Over the years there have been many, often heated debates over the differences between 8 bit and 10 bit codecs. This is my take on the situation, from the acquisition point of view.

The first thing to consider is that a 10 bit codec requires a 30% higher bitrate to achieve the same compression ratio as the equivalent 8 bit codec. So recording 10 bit needs bigger files for the same quality. The EBU recently evaluated several different 8 bit and 10 bit acquisition codecs and their conclusion was that for acquisition there was currently no advantage to be gained by using any of the commonly available 10 bit codecs over 8 bit because of the data overheads.

My experience in post production has been that what limits what you can do with your footage, more than anything else is noise. If you have a noisy image and you start to push and pull it, the noise in the image tends to limit what you can get away with. If you take two recordings, one at a nominal 100Mb/s and another at say 50Mb/s you will be able to do more with the 100Mb/s material because there will be less noise. Encoding and compressing material introduces noise often in the form of mosquito noise as well as general image blockiness. The more highly compressed the image the more noise and the more blockiness. It’s this noise and blockiness that will limit what you can do with your footage in post production, not whether it is 10 bit over 8 bit. If you have a 100Mb 10 bit HD compressed recording and comparable 100Mb 8 bit recording then you will be able to do more with the 8 bit recording because it will be in effect 30% less compressed which will give a significant reduction in noise.

Now if you have a 100Mb 8bit recording and a 130Mb 10 bit recording things are more evenly matched and possibly the 10 bit recording if it is from a very clean, noise free source will have a very small edge, but in reality all cameras produce some noise and it’s likely to be the camera noise that limits what you can do with the images so the 10 bit codec has little advantage, if any.

I often hear people complaining about the codec they are using, siting that they are seeing banding across gradients such a white walls or the sky. Very often this is nothing to do with the codec. Very often it is being caused by the display they are using. Computers seem to be the worst culprits. Often you are taking an 8 bit YUV codec, crudely converting that to 8 bit RGB and then further converting it to 24 bit VGA or DVI which then gets converted back down to 16 bit by the monitor. It’s very often all these conversions between YUV and RGB that cause banding on the monitor and not the fact that you have shot at 8 bit.

There is certainly an advantage to be had by using 10 bit in post production for any renders or effects. Once in the edit suite you can afford to use larger codecs running at higher bit rates. ProRes HQ or DNxHD at 185Mb/s or 220Mb/s are good choices but these often wouldn’t be practical as shooting codecs eating through memory cards at over 2Gb per minute. It should also be remembered that these are “I” frame only codecs so they are not as efficient as long GoP codecs. From my point of view I believe that to get something the equivalent of 8 bit Mpeg 2 at 50Mb/s you would need a 10 bit I frame codec running at over 160Mb/s. How do I work that out? Well if we consider that Mpeg 2 is 2.5x more efficient than I frame only then we get to 125Mb/s (50 x 2.5). Next we add the required 30% overhead for 10 bit (125 x 1.3) which gives 162.5Mb/s. This assumes the minimum long GoP efficiency of x2.5. Very often the long GoP advantage is closer to x3.

So I hope you can see that 8 bit still makes sense for acquisition. In the future as cameras get less noisy, storage gets cheaper and codecs get better the situation will change. If you are studio based and can record uncompressed 10 bit then why not? Do though consider how you are going to store your media in the long term and consider the overheads needed to throw large files over networks or even the extra time it takes to copy big files compared to small files. For field production I think the NanoFlash hits the sweet spot with compact, clean, low noise files that are easy to store and easy to work with.

Nicely done Alister. Although I would be interested in knowing what the EBU considers "commonly available". Based on what I am seeing from my work, I am trying to move away from DCT based compression codecs and more toward wavelet. There is an efficiency there and a quality there that I believe would allow us to compress a 10-bit signal into the space of our current 8-bit codecs and still receive a boost in quality and performance.

We are currently seeing that from the Cineform folks both in the edit suite and from Silicon Imaging's use of the Cineform RAW codec. Similarly, RED is doing amazing things with it's wavelet based codec. Essentially giving us 4K based recording (with 4 channels of audio) at ~145Mbps. My own testing shows that on HD material, 50Mbps Jpeg2k (Wavelet) is visually indistinguishable from 220Mbit DCT (DNxHD).

So while I agree that staying 8-bit makes sense using the codecs that are commonly used today, including mpeg2 and mpeg4 variants, once the concept of wavelet based recording becomes commonplace, we'll be able to rethink this and go to 10-bit or 12-bit acquisition without much penalty and potentially, a lot of gain.

