Robert Harris on The Digital Bits - 12/20 column - OFFICIAL THREAD (1 Viewer)

[Mikey]

Hi All
And in reply to John and Michel,

The 34 million pixel figure was for a full frame 35 neg in std. still home photography (24 x 36 mm). As previously mentioned, the 100 line pairs per mm (= 200 pixels per mm)is the native resolution for the film itself. As the film travels from camera to your screen, resolution is lost and grain is increased (due to camera motion/jitter, shutter, camera optics, processing, duplication, etc..).

For a 1.85 aspect ratio neg of 20.95 x 11.37mm (assuming a film with native resolution of 100lp/mm):
200 pixels/mm x 20.95 mm = 4190 pixels
200 pixels/mm x 11.37mm = 2274
4190 x 2274 = 9.528 million pixels

Of course, keep in mind that one cannot strictly equate film resolution (analog) to digital resolution. Film has a 3 dimensional structure due to the silver halide crystals' (random) arrangement within the emulsion layer.
A CCD array or digital tv is 2 dimensional, fixed, and affected differently (usu slightly more severely) by Nyquist frequency limitations.

Film Image quality is really a composite of all the film's attributes including Speed, granularity, Resolution, Contrast, Acutance, Modulation Transfer Function (MTF is the normalized ratio of output contrast to intput contrast vs frequency).

Final image quality as seen on your screen is the cascading of several, and unfortunately slightly degrading steps:

Camera MTF x Film MTF x Processing MTF x Dupe film MTF x Duplicator MTF x Dupe film Processing MTF x Projector MTF x(MTF), etc...

Time to go watch a dvd...

 

[Michel_Hafner]

#Final image quality as seen on your screen is the cascading #of several, and unfortunately slightly degrading steps:

Slightly degading I would not call it. Loss of high
frequency content of release prints versus camera negative
is 50% or more. Add to that that many films are not that
sharp on the negative for various reasons and you end up
with release prints that are less sharp than HD.
The day wee see real 4K resolution projected in
regular cinemas is the day a lot of jaws will drop to the
floor since normal people have simply never seen anything
this sharp before.

 

[Michel_Hafner]

Mr. Harris,
I think it would be a very good idea if you would dedicate
a whole article to the subject "What a good DVD looks like
in my opinion".
Fact is that so many different opinions, reviews, statements
etc. are made about specific DVDs and transfers, by laymen
and (presumed) experts alike which totally contradict each
other so that one could get the impression that anything
goes. It's just a totally subjective opinion and there are
no criteria to reliably tell the good from the bad and the
ugly.
It would be helpful to see what you look for and what does
not bother you much (while it might bother others a lot).
Since you mentioned XXX as a good new transfer here's what
Widescreen Review says (it gets 4.5/5, not the highest
rating): ... Pixelisation is evident, causing minor image
breakup in finer details and creating a digital appearance.
There is also some edge enhancement noticed.
I have not seen this DVD (yet) but if it does have EE it
fails to provide a look as close to film as DVD can and
should provide.
The discussion applies to HD transfers as well. Shortcomings
of DVDs are usually also visible on the HD original to a
lesser degree, or digital intermediates used for print
production.
Example: While watching a 35mm print of "Panic Room" I
noticed scanline artifacts and in the end credits massive
aliasing problems. It turned out they used a digital inter-
mediate (of not acceptable quality IMHO). The aliasing is
also visible on the Superbit DVD and therefore on the HD
master made from (presumably) the 2K intermediate.
Another example: Lowry Digital Images cleans up material
with sophisticated noise reduction filters. Not
sophisticated enough to completely avoid adding digital
artifacts, though. Acceptable or not?

 

[Robert Harris]

Michel,

This is a superb idea, and we'll work toward it.

Re: Mikey's comments on line pairs, etc.

The list should understand that he is speaking of 35mm still photography, which in the cinematographic world is known as VistaVision.

