Super 16 D.I. ... Including 4:4:4 Super 16 D.I.

I've found some great info in the archives about D.I.'s and Super 16 blow-ups, but I still have some questions. I've been making the rounds at Cinesite, Technique, and Efilm for a Super 16 feature that I recently shot. One of them wants to try and save us some money by using D5 instead of raw data, but they insisted that it would NOT be a quality issue at all with 16mm. The guy we spoke to told us that in 16mm, you're never getting a real 2K scan anyway, that the data is only as good as HD to begin with (or something like that). It's always an HD scan, and then it's a matter of how you move the media about. Is that true?

I believe what he said is that our negative would be scanned on a 2K machine, but then the data would be stored on D5 because it would be a cheaper way of bringing the data in-house than by using the raw data (this particular company has to contract the 16mm scans out of house). But again, he said the quality would be the same. And this is a VERY reputable house, so I want to believe it. But everything else I've read about D.I.'s definitely states that D5 is NOT as good as raw data for 35mm. But is there any difference for 16mm for some reason.

Do the currently available processes for scanning 16mm already limit you to data that is just as good stored on D5 as it would be if it were stored raw? And is having a scan stored on D5 different than having a negative telecined to D5? And does my rambling make any sense at all?

Thanks,

Erin Harvey
L.A. cinematographer

In terms of pixel resolution, a 2K scan (2048 x 1550?) is only slightly bigger than an "HD" scan at 1920 x 1080. It's true that a Spirit Datacine will scan at 1920 x 1080 and uprez this to a standard 2K data format. But there's still a difference between a 2K 10-bit 4:4:4 RGB data file and a HD-D5 video recording, which is 10-bit 4:2:2 with some mild compression -- the difference is more in color information than resolution.

So I wouldn't say that there's no difference between 2K data and 24P/1080 HD-D5. The issue for you is if you are happy enough with the quality of HD-D5. The difference may not be enough in your mind to warrant spending more money on 2K. But there is a difference. Now if we were talking about a 4:4:4 HDCAM-SR hi-def recording versus 2K data, you might be able to argue that you're getting closer to a comparable quality level.

Others will be answer this question more accurately than my layman's attempt!

David Mullen, ASC
Los Angeles

Thanks for the info David. I was hoping you would chime in because I've seen so many other great posts by you about this subject. I have no idea what 4:4:4 and 4:2:2 means, but I do know from perusing the archives which one is better, so I guess I'm getting somewhere.

You said, "But there's still a difference between a 2K 10-bit 4:4:4 RGB data file and a HD-D5 video recording..."

Now when you say HD-D5 video recording, is that different than simply having data recorded onto D5, or would it be essentially the same thing? The post house wasn't talking about a telecine to D5 - they were talking about scanning and then simply moving the data via D5. Are they the same thing? I remember when some old Radio Shack computers used standard audio tapes to store data. Some of that data might have represented music, but it was recorded in a very different way as data than it would have been as a standard audio recording. Is this the same, or is it impossible for D5 to record the full data that would come in a scan?

Thanks again,

Erin (digital beginner-intermediate) Harvey
L.A. Cinematographer

4:2:2 basically means that you have half of your original red and blue color information compared to keeping all of the green information, which also contains all of your luminance (brightness) information. So red and blue are "subsampled" to reduce overall data. A 4:1:1 recording format like DV keeps only a quarter of the original red and blue information.

I don't see how they are using D5 if not for the storage and cost savings from working in compressed 4:2:2 YUV video instead of uncompressed 4:4:4 RGB data. That's what makes D5 what it is -- it's not simply the physical tape medium. Otherwise, I don't see what the significant cost benefits are if they claim they are storing uncompressed 4:4:4 data onto a D5 tape.

Whenever somebody offers you two options and says one is cheaper but just as good, a little scepticism is a good thing (although sometimes they are correct, as with generic versions of expensive drugs.)

David Mullen, ASC
Los Angeles

Erin Harvey wrote:

>Is this the same, or is it impossible for D5 to record the full data that >would come in a scan?

D5 is a video format. It is not a data format.

Basically, you asked the facility for a cheaper alternative, and they're offering you one. That alternative is to use HD video instead of data. By doing so, you limit the color depth and thus the amount of information you have to perform color correction. You also "scan" into video colourspace, which is different than RGB colourspace, thus limiting the color palette in certain hues. How much this would affect your project depends on how much color correction is necessary, the color palette in the original piece, and your own level of satisfaction and expectation.