I agree Perrone. In the future it is likely that we will be using 10 bit codecs.

Here is the full EBU document:

http://www.ebu.ch/en/technical/trev/trev_2008-Q3_HD-Prod-Codecs.pdf

And the EBU's summary:

A 10-bit bit-depth in production is only significant for post-production with graphics and after
transmission encoding and decoding at the consumer end, if the content (e.g. graphics or
animation) has been generated using advanced colour grading, etc.
For normal moving pictures, an 8-bit bit-depth in production will not significantly degrade the HD
picture quality at the consumer’s premises.

This statement was particularly telling:


For acquisition applications an HDTV format with 4:2:2 sampling, no further horizontal
or vertical sub-sampling should be applied. The 8-bit bit-depth is sufficient
for mainstream programmes, but 10-bit bit-depth is preferred for high-end acquisition.
For production applications of mainstream HD, the tests of the EBU has found
no reason to relax the requirement placed on SDTV studio codecs that “Quasi-transparent
quality” must be maintained after 7 cycles of encoding and recoding with
horizontal and vertical pixel-shifts applied. All tested codecs have shown quasitransparent
quality up to at least 4 to 5 multi-generations, but have also shown few
impairments such as noise or loss of resolution with critical images at the 7th generation.
Thus EBU Members are required to carefully design the production workflow
and to avoid 7 multi-generation steps.

Thanks for this explanation Alister.

The EBU recently evaluated several different 8 bit and 10 bit acquisition codecs and their conclusion was that for acquisition there was currently no advantage to be gained by using any of the commonly available 10 bit codecs over 8 bit because of the data overheads.....What an odd conclusion. I guess the EBU didn't try any Primary & Secondary Color Correction then ;-) In order to get a specific look I want get (The Super 16 MM Film look), I have to set Cine Gamma 1 in my XL H1's rather extensive image settings, then play with the Knee settings and tweak the skin tone settings (All in 8 bit) in camera. In post I immediately transcode after finishing the picture elements editing and go to DNxHD 220 X (The *X* stands for 10 bit) and go into Media Composer's Color Correction Mode. Sometimes it takes me days of fiddling looking at a warmed up monitor before I finally get the look I'm searching for. If I stayed in 8 bit I wouldn't get there. I've noticed a big difference in Avid between the performance of 8 bit CC and 10 bit CC. Episode 2 of my Internet Tv Series Please Stand By was shot in 25 Mbps HDV (4:2:0 Color space), so I need all the help I can get. I'm also now looking into Color in the Final Cut Pro Suite to see what it can do for me, since Media Composer's Color Correction leaves allot to be desired IMHO. Apple's Color correction suite looks quite interesting. Avid Symphony Color Correction is Awesome. There has been some talk of Avid rolling Symphony into Media Composer, as they did with Avid Xpress Pro.

.....What I also found immensely helpful was shooting with the XDR, which did three things:

A) Raise my HD Raster from 1440 x 1080 to 1920 x 1080.

B) Increase my color space from 4:2:0 to 4:2:2

C) Lower the amount of compression I was recording clips at. from 25 Mbps to 160 Mbps.

In conclusion the XDR will transform to 10 bit recording once uncompressed recording is enabled, since all HD-SDI is inherently a 10 bit log input and since uncompression will bypas the XDR's encoding engine - Voila ! Our digital cinema shooting will be amazing in 10 bit !

If you read the report you will see that they are quite specific about there being little difference between 8 bit and 10 bit for acquisition, however they do recommend 10 bit for any post production. Colour correcting from 8 bit to 8 bit will very often lead to banding and other possible artifacts while going from 8 bit to 10 bit will not.

From my point of view I believe that to get something the equivalent of 8 bit Mpeg 2 at 50Mb/s you would need a 10 bit I frame codec running at over 160Mb/s. How do I work that out? Well if we consider that Mpeg 2 is 2.5x more efficient than I frame only then we get to 125Mb/s (50 x 2.5). Next we add the required 30% overhead for 10 bit (125 x 1.3) which gives 162.5Mb/s. This assumes the minimum long GoP efficiency of x2.5. Very often the long GoP advantage is closer to x3.
.

Nice explanation Alister, thanks.
Am I right in thinking that the AVC-Intra from the HPX2700 for example should by your calculations be rated as top draw, as its 100 mb/s I frame codec is reckoned to be about twice as efficient as other I frame codecs, so equating to a 200 mb/s mastering codec in 10 bit?