On our current project we're scanning Vista frames at 6k, which resolves as a digital image of approximately 24 megapixels, or twice that resolution of the best of the currently available (35mm) digital still cameras.

We're getting their quickly as far as digital photography is concerned. It will be most interesting to be able to compare a digitally captured image against one captured on a film emulsion.

End credits on film are one of the most difficult to project properly. Because of their rate of movement and high contrast, they will reveal the inadequacies of shutter synchronization and focus, which may not be noticed in the projection of normal images. The ability to note scan lines or digital artifacts in the 35mm projection is based, unfortunately, upon the optics used in that projection, as well as the rest of the system, ie. alignment, etc.

2k, however, is usually enough, when properly processed, to create a more than adequate dupe neg for the production of prints. This is not to say that 2k will carry the total resolution of a photographed film image.

It will not.

As a final note, I wonder if Michel might care to offer a CV to the group?

 

[DeeF]

An excellent suggestion from Michel. Not only do we find contradictory opinions from the experts, but also, I have reversed my own opinion when showing the DVD on different equipment, using different user settings. It's almost impossible to judge the "absolute" value/quality of the DVD. Though this seems a bit of a conundrum, it is little different from the varying quality of projected film in the theaters. Just as aspect ratios change from theater to theater, so does contrast, brightness, focus, and quality of sound.

I'm also not sure if what I experience as a poor quality DVD might be exacerbated by the problems of plasma technology. Here's my equipment: Fujitsu 50" plasma (shows plenty of "false contouring" and banding), Panasonic RP-82 (with Faroudja deinterlacing chip) DVD Player, SA3100HD High-Def cable decoder box from Time Warner, Boston Acoustics 1000 surrround-sound system, and a line transformer/voltage meter for consistent voltage. It's a quality system, but it does enhance the problems on DVDs.

Three examples of what I consider poor quality DVDs.

High Noon -- though an improvement on its original release, this movie is too smooth, too much contrast, so that the actors look like Greek sculptures. It's just so apparent that way too much filtering has happened.

Oklahoma! -- This DVD arrived in 1998, and it shows. Non-anamorphic, with digital artifacts, odd shimmer, color shifts and other very distracting problems. The movie was the first Todd-AO production, and it deserves a restoration (Mr. Harris!) and a superb new transfer. (Speaking of this, would 2K really be enough resolution to properly represent these larger films? The Oklahoma! negative must be enormous).

To Catch A Thief -- The problem with this DVD is that the transfer seems to have been made from substandard elements, which may be the only surviving original pieces. There is just so much contrast that the faces literally look muddied. The picture is disappointing, but this is one movie where I could alter the user settings, and make what I thought was an acceptable picture, turning contrast way down, brightness up, basically washing it out.

To Catch A Thief looks grainy to me, but in a distracting way, grain on top of grain. Perhaps what I'm calling grain is really video noise, but I don't know enough about what I'm looking at to call out the difference. Alternatively, Minority Report and A.I. Artificial Intelligence are grainy, but this is very artistic use of grain, not distracting, but enhancing these films.

Excellent article, RH; I always look forward to your ruminations.

 

[Michel_Hafner]

Hi,
#On our current project we're scanning Vista frames at 6k, #which resolves as a digital image of
#approximately 24 megapixels, or twice that resolution of #the best of the currently available (35mm)
#digital still cameras.
The new Kodak has 14 Mio Pixels. The current world
record holder is 22 Mio Pixels (http://www.sinarbron.com/photokina/sb54kina.html)
#2k, however, is usually enough, when properly processed, to #create a more than adequate dupe neg for
#the production of prints. This is not to say that 2k will #carry the total resolution of a photographed film image.
The Panic Room credits looked all weird. Like the font was
not antialiased or scanlines were swapped. 2K can not be
super sharp but aliasing free when properly done. Something
went wrong with these credits.
#As a final note, I wonder if Michel might care to offer a #CV to the group?
I have a degree in business oriented computer science and
worked some years on a thesis about digital film
restoration when it was not yet really workable unlike
today.
The last ten years I have worked for www.imdb.com which I
helped founding. I write sometimes DVD reviews as a hobby
based on what I see in my pretty good home cinema.