Mike Most
VFX Supervisor
IATSE Local 600
Los Angeles

I recently read that "Never Die Alone," Ernest Dickerson's film shot by Matthew Libatique, went through a DI that used an HD film-out (ICG magazine, April 2004, page 14). The fact that two ASC members were happy enough to go through this process is a pretty strong endorsement.

I thought the movie looked terrific, though at times it was a true celebration of 16mm granularity, so perhaps it isn't the best way to asses the true potential of this process. The article said that they did their work at Cinesite and I saw none of the aliasing, jaggies or color banding you might think you would be in danger of getting with this process.

Rob Barocci
SP - NYC

Rob Barocci wrote:

>I recently read that "Never Die Alone," Ernest Dickerson's film shot by >Matthew Libatique, went through a DI that used an HD film-out....

Not to be picky in terms of the terminology, but to me, an HD "DI" is basically finishing a project as you would a television program, then doing an HD tape to film transfer. This is considerably different than scanning film as data in a much wider colourspace that provides far more manipulative capability for color correction (thanks to 10 bit log encoding) and other effects (such as resizing) and a higher resolution (even if only slightly higher). As I previously stated, the amount of difference this makes in the final product, and whether or not the additional trouble and expense is warranted, is dependent upon the circumstances, and can only be determined by worst case scenario tests.

Frankly, my guess is that the original HD video version of the picture you mentioned likely looks far better than the film prints. But that's only a guess.

Mike Most
VFX Supervisor
IATSE Local 600
Los Angeles

I know "Never Die Alone" was in fact finished on 35 mm anamorphic from a 2.40 extraction of the super 16 frame.

I am wondering how is this done with a Digital Intermediate. Is it the same exact process as extracting 2.40 from asuper35? Do you just squeeze resize and output to film?

It's amazing to me that one can now end with a cinemascope print from a super 16 negative and have it look good.

Armando Estrada
Cinematographer
Gto. Mexico

Rob Barocci wrote:

>I recently read that "Never Die Alone," Ernest Dickerson's film shot by >Matthew Libatique, went through a DI that used an HD film-out....

I didn't read the article, but to the best of my knowledge, the film was scanned as data on a Spirit and output to Neg. The only HD involved was one made from the neg for DVD etc...

Christopher C. Pearson
Director of Photography
Sherman Oaks, Ca.

Thanks so much to all of you for your informative responses.

I now have a much better grasp of the technical issues I'm facing while taking our Super 16 negative through a D.I. We have now narrowed our search down to two facilities. But I am considering more than just the technical issues while trying to decide between them. I have seen a few 35mm films from each place. The films that I have seen from one looked great, and I could not tell that they were D.I. The films that I have seen from the other looked great most of the time, but then there were those scenes with colors that were a little askew or highlights that were a little weird that made the image look like a bad telecine to me. This is my biggest fear of D.I. So now I am faced with a dilemma:

I can choose facility A which will do a 2K scan and work from raw 4:4:4 data, but whose previous 4:4:4 projects looked a little "video" to me at times

-- OR --

I can choose facility B which will do a 2K scan transferred to 4:2:2 D5, but whose previous 4:4:4 works looked beautiful to me and indistinguishable from film

I have never seen a 4:2:2 D5 project from facility "B" before, but when comparing apples to apples, I definitely liked their apples better. From my visits to the facilities, I believe the difference in quality that I have seen with regards to the previous works is because of the relative skills and talents of the colourists. So would I be better off with a more skilled colorist working with an inferior data set, or would I be better off with the less skilled colorist working with the full data set? Which one would be better able to keep my film from looking like video? I realize that there is no "right" answer to this question. But to hear a few of your thoughts would help me get a handle on what's most important to me. With regards to the differences between Super 16 and 35mm, I would always prefer the quality of image produced by a more talented D.P. shooting Super 16 than a less talented D.P. shooting 35. Would this be a fair analogy to the difference between raw 4:4:4 and 4:2:2 D5?

Thanks again,

Erin (shot film and don't want it to look like video) Harvey
L.A. cinematographer

Erin Harvey wrote:

>So would I be better off with a more skilled colorist working with an >inferior data set

Yes. But that answer should be qualified with "if the inferior data set is not too inferior."

Mike Most
VFX Supervisor
IATSE Local 600
Los Angeles

Sorry I'm late into this discussion ... I was shooting and then drinking in the Crown in Belfast

Never believe less is more, more is always more.

4:4:4 gives you more room to maneuver.