Steve

Hi Alister:
This makes sense. I have to go read that EBU report.

Nice explanation Alister, thanks.
Am I right in thinking that the AVC-Intra from the HPX2700 for example should by your calculations be rated as top draw, as its 100 mb/s I frame codec is reckoned to be about twice as efficient as other I frame codecs, so equating to a 200 mb/s mastering codec in 10 bit?

Steve

I'd place it behind wavelet at the same bitrate and HDCamSR which is also i-frame mpeg4 but at 440 Mbps and not subsampled for color. So no, not top drawer, but better than most.

Hi Perrone:
It was because of the HDCAM SR MPEG 4 quality that we want to capture to MPERG 4 compression in our SD Card SSDR. If you folks like MPEG 2 Long GOP, then you're going to faint when you see how good MPEG 4 can look ! IMHO Mpeg 4 blows away Long GOP Mpeg 2, and we all know just how good Long GOP MPEG 2 can look.

I have not seen any information to corroborate Panasonics claim that their I frame only codec is 2x more efficient than Mpeg 2. Side by side comparisons that I have seen show 100Mb/s AVC-I to perform no better than Mpeg 2 at 50Mb/s. In addition AVC Intra's predictive coding has a real problem with certain types of in-frame motion, for example fine repeating patterns or areas of fine detail moving across the screen at slow speeds.

Hi Alister:
Yeah, that's Panasonic's MP4 not Sony's. MP4 can be *Far* superior to MPEG 2, but no one is bothering to experiment with high data rate hardware encoded MPEG 4 (Except us and Sony). BTW, you can cram so much more onto an SD card at 2 Gig or 32 Gig size (Not to mention SDXC 1 Terrabyte Card) in hardware encoded MPEG 4 versus MPEG 2. There's also one even better than MPEG 4 and it's called MPEG 7 friends !

The Panasonic implementation is interesting as it appears to gets less efficient the more detail there is in the picture, which obviously is undesirable. I regularly encode HD using Mpeg4 at 20Mb/s and it always looks very good, but it's a swine to then re edit because of the CPU overheads to decode. Perhaps in the future there will be hardware decoders cards that can work at high bit rates.

Hi Perrone:
It was because of the HDCAM SR MPEG 4 quality that we want to capture to MPERG 4 compression in our SD Card SSDR. If you folks like MPEG 2 Long GOP, then you're going to faint when you see how good MPEG 4 can look ! IMHO Mpeg 4 blows away Long GOP Mpeg 2, and we all know just how good Long GOP MPEG 2 can look.

Is it REALLY so difficult to license Jpeg2k compression? I'd MUCH rather deal with that in post than ANY mpeg4.

Very interesting Alister.
I think as well that the AVC-Intra is too short of data rate.
About the "10b recording" I don't know which are the expectations of the people.
There is more improvement going from 420/36Mbps to 422/100Mbps than the one going from 422/100Mbps to 10b Uncompress.
Some people expect the footage already color corrected.
rafael

Some people expect the footage already color corrected.
rafael

How True!!

You are quite right, the jump from 35Mb 4:2:0 to 100Mb 4:2:2 is really quite dramatic while the step from that to uncompressed 10 bit is really very small and very difficult to see, even after grading.

Hi Alister:
You wrote: "How True!!

You are quite right, the jump from 35Mb 4:2:0 to 100Mb 4:2:2 is really quite dramatic while the step from that to uncompressed 10 bit is really very small and very difficult to see, even after grading. "


....It depends on what your presentation medium is. The difference between 100 Mbps Long GOP and 10 bit uncompressed in digital projection on a large screen is significant. This difference is also significant when you output to motion picture film. If you are looking at a TV set then it's not so appearent.

I disagree. Projection tends to hide any small artefacts. Air currents, suspended particles etc tend to soften the picture by a small amount. It is my personnal experience that I see more issues in an image on a large monitor or TV than projected, even at 4k. I have seen my material projected on many occasions on to some very big screens, including the main theater at IBC and also at NAB and it allways amazes me how good it looks, even 35 Mb footage. Transfered to 35mm the film grain hides even more digital artefacts.