 

[Robert Harris]

Re: DeeF's comments...

High Noon.

No problem here as the original negative survives.

Oklahoma.

A faded Oneg. Sep masters may or may not work. This film was most likely shot on 5248, which may well fit within the parameters of 2k, based however upon a larger scanned area of 65mm. We're talking ratios here.

To Catch a Thief.

This can look appreciatively better, but not based upon the current dupe neg used for the transfer. This may well be a problem film, with more inherent difficulties than meets the untrained eye.

 

[Rob Tomlin]

Film grain is not the enemy

Exactly!

Film grain is something that can be creatively controlled by the photographer/cinematographer in order to acheive a desired affect/mood. Mikey did a great job of explaining how the grain structure can be manipulated in his post.

With people like George Lucas beginning to use Digital cameras, I am very worried that cinematographers will lose the ability to creatively control grain, resulting in a much more "flat" look to these movies.

Excellent article once again, Mr. Harris!

 

[Mikey]

Seasons greetings!
Yes excellent suggestion Michel! I'd love to see a full treatise on dvd quality (from orig camera neg/dupe quality, to transfer charac, high freq rollof, MPEG2, etc..)
True, Michel, there can be some poor print quality out there as you mentioned. For example, looking at high frequency info, let us assume our camera neg has an MTF of 0.99 at 50cycles/mm and 0.90 at 75cycles/mm. Let us also assume the same MTF for the duplication step. Then each successive dupe generation would for MTF50 be(.99 x.99 x.99x.99x etc..). For the MTF75 the case would be (.9x.9x.9x.9 etc..). After a 4th generation the MTF at 50 cycles would be 0.96, but only 0.66 at 75 cycles/mm! It gets worse at higher frequencies.
I am curious on some input from folks who have seen the recent Star Wars Pt II that was shot digitally (1080 res?). How was the digital projection compared to the film projection? How did these compare to other recent all film presentations? I would venture to say that Star Wars might have been a bit soft compared to other recent movies.
Finally, for grain fans (Rain, Rob)... After just watching "Minority Report" it seems that Steven Spielberg/Amblin Entertainment is a big fan of grain, contrast, and desaturated color (MR, Sav priv Ryan, A.I.). Or did Amblin get a deal on a semi-trailer full of Expired camera negative film that was stored in Death Valley for a couple of summers and then repeatedly run through an industrial xray machine?
Happy Holiday!

 

[Rain]

...it seems that Steven Spielberg/Amblin Entertainment is a big fan of grain, contrast, and desaturated color (MR, Sav priv Ryan, A.I.).

Some of the "blame" undoubtedly belongs to the DP on those films, Janusz Kaminski.
Oh, by the way, he won the Oscar for his astounding work on Saving Private Ryan and I'm willing to bet he gets nominated for Minority Report, another very visually striking film.
I personally love the look of these films.

 

[John Kanan]

Interestingly, John Lowry, founder of the company behind the digital clean up of North By Northwest, Doctor Zhivago, Snow White, Citizen Kane, Singin' In The Rain,
and Sunset Boulevard, is no fan of grain-free DVD transfers.
He says: "Our goal is not to create grain free
images, but to create images that match the quality of the original film dailes where possible."
He talks about his company and its ever-improving technology here: http://www.hometheaterforum.com/files/lowry.txt

 

[Robert Harris]

The "goal" is not being achieved.

 

[DeeF]

RH --

Did you happen to catch the High-Def airing of "The Sound of Music" on Sunday night (ABC)? One wonders what you might have thought of the quality of the picture, transfer, etc., compared to what we see on the DVD (which is poor, by any estimation).