If you have seen bad images from places working in 4:4:4 then that's just what they were, bad images, that doesn't mean that 4:4:4 isn't better than the 4:2:2 that you've seen and liked, it just means that you've seen bad 4:4:4.

If you intend to really exploit what a DI can give you in image creation then you NEED the 4:4:4, if on the other hand you are happy with a very simple grade you'll get away with 4:2:2 but of course you won't be able to resist the urge to tweak

Cheers

Geoff Boyle FBKS
Director of Photography
EU Based
www.cinematography.net

>So would I be better off with a more skilled colorist working with an >inferior data set, or would I be better off with the less skilled colorist >working with the full data set

It depends, as usual. If you plan to do heavy-post (effects, more color corrections) is better to handle has many information you can get from the film. Otherwise if you are going to cut and end up with a DVD or so, don't bother (4:4:4 takes about twice the memory space of 4:2:2) and go for the good colorist source.

I'm going to shoot a movie in S16 this summer, but I have no idea of where I could get a 2K scan in Italy or France. Any help?

Leandro Pedroni

Well, between the Spirit scan and the laser recording back to film, "Never Die Alone" had to be stored on something as something, whether as 2K data or as HD on videotape. It's not a direct path from the scan back to film. I also remember reading in "Millimeter" that they had decided to use HD-D5 as a cost saver, but I could be mistaken.

The crop to 2.40 and the 2X squeeze is done digitally just before recording the image to film. No anamorphic optics are necessary in that transformation, just at the projection stage.

Yes, it's amazing that we can crop lower-rez formats to 2.40 and "get away with it" (whether Super-16 or HD, etc.) but there are usually consequences (like more grain). However, there are times when the wider aspect ratio is artistically worth the other artefacts that might result.

I certainly don't hesitate to shoot HD on the F900 and frame from cropping to 2.40 -- I've tested it enough times to know that the hit in quality is not so bad. It's not like the average viewer looking at "Once Upon a Time in Mexico" said "hey, that's 25% less vertical resolution than HDCAM-to-35mm should have -- what's going on?" A simple thing like what filter you use or lens you use will also affect resolution, so it's not so clear-cut to the eye.

Uh, don't respond to that last section without changing the CML category & subject heading...

David Mullen, ASC
Los Angeles

>It's amazing to me that one can now end with a cinemascope print from >a super 16 negative and have it look good.

I briefly read the article on AC and I think I recall the use of 7279-even though '218 has already been available for quite a while. Perhaps '79 was purposely chosen for certain scenes(?)

John Babl

I find it extraordinarily funny to hear everyone speak of the digital intermediate color timing process as if they invented it, or it is a new discovery. Although the output back to film is new, and the potential for color correction greater cinematographers in television have been modifying color and dynamic range for YEARS. I recall sitting in a room with Mike Most many years ago doing somewhat the same thing that everyone has suddenly invented or embraced. It use to be that those in features would look down upon anyone in television who could modify an image in the middle of a shot. Welcome aboard!

Roy H. Wagner ASC
director of photography

>It's amazing to me that one can now end with a cinemascope print from >a super 16 negative and have it look good.

Well I've done it optically and it looked great. The DI takes away certain issues but adds others. But for S-16 to 35-Scope this is not a new thing.

Mitch Gross
NYC DP

Roy Wagner writes :

>I recall sitting in a room with Mike Most many years ago doing >somewhat the same thing that everyone has suddenly invented or >embraced.

Indeed. I don't think either of us could have gotten through "Cop Rock" without being able to do those things. I must confess that I too find it endlessly amusing to read AC or ICG articles in which the cinematographer discusses the miracle of Power Windows (a trademark of DaVinci, BTW, not a generic term, although it seems to have become one), as if it's something that was invented last year by Technique and Efilm.

Mike Most
VFX Supervisor
IATSE Local 600
Los Angeles

E. Harvey wrote :

>One of them wants to try and save us some money by using D5 instead >of raw data, but they insisted that it would NOT be a quality issue at all >with 16mm.

The biggest issue between a 2k DI and D5 is not the resolution. It's the latitude of the digital format.

As far as resolution goes, HD is more than sufficient to contain all the data you need to produce the highest quality film out to an academy projection aperture.

However, DI uses a 10 bit log color space which has a latitude closer to film. You can therefore preserve all of the highlight rolloff and shadow detail, and generate a file and a filmed out negative which can be printed down just like your camera original.