I have not seen any information to corroborate Panasonics claim that their I frame only codec is 2x more efficient than Mpeg 2.
There are so many variables that I don't think it's possible to claim any single, definitive figure for the improvement factor. Under some circumstances the figure may be better than 2x, under others nowhere near as much. Those variables include the material under compression, and the actual bitrates - compare AVC-Intra v MPEG2 I-frame only at low bitrates and you may get a widely different improvement factor to doing it at high bitrates. Note the EBU did not give detailed results of their tests - only their conclusions, that both XDCAM 422 and AVC-Intra are considered suitable for "future general broadcast HD acquisition".

Note the word "general" - they are not saying that better than either may not be desirable for real top end acquisition (assuming the camera front end is up to it). What they ARE saying is that both AVC-Intra and XDCAM422 are both very good, and trying to argue one is better than the other really is splitting hairs.

And note that the 2x claim infers like-for-like, so AVC-Intra v MPEG2 I-frame only. Generally, we're more likely to be comparing it to long-GOP, with the efficiencies that itself brings (and which vary with GOP length), so in very broad terms, the figures tend to bear out that there's not a lot to choose between AVC-Intra 100 and XDCAM422 OVERALL. That's not to say that one may look better in some circumstances, the other in different circumstances.
....What an odd conclusion. I guess the EBU didn't try any Primary & Secondary Color Correction then ;-)
What they did do was compress, decompress, then put in a series of shifts so the next compression would differ from previous ones - not just simply recompress the material without alteration. For XDCAM, they additionally made sure that original I-frames recompressed as difference frames, to deliberately tax the codec as much as possible.

As Alister says, this is for acquisition codecs - if you wish to do a lot of post work, then a better than either post codec may well be sensible.

Hi David & Alister:
I claim MPEG 4 is better than both the aforementioned codecs and compresses way, way more. In post production one can see a distinct difference between 8 bit and 10 bit acquisition, but I do submit that codec related exccentricities can certainly color, if not mask any advantages provided by 10 bit. I have not seen 4:4:4 12 bit acquisition, but I have been informed the MPEG 4 HDCAM SR codec is stunning. I can tell you the 4:2:2 version of it is stunning as well.

The SI-2K is 12bit acquisition. The Viper is 10-bit Log, Panalog is 10-bit log (from 14-bit linear), the Arri D21 is 12-bit linear.

Of course, these cameras are in a completely different league than the common "broadcast" cams most likely considered in the EBU test.

Hi Perrone & Alister:
My personal application is less broadcast and more Internet and Digital Cinema. I was very surprised by your analysis of large screen projection loss, and what you concluded from this Alister. (??) IMHO, it is precisely the "loss" factor in digital large screen projection, which you described well, that makes the differences between compressed and uncompressed easy to see. Since you're losing so much, you need to originate with as much information as you can possibly squeeze out of your camera.

.....Another reason for uncompressed is because such a signal is inherently less noisy, and when you have to go into a crippled video format like those currently employed on the Internet, then the less noise the better. As I think you pointed out earlier, encoders don't know what to do with noise. Noise in the signal doesn't encode well. Since my Tv series is on the Internet, I want to be able to take steps to improve what I consider to be not very good video quality. At least Adobe Flash and Windows Media Video have been steadily improving to the point where their HD professional codecs are quite acceptable now. I state *acceptable* but not great.

Without question, you guys always have the most interesting "conversations" when I'm doing my real job. I miss the best ones for sure. Alister pretty much said it all in the opening post though.

Hi Aaron:
Yeah, I'm sort of debating, deliberating, and inquiring because I'm also actively involved in developping my own dream box to capture video my way, so we will see what we can see.

I claim MPEG 4 is better than both the aforementioned codecs and compresses way, way more.
I'm not disagreeing with you, but it's necessary to define "better", silly though it may sound.

MPEG4 specifies many different subsets - including AVC-Intra, itself a specific of MPEG4 Pt 10 or H264 - see H.264/MPEG-4 AVC - Wikipedia, the free encyclopedia (http://en.wikipedia.org/wiki/H.264/MPEG-4_AVC) .

And you have to specify bitrate. It shouldn't cause any surprise that MPEG2 at a high enough bitrate will outperform MPEG4 at a much lower bitrate.

But what is likely is that an MPEG4 variant will offer comparable quality to MPEG2 at a lower bitrate, so in that respect it may well be seen as more efficient. But it comes at a price - and that tends to be complexity, more difficult and requiring more power to code/decode. Hence, better in terms of bitrate efficiency, worse in terms of editability.

