lastly, when you throw your video on the web in some inferior codec and resolution than you even shot with in the first place, I can tell you for sure and certainly that it really won't make much, if any difference at all.
Last edited:
BobbyMurcerFan
Well-known member
How can he explain what someone else saw??
starcentral
Veteran
Well he can't obviously explain what someone saw or didn't see, but does he suspect that perhaps the observations of others are simply a lie?
I don't agree with what's been posted about the noise levels and its relationship to the visibile differences between 10 and 8 bit.
Noise levels affect the blacks and shadows. We would see problems with 8 bit material on projection systems (Barco, Sony and Christie) even in midtones and especially in the highlights. This was usually a problem with the workflow - somebody basically f***ed up up stream and worked off of or rendered 8 bit sources. Yes, it's very hard for the human eye to see differences between 10 and 8 bit. But they are there and become more visible when there are problems with the source that are exaggerated in grading or image processing.
Noise levels affect the blacks and shadows. We would see problems with 8 bit material on projection systems (Barco, Sony and Christie) even in midtones and especially in the highlights. This was usually a problem with the workflow - somebody basically f***ed up up stream and worked off of or rendered 8 bit sources. Yes, it's very hard for the human eye to see differences between 10 and 8 bit. But they are there and become more visible when there are problems with the source that are exaggerated in grading or image processing.
MattDavis
Well-known member
The enemy of 'good' is 'great'. For the most part, 8 bit is fine - absolutely fine. Especially when your rushes will look 'sort of' the same when it gets to the audience.
However (with my editor hat on), 8 bit runs out of puff very quickly if you need to match shots from different locations or different times of day.
8 bit is great for news, event coverage, docco, reportage. Essentially, stuff that doesn't need to be messed around with. Good 8 bit that doesn't need work, repair or repurposing is fine, always welcome. It's what I shoot for my little niche of the Corporate world and it serves me well. It will project onto big screens and look good, so long as you're not stylising it or correcting for seriously big fubars.
10 bit gives you a lot more to play with - when well shot and exposed, you can do a lot with it to bend it into your look and feel. But it has to be real 10 bit. No point shooting an 8 bit output on a 10 bit device. All you've got is more tarmac in which to crash and burn. You can pull back some Foos from the brink, but not the FUBARs.
12 bit is serious territory that I'm still exploring (BMCC). For those of you who deal with Red footage, you'll know the score. Everything is there, it's now down to the Post guys to do the stuff the DoP used to do, but one still needs to practice 'digital hygiene' - please peg your 12 bits and 2.5k at the mid-point, and we'll work magic. Fail to park it at any recognised point, and we're just using all that data space to bring us back to where we'd be with well exposed and well shot 8 bit.
However (with my editor hat on), 8 bit runs out of puff very quickly if you need to match shots from different locations or different times of day.
8 bit is great for news, event coverage, docco, reportage. Essentially, stuff that doesn't need to be messed around with. Good 8 bit that doesn't need work, repair or repurposing is fine, always welcome. It's what I shoot for my little niche of the Corporate world and it serves me well. It will project onto big screens and look good, so long as you're not stylising it or correcting for seriously big fubars.
10 bit gives you a lot more to play with - when well shot and exposed, you can do a lot with it to bend it into your look and feel. But it has to be real 10 bit. No point shooting an 8 bit output on a 10 bit device. All you've got is more tarmac in which to crash and burn. You can pull back some Foos from the brink, but not the FUBARs.
12 bit is serious territory that I'm still exploring (BMCC). For those of you who deal with Red footage, you'll know the score. Everything is there, it's now down to the Post guys to do the stuff the DoP used to do, but one still needs to practice 'digital hygiene' - please peg your 12 bits and 2.5k at the mid-point, and we'll work magic. Fail to park it at any recognised point, and we're just using all that data space to bring us back to where we'd be with well exposed and well shot 8 bit.
starcentral
Veteran
Here's an 8-bit originating frame shot on the F3 in slog mode and post done to de-SLOG it and add some contrast and just a tweak to the coloring a bit.
We can see a lot of dithering, or dots in areas where there are tones changing shades ie. arm, chest, forehead, the wall between the bed and the chair in the background.
Is it noise? characteristics of the sensor? fact it was shot in 8-bit and pushed? limitation of 4:2:0? I'd be interested to hear peoples view on that.
We can see a lot of dithering, or dots in areas where there are tones changing shades ie. arm, chest, forehead, the wall between the bed and the chair in the background.
Is it noise? characteristics of the sensor? fact it was shot in 8-bit and pushed? limitation of 4:2:0? I'd be interested to hear peoples view on that.