The D5 uses an 8-bit linear format which saves space but has only a 5 stop latitude, so you have to be extremely careful about scanning the image in, so that you are sure of which data you are preserving, and which data you are clipping off.

Also, if you need to do a severe color shift or exposure correction, you can very easily create banding or expose the data clipping (printing down to such a degree that you see the whites turn in to broad areas of flat color).

In my opinion, the 4:2:2 color compression is less of an issue unless you are doing blue or green screen work, then it will be an issue for your FX house. Severe color corrections may also expose the color compression as well.

(4:2:2 means Full rez Luminance (B&W) and half rez Chroma and Saturation) (4:1:1 means Full rez Luminance and quarter rez Chroma and Saturation)

Anyway, if you can afford 2kDI it's the way to go. If D5 is your only option, just be really careful - especially during the scanning.

Best of Luck!

Rachel Dunn
-------------------------
Cinematographer
818-904-1124
-------------------------

Rachel Dunn writes :

>As far as resolution goes, HD is more than sufficient to contain all the >data you need to produce the highest quality film out to an academy >projection aperture.

Rachel :

After enjoying many of your prior posts I am surprised that you make such a bold statement.

John Lowry

>After enjoying many of your prior posts I am surprised that you make >such a bold statement.

I was talking only about the number of pixels, not color resolution or latitude. Perhaps I should have clarified that point better.

HD Rez is 1920x1080 which is only about 120 pixels shy of the 2048x1107 needed for a super35 1.85 film-out aperture.

For an academy 1.85 aperture, you need an image rez of 1828x988 which falls well within the 1920x1080 resolution of HD.

Of course, these should be padded out to 2048x1556 before filming out, but, yes, I maintain that 1080p HD has more than enough pixel resolution to produce an extremely high quality film-out image for an Academy aperture.

Color resolution and latitude are a different story, which is why I suggested that if he decides to go the HD route, he needs to be extremely careful on the Image scanning, since he will be severely limited in how much he can alter the image without introducing visible artefacts.

I hope that clarifies my position.

Thanks,

Rachel Dunn
-------------------------
Cinematographer
818-904-1124
-------------------------

Rachel Dunn wrote :

>HD Rez is 1920x1080 which is only about 120 pixels shy of the >2048x1107 needed for a super35 1.85 film-out aperture.

The difference between the two is more like 200,000 pixels (1920x1080 = 2073600, 2048x1107 = 2267136). But I don't necessarily disagree with your overall point.

Mike Most
VFX Supervisor
IATSE Local 600
Los Angeles

An update to my original questions on this thread :

As it turns out, both houses I was talking to were going to be using 4:2:2 images. I was mistaken about the option for 4:4:4. One would be using raw data in 10 bit log space, the other would be transferring to D5. I have elected to stay away from the D5 altogether, even though both are 4:2:2. But now I'm curious about earlier posts regarding Spirit 2K scans and 4:4:4. All the information I've gotten from Efilm, Cinesite, and Laser Pacific suggest that there is no such thing as 4:4:4 data from a Spirit 2K. A Spirit 2K only offers 4:2:2 scans. Or am I missing something? As a reminder, this is a Super 16 project, so perhaps that makes a difference.

Erin Harvey
L.A. cinematographer

Dear Michael and CML friends,

What concerns me most is that many are venturing onto the digital intermediate playing field without taking advantage of what has been learned over the last 15-20 years in the digital darkroom that cinematographers worked through in television. There are numerous problems that I hear about that have "old hat" solutions. It is extremely important to learn from our past. I too applaud the digital intermediate path for their are boundless possibilities that are unavailable with photo chemical. Those options and possibilities grow everyday. What a great journey and incredible time to be practicing our Art.

Respectfully,

Roy H. Wagner ASC
Director of Photography

>4:4:4 SUPER 16 D.I.

Thanks for the continuing stream of great info regarding the Super 16 D.I. Thanks in part to the collective wisdom of those on this list, I have decided to stay far away from D5 for this project. I am now only looking at using raw data, most likely originating with a 4:2:2 scan on a 2K Spirit.

But I was surprised to find that there really is no option, at least that I've found in Hollywood, to do a 4:4:4 scan from 16mm. Everyone has sufficiently convinced me that a 2K 4:2:2 scan (or even an HD telecine) captures enough resolution for the Super 16 frame. But when it comes to things like color depth and latitude, most would say that 4:4:4 is better, even if it's only slightly so. Unfortunately, I don't even have the option of looking into the extra cost of a 4:4:4 scan because nobody even offers that as a possibility for 16mm. Why aren't more people asking to do 4:4:4 scans for 16mm? I'm using the exact same film that someone shooting 35 would be using. Just because I've decided to create an image that has more grain does NOT mean that I want to give up any more of my color rendition than a 35 project would.