Hi David:

You wrote: "But what is likely is that an MPEG4 variant will offer comparable quality to MPEG2 at a lower bitrate, so in that respect it may well be seen as more efficient. But it comes at a price - and that tends to be complexity, more difficult and requiring more power to code/decode. Hence, better in terms of bitrate efficiency, worse in terms of editability."

......Try 1.5 X that of MPEG 2 in terms of quality enhancement (Compared to Long GOP) @ a little less data rate (I don't know how much less yet, but it is less. I do know you can cram so much more data on a memmory card with it however.) I don't know anything about MPEG 4's editability. I do know there is a royalty to pay to the MPEG consortium *IF* we elect to offer any sort of MPEG 4 compression in our box. We pay more again if we offer built in DVD image authoring (MPEG 2 Again). We don't pay anything if we go uncompressed ! ;-)

.....We're including an above average data processing chip in our box, because we literally want it to handle complex instruction sets for things we can't even figure out how to program yet (Which is allot of things actually), at as rapid a speed as possible.

......The XDR/Nano is a complex and unusually sophisticated system when you think of it, because the designers made it to run all kinds of operations and programs they can keep on adding to at a later date. I don't know how many programs the XDR and Nano can hold in their memmory, but in theory, I suspect the software engineers can just keep on going for quite sometime adding new programs to their manufactured units until an internal memmory limit or a hardware limitation is reached. This is really a very fascinating way to design and build something. If you consider there probably is no 100 % foolproof way CD could of programmed everything possible into these units before an initial release anyway, so they would need to have a way to desin and impliment based on user feedback.

With regard to projection. My observation is that projection often hides small nuances and artifacts that are more visible on a large LCD or Plasma screen monitor. Now I am not saying that starting out with a better image is in any way less desirable when projecting, simply that projecting is not as tough a test as of image picture quality as some may think. That's my opinion at least, based on actual observation of my own material on both large screen (60") monitors and projected at both 2k and 4k onto 30ft screens.

Codec decoding is a very important issue when choosing codecs for professional video applications. I believe the way the RED system works with its convoluted post production workflow is it's biggest downfall. Certainly RED can produce stunning pictures but very few producers want to deal with the cumbersome workflow. I have several Sony HXR-MC1 mini cams. These use AVCHD which is a type of Mpeg 4. Working with the files from these cameras is a complete PITA. Yes there are advantages in terms of file size and on a mini cam being able to have small files and thus small amounts of storage is essential. But even though it means more complexity and a second box I use the NanoFlash as a recorder whenever I can as the Mpeg 2 files are so much easier to deal with in Post. It's also a happy bonus that the picture quality is better, but then the Nano runs at a higher bit rate so thats expected.

You can argue over the numbers, figures and bitrate and the intricacies of all the various codecs forever and it's often an interesting debate. My observations in my original post were based on my real world experience working with products that are available now. Right now I believe that the NanoFlash is sitting in the sweet spot of size, usability, compatibility and picture quality. I could quite easily record 10 bit uncompressed from my cameras, but the files would be massive with all the storage and handling implications that brings. But visually the difference between acquiring at 100Mb/s 8 bit and uncompressed 10 bit is just about invisible. Even after heavy post work, provided in the studio I do any renders at 10 bit the difference is so tiny that it gives me no desire to have to deal with bigger files or slower codecs for acquisition.

One observation I have is that as a FCP user I find the Apple ProRes HQ codec rather disappointing. It seems to be fairly noisy and for my broadcast productions I am now rendering to uncompressed and this is giving me a cleaner picture.

You can argue over the numbers, figures and bitrate and the intricacies of all the various codecs forever and it's often an interesting debate. My observations in my original post were based on my real world experience working with products that are available now.
The point is that the numbers generally back you up - providing you look at the whole subject, and not just focus on one factor - such as 10 bit. I fully agree with what you say.

Two things can give contouring/banding in a post workflow - insufficient bitdepth or too much compression. The first seems intuitive, the second much less so. What it means is that 10 bit is only really worthwhile if it doesn't compromise the overall compression.

I could quite easily record 10 bit uncompressed from my cameras, but the files would be massive with all the storage and handling implications that brings. But visually the difference between acquiring at 100Mb/s 8 bit and uncompressed 10 bit is just about invisible.....Hi Alister: Yeah, in your application, but not in mine. In Internet Web Encoding (Which is more of an art than it is an exact science), starting post from an uncompressed source and rendering to an uncompressed source if you require rendering (As I think you have already found out for yourself) yields a much cleaner delivery than working from a compressed source. In our web site, you can double click on the screen and stream full screen. This is programmed as a dual purpose. If you have a laptop or desktop with an HDMI output which can connect to your large HD flat screen TV, then when you stream my production you are receiving Blu-ray quality audio and video. We are going to start to use Apple's Soundtrack Pro, because it allows us to mix, monitor and build a true Dolby Digital 5.1, which we will encode to stream straight from our website for the audio portion of our video streams.

Hi Mark,
How are you acquiring your Uncompress material at the moment?
rafael

Hi Rafael:
I'm not. I'm waiting for the Flash XDR to be upgradeable to full 10 bit uncompressed hopefully, in the near future. Originally, Convergent Design published a broshure stating full uncompressed would become an available feature via a paid upgrade sometime in the future. This was last year. Right now if I have to render out something I know I'm publishing to web, then I do it to full uncompressed frames or I use Quicktime Reference (Amounts to the same thing actually). The camera I own is capable of full uncompressed HD output via HD-SDI, but I have had no recording facility to capture source uncompressed. I'm waiting to have this feature added to the Flash XDR.

Not sure if CD ever said they'd do 10 but though did they?
Also, am I being dim, but I never considered web publishing to really be the most demanding of quality material - surely anything that's good enough for HD (or even SD) TV is miles above adequate for anything on the web? What am I missing here?
Steve

Dear Steve and Mark,

Last year, we were over optimistic concerning our workload regarding feature development.

Yes, we promised, for the Flash XDR, an extra cost option, for recording uncompressed.

The uncompressed option allows for recording 10-Bit and this was priced as a $995 (US) extra cost option.

We try hard to listen to our customers and we try hard to develop features that are in the most demand, while balancing our obligations to fulfill our promises.

At this time, we have failed to deliver uncompressed with 10-bit support as promised.

I apologize for the delay.

Thanks Dan, missed that.
I remember looking at your explanation of your uncompressed scheme and noticing 2 things: first the data rates were so huge it was crazy, and secondly that you'd done tests and the difference between uncompressed and 100 mb/s GOP was small. Because of these 2 things I didn't look any further as for almost all purposes it didn't really seem worthwhile.
Steve

Dear Steve,

Once we had confirmation from others that our 100 Mbps Long-GOP was very close to uncompressed, "visually indisinguishable" as we like to say, we changed our priorities. There were so many other features that we had to implement.

Now that we have 140 and 160 Mbps Long-GOP as options, the very small difference is even harder to justify.

We also have to taking into consideration the difficultly of actually working with these huge uncompressed files. Some are prepared to work with these files, most are not.

Uncompressed is somewhere between 120 and 150 Megabytes per second.

At a minimum this is 7,200 Megabytes per minute, or roughly 432,000 Megabytes per hour (approximately 432 Gigabytes per hour). Note: these are just rough estimates.

This is a serious burden on any editing system and post process.

Dear Steve and Mark,

Last year, we were over optimistic concerning our workload regarding feature development.

Yes, we promised, for the Flash XDR, an extra cost option, for recording uncompressed.

The uncompressed option allows for recording 10-Bit and this was priced as a $995 (US) extra cost option.

We try hard to listen to our customers and we try hard to develop features that are in the most demand, while balancing our obligations to fulfill our promises.

At this time, we have failed to deliver uncompressed with 10-bit support as promised.

I apologize for the delay.,,,,Hi Dan: Not a problem at this time, but I'm hoping this will become available in the near future. I also hope RS-422 and the firewire interface on the Flash XDR will become enabled so I have the option to work with four 128 GB CF cards plugged into my XDR and intefaced into my Avid NLE via FW and using Avid's Super AMA, or I will edit the uncomp via proxies. With AMA you don't ingest anything, so digital cinema becomes a practical proposition at this point.

Dear Steve,

Once we had confirmation from others that our 100 Mbps Long-GOP was very close to uncompressed, "visually indisinguishable" as we like to say, we changed our priorities. There were so many other features that we had to implement.

Now that we have 140 and 160 Mbps Long-GOP as options, the very small difference is even harder to justify.

We also have to taking into consideration the difficultly of actually working with these huge uncompressed files. Some are prepared to work with these files, most are not.

Uncompressed is somewhere between 120 and 150 Megabytes per second.

At a minimum this is 7,200 Megabytes per minute, or roughly 432,000 Megabytes per hour (approximately 432 Gigabytes per hour). Note: these are just rough estimates.