I suppose it's because most 16mm productions can barely afford D.I. to begin with. But everything that I'm reading suggests that D.I. has helped create a new boom in Super 16 production. Currently, the D.I. choices for 16mm tend to be about $100,000 for an HD digital intermediate or $150,000 for a true digital intermediate with a Spirit 2K scan. But certainly those films with a budget of one million or more might like the option of a higher quality, $175,000 or $200,000 digital intermediate that captures all of the color information in 4:4:4. It costs about $95,000 for one of these post houses to get a 16mm gate for their scanner. But I would expect that something like that would pay for itself fairly quickly.

Am I the only one that wants this? Forget writing to your congressmen - write to your local post house and demand 4:4:4 scanning for 16mm!

Erin (RGreenB Party candidate) Harvey
L.A. cinematographer

"E. Harvey" wrote :

>But I was surprised to find that there really is no option, at least that I've >found in Hollywood, to do a 4:4:4 scan from 16mm."

Erin:

You must have mis-understood the salesperson. On second thought there are many sales persons who don't know what they can do.

EVERY Spirit in Hollywood can scan out dual link HD at 4:4:4 from Super 16. I don't know any facility that has a Spirit that does not have a 16 mm gate. Maybe Technique does not. But Hell they are owned by Thomson who makes the Spirit and can get one if they do not have one. Technique and the local Thomson office are 300 feet apart.

If they can record dual link 4:4:4 High Definition has nothing to do with having a gate. There are two methods to record dual link HD 4:4:4. One way is to record to the New Sony HDCam SR recorders that are just being delivered with the 4:4:4 card. The other method is to use a dual link disk array that can record Dual link HD 4:4:4. Most facilities that you would consider would have one of these disk arrays.

Contrary to anything that has been said before the only facility to have extensive experience with 16 mm Digital Intermediate is Cinesite. They have been doing it for a least 3 years with great success.

In additional to the mentioned feature "Never Die Alone" that was Super 16 to anamorphic 35 mm they are finishing Spike Lee's latest effort. It is reported that he selected Cinesite because they are the only one that can scan to data the 16 mm negative and keep it as 4:4:4 data through to the laser film out, and they are scanning on a Spirit with S 16 mm so it is being done on a regular basis at Cinesite.

Spike Lee's movie : "She Hate Me"
http://www.imdb.com/title/tt0384533/

(Insert disclaimer here... No association with Cinesite but appreciate their good work with Digital Intermediate S16 mm.)

Regards,

Bill Hogan

"E. Harvey" wrote :

>But I was surprised to find that there really is no option, at least that I've >found in Hollywood, to do a 4:4:4 scan from 16mm."

I think people are being very specific with their terms, when somebody says you can't get 2K 4:4:4 out of a Spirit, what they are saying is that the Spirit's RGB CCDs are not 'full bandwidth' vs the detail sensors, and are kind of 4:2:0 in size terms. All Spirit 1 machines are like this, can't do anything about it.

Other scanners/telecines have a different arrangement for capturing images and have more equal capabilities for RGB so fit better the '4:4:4'ness people are talking about, this includes the Spirit 2.

On 16mm, the lens gate assemblies on the Spirit 1 actually give 16mm a good chance of looking decent by effectively optically zooming the image 2X vs the 35mm, so the pixels from a 16mm is kind of the same size as a 4K scan, thus the grain is larger, etc, but its not like some people might say a '1K' scan, i.e. the image is mapped across the full sensor not a cropped half.

Fussy people would say that no 2K is 4:4:4 but should be a higher number but that's a different story.

E. Harvey wrote:

>But I was surprised to find that there really is no option, at least that I've >found in Hollywood, to do a 4:4:4 scan from 16mm.

Erin,

Don't get confused between 4:4:4, 4:2:2, 8-bit linear and 10-bit Log.

You can have your info recorded in 4:4:4 and still not have the info needed to produce a good negative.

The REAL difference you need to look for when doing DI is weather you are working in 10-bit Log space or 8-bit Linear space. (Both of which can be 4:4:4)

10-bit Log space (cineon format) has a latitude which more closely resembles film. It has 1024 logarithmic increments per channel in it's format, therefore it's capable of recording nearly the full latitude of the film. The typical size of a 2k file is 12MB and there is no compression available for this format.