This is a serious burden on any editing system and post process....Hi Dan:

I don't see the huge storage requirements as such a big deal anyomre, when you can get a 2 TB G-Drive in Montreal for about $450.00. Uncompressed was out of the question before due to storage requirements and HDD performance, but now with Avid's AMA, there's no longer any question about ingesting such large files anymore since you don't have to do that to deal with them.

But you wouldn't be able to edit uncompressed 10 bit of a single drive or G-tech drive pair. You will need a really good raid array, probably 4x fast sata drives with a top spec PCI controller as a minimum. Especially if you want 2 streams at a time (around 2Gb/s). Do not underestimate the kind of drive performance you will need to work with uncompressed. Also consider how you will manage your footage in the field. Even with 4x 64Gb cards you will fill them after less than 30 mins of shooting. Offloading to USB drives is not realistic as it would take over 2 hours to backup your 4 cards. Esata would be faster and take around an hour with a fast drive.

One of the key advantages of file based systems is the possibility of faster than realtime workflows. Going back to a slower than realtime workflow is a hit most don't want.

I've done several features uncompressed.

A single frame is about 8MB (1920x1080p, 24, 32bit*).

Do the math and it's nearly 700GB per hour.

The only problem for me is my 10TB RAID filling up and the TIME it takes to move this stuff off. Even with eSATA it's "come back tomorrow."

Uncompressed 10bit is key for big screen. For Web or eyePhone, even tee-vee? SD still looks fine.




*24 bit is smaller, but the studios add the alpha for no reason except ignorance

I've done several features uncompressed. <snip> Uncompressed 10bit is key for big screen. For Web or eyePhone, even tee-vee? SD still looks fine.
I don't usually get involved in religious discussions (arguments?), but I can't help myself sometimes.

I'm the Senior Video Engineer for all the Metropolitan Opera Presents telecasts. We send 1080i/5.1 to over a thousand theaters around the world, live, every couple of weeks. It's sent 8-bit compressed. Not only has the quality of the viewing been praised by every single review, but we just won an emmy award for it. I've also shot a number of shows in video for theatrical release, including a series of outdoor concerts in HiDef 3D shot this summer.

The extra two bits provide nothing to speak of except for a higher signal to noise ratio. The best state of the art video camera doesn't come close to the signal to noise ratio of 8-bit. (And ONE of my cameras costs over a hundred grand). Editing/grading/keying is another story, as you're performing mathematical changes to the bits. So you acquire in 8-bit and you add two bits (of zero) to make it 10-bit in the editor. That way the manipulations don't truncate some of the low bit info in the 8-bits you acquired.

Sometimes I want to just yell "HEY those of us that work in the real world really think these esoteric discussions are no different from the ones about needing $750 power cords for your audio amplifier".

But I try to remind myself I don't know everything, and that my needs aren't necessarily the same as others needs. I understand striving for the very best quality, even if others can't see it, but recording 10-bit uncompressed in the field strikes me as having far more down sides than up sides.

Billy

At least in my part of the woods, the customer is always right.

Not so much a religion as keeping one's options open.

So you acquire in 8-bit and you add two bits (of zero) to make it 10-bit in the editor. That way the manipulations don't truncate some of the low bit info in the 8-bits you acquired.

if I could just convince the 10 bit purists to read, re-read and actually understand this, it would short circuit a lot of this heated discussion.

i'd like to do a double blind study of my own where the 10 bit purists are forced to reveal in an A/B comparison (AT VIEWING DISTANCE) of which footage was acquired 10 bit uncompressed and which which an 8 bit compressed, like a Nano.

I certainly don't care what anybody else thinks about this topic as we all have our own worlds with clients calling the format shots.

I certainly don't do uncompressed for my health. My clients REQUIRE it. To refuse simply makes them go to someone who will.

If you want 10bit, solutions are well known.

If 8bit will work, you're in the right place.


Can't we all get along and just say "go with what floats your boat?"

Then we can get off this 8/10 RAMBLE (look how many pages are devoted to this "discussion") and get back on the subject at hand: CD's stellar products.

Not sure if CD ever said they'd do 10 but though did they?
Also, am I being dim, but I never considered web publishing to really be the most demanding of quality material - surely anything that's good enough for HD (or even SD) TV is miles above adequate for anything on the web? What am I missing here?
Steve...Hi Steve:
It is precisely because the web formats are so relatively low quality, that one must begin with as high a resolution possible with low noise, so that the resulting encoded web video looks as prestine as it possibly can.