8-bit Linear (video) has 256 increments per channel which are uniformly distributed between black and white, and therefore can only record about half of the tonal range of film. With lossless RLE compression and no alpha channel, typical file size is about 8 MB/2k frame)

In order to represent the full range of film in 8-bit linear, you must reduce the contrast of the scan so that it fits in the latitude of the 8-bit file. You will have a file which looks flat and muddy, just as if you were to view a cineon (10-bit) file on a video monitor (all of which are 8-bit)

Were you to stretch out the flat 8-bit file to cineon format digitally, you may introduce banding because the original recording format latitude will have to be stretched out to match.

In order to achieve a decent amount of contrast of the image in an 8-bit format, and reduce the likelihood of banding when converting back to cineon format for filmout, you are forced to start clipping of either end of the tonal scale depending on whether you want to preserve the detail in the highlights or shadows.

Once you eliminate this data, you will be unable to print the image up or down without exposing the clipped areas which would normally have image data on film.

Now don't get me wrong, 8-bit is a fine format for this kind of work, and 8-bit files are filmed out every day with excellent results, but you really have to be aware of the format's limitations, and that you may have some difficulty making drastic colour changes, and that you should be very conscious of what data you are preserving during the initial scan.

Thanks,

Rachel Dunn
-------------------------
Cinematographer
818-904-1124
-------------------------

Rachel, your description of 8-bit linear scared me more than "The Exorcist". I am definitely sticking to 10 bit log, if for no other reason than so that I can sleep better at night.

Still perplexed by the varying accounts I've heard about Spirit 2K's and whether they're 4:2:2 or 4:4:4. Even different technicians from the same company seem to differ on this point. But if it's really just a technicality, then I'm not so concerned.

Erin Harvey

Hi all interested.

Regarding 4:4:4 vs. 4:2:2 scans using Spirit:

There seems to be some confusion here. The frame size or format does not have anything to do with the bit depth. All though the Spirit 1 does use chrominance sub-sampling (Y-RGB) this should not be confused with 4:2:2 color space conversion.

The "front end" of the Spirit is 14 bit RGB (after combining the lum and chrominance samples) after which it is converted to 10bit which is dictated by the hardware architecture. This is why precise primary grading at the front end can utilize the max. bandwidth of the 10bit hardware. If scanned as RGB data in 10bit (as we do) you can compare it
to a 4:4:4 HD scan.

Interestingly enough a new and very viable workflow it seems would be to use the Sony SR 4:4:4 tape format as an alternative to the slow data scans.

A precise "technical" best light to SR 4:4:4 could eliminate the need for a neg. cut and re-scan all together. We will be testing this route as soon as the first machine can be delivered (2-3 weeks).

Bets regards

Kris Kolodziejski

Kris Kolodziejski wrote:

>Interestingly enough a new and very viable workflow it seems would be >to use the Sony SR 4:4:4 tape format as an alternative to the slow data >scans.

Provided you happen to like 10 bit linear images as a source. If you want log format from a Spirit (which puts more levels of color information in the portions of the curve that are useful for color correction, thus allowing retention of more detail at both ends even when doing significant color correction), you need to go the data route.

Mike Most
VFX Supervisor
IATSE Local 600
Los Angeles

So just to confirm what you're saying Mike...

It's more important to have 10 bit log data than it is to have 4:4:4? 4:2:2 recorded as raw data would be superior to 4:4:4 recorded on that new SR tape? I've been told that the new SR tape can record 4:2:2 with no compression or 4:4:4 with minor compression. Is it that minor compression that converts it from 10 bit log to 10 bit linear? That new tape has come up a few times with different post houses as a possibility.

Erin Harvey

Erin Harvey wrote :

>I've been told that the new SR tape can record 4:2:2 with no >compression or 4:4:4 with minor compression. Is it that minor >compression that converts it from 10 bit log to 10 bit linear?

Go to see :

http://bssc.sel.sony.com/Professional/markets/

production/productsite/hdcamsr/faq.html

What is the compression ratio?

In 4:2:2 recording the compression ratio 2.7/1
In 4:4:4 recording the compression ratio is 4.2/1

And with the additional SRW-1 unit : 444 is 2 to 1

Its on tape @10 bit (log or linear). Its just that log will require a LUT for viewing & manipulation.

Tom
Telecine Ass't.
Nice Shoes /NYC