I agree Daniel. The client doesn't know any better and it's not necessarily our job to educate them. I'll gladly shoot and post uncompressed for anyone willing to pay for it. I'll use Zeiss lenses or whatever else they want if they're willing to pay for it.

But what I choose to own and use for my own projects will more closely reflect the technology that I consider to be "good enough". Bang for my buck is the most important thing for me and the CD products are absolutely exceeding in that regard.

However, I'm still pursuing an uncompressed capture system for myself. Not to contribute to the 8 vs. 10 bit discussion but more to use my camera's dual link SDI and 4:4:4 Filmstream output.

If you all agree that 4:2:2 is much better than 4:2:0, then we can also agree that 4:4:4 is even better yet.

However, even if I had a 4:4:4 capture system, I'm not planning on giving up my XDR anytime soon. This box is awesome and getting better all the time.

Agreed, Aaron.

And "even" I'll say there's such little (as in VERY) visual difference between 4:4:4 and 4:2:2, but ask the green screen guys. There is no doubt you get a cleaner pull...yet, the average observer will never see it.

My motto: if the (end) viewer can't see it, it isn't a problem. "Good enough" is powerful medicine.

But you wouldn't be able to edit uncompressed 10 bit of a single drive or G-tech drive pair....Hi Alister: I run an Avid certified HP Workstation Laptop PC (Model nw9440). I have the nVidia Quadro graphics card ( FX1500 M but not the 256 MB version = I run a special version of this model with the 512 MB of DDR 3 on board video RAM ). My laptop runs on 4 GB of dual channel DDR 2 667 Mhz RAM together with a special version of system HDD @ 7,200 RPM. Now this baby is plugged into the special HP advanced docking station, which is equipped with Express Card *54* bus running a firewire 800 interface to a 1 TB Raid 0 array. My Avid system runs one stream of uncompressed HD already without studdering. I can do video mix downs if I need to pile on lots of lower thirds and effects, CC, etcetera.
You will need a really good raid array, probably 4x fast sata drives with a top spec PCI controller as a minimum. Especially if you want 2 streams at a time (around 2Gb/s). Do not underestimate the kind of drive performance you will need to work with uncompressed......A G-Raid 2 TB is enough. I will switch from FW 800 to an express 54 speed eSata interface card. No problem.
Also consider how you will manage your footage in the field. Even with 4x 64Gb cards you will fill them after less than 30 mins of shooting.....Nope. I have already made the decision not to purchase any 64 GB CF nedia unless it's offered to me dirt cheap. I will wait for the 128 GB CF cards.
Offloading to USB drives is not realistic as it would take over 2 hours to backup your 4 cards. Esata would be faster and take around an hour with a fast drive.....I was thinking a couple of 500 GB Nexto DI's. Soon the CF media will be @ 100 MBps, so not too long to wait. I will learn to integrate dump times into my production workflow.
One of the key advantages of file based systems is the possibility of faster than realtime workflows. Going back to a slower than realtime workflow is a hit most don't want....To be able to have uncompressed I am willing to accept certain considerations.

* It doesn't really matter if your system isn't quite fast enough to playback uncompressed HD or SD anyway. You put up with the studdering for editing purposes and then copy out your completed sequence to another CF card and put that into your XDR or Nano and then play it out in realtime. No problem. If you are delivering on the web or on Blu-ray, then you export via QT Ref into your DVD authoring program or Web encoder and/or render out to Blu-ray DVD .ISO and burn baby burn. Frankly, I don't see where there's a problem Alister. (??)

At least in my part of the woods, the customer is always right.

Not so much a religion as keeping one's options open.

Here I don't disagree with you at all. But note that your post that I responded to had nothing to do with pleasing a client, or keeping one's options open, it just said that "Uncompressed 10bit is key for big screen", which I wanted to point out was just your opinion (and that of your clients), not a fact, or based on any real world truths.

Billy

Hi friends:
I agree with most of what you've been saying about those 8 wonderful bits, but I was a bit surprised (Ha ! Ha ?) you folks had great difficulties seeing how or telling the difference visually between 8 or 10 bits. Man ! I can clearly see the difference in a big screen projection or on a Sony HD Monitor (Which I have) between the color depth and digital vibrance of the signal. I really don't find this difference hard to see at all. I just shot some live action Canon 24 F @ Long GOP 140 Mbps on my XDR and it looked so close to uncompressed that it's scary-real scary, but 10 bit uncompressed 4:2:2 HD wins :-)