Despite the endless commercials, I thought the picture looked terrific, with bright colors and smoothly rendered transitions throughout. It was not in the original aspect ratio, but a simple 16:9, so some cropping was done left and right. But this is the best I have ever seen this movie look (I was 7 when I saw it in the theaters, originally).

Would this transfer have been the same one used on the DVD? If so, then I think the movie is in fine shape. Of course, I don't know how well the original elements have survived.

Hope everyone had a great holiday.

Dee Fletcher

 

[DaViD Boulet]

Dee,

did the opening credits have that horrid print damage with that red mark that is so distracting on the DVD?

Even without the 2.35:1 aspect ratio, it must have been a pleasant experience to view that film without that copious amount of EE/ringing present on the DVD version!

I *really* hope that Fox revisits the Sound Of Music done right. Maybe once they start taking their classic musicals more seriously we might see some action. In the meantime...I just PRAY that the Hello Dolly DVD does the film justice!!!

-dave

 

[Rain]

The "goal" is not being achieved.

Well, perhaps we should stay optimistic.
The fact that Mr. Lowry is seemingly aware of the importance of maintaining the look of the original film and (including the grain structure), and has referred to this as a "goal," could be a positive thing for future transfers he and his company end up working on.
Hopefully.

 

[Jon Robertson]

I completely agree in regards to the "look" of Minority Report.

What struck me during my viewing of this film was how the graininess served to hide the "seams" between the filmed images and the CGI, much in the same way it would help to hide certain "effects" in older films (as was briefly mentioned in this article).

One of my favourite examples of the benefits of grain is on Blue Underground's superb disc of Ken Wiederhorn's much-underappreciated Shock Waves.

The transfer is from a less-than-ideal print source: the super-16mm camera negative went missing decades ago, so a 35mm blowup theatrical print, provided by the director himself, was used and grain runs riot throughout much of the picture.

However, it serves to make the sunny tropical settings more than a little menacing and eerie, and one shot at the 41-minute mark, when a character is wading through the swamp, there is a perfect shot of a member of the Death Corps watching him through the dense undergrowth. He is partially hidden by the camouflage and shadow, but also the grain allows him to blend almost invisibly into the surroundings. Only when he begins to move is he truly visible, and, let me tell you, it is a spine-chilling moment!

 

[Mark Zimmer]

I often see segments of DVDs where the background is alive with highly distracting sparklies. I'd assumed this was either video noise or improper compression or both. But now Mr Harris tells us that this may be part of the original film?

Like film grain, a bit of minus density dirt, known as "sparkle," is part of that film image, and is perfectly acceptable transferred to DVD.

Or is this some other phenomenon altogether? Perhaps Mr Harris could indicate DVDs where the sparkle he's referring to is visible. I have to say, I don't ever recall movies in a theater looking like this.

 

[Robert Harris]

"Sparkle" can be seen in the Rear Window and To Catch a Thief dvds.

It refers to just an occasional tiny pinpoint of dirt which shows a quick almost imperceptible white spot.

This is nothing obvious.

 

[Jim Peavy]

I heartily agree with Rain's qoute above, but it seems like many are misunderstanding the topic to the extent that they label some "grain fans".
When me and my buddies would record movies on VHS back in the day, we could always tell the difference when a station played a 35mm print or a (more than often) 16mm. We of course loved the 35mm, for it's increased clarity and, yes, less grain. Grain may not be the enemy, but it's not a desired option, either. It's not like, "yeah! The more grain the better! And while you're at it, throw in some film splices and missing frames as well. That'll make it look even more 'film-like'!" Certainly no one would agree with this.
It seems appropriate to tell film restorers to just respect the look of the original film. When repairing the damage accumulated over the years to a film source, don't throw the baby out with the bath-water and attempt to "smooth" everything out, including the limitations of film stock of the day, which the filmaker no-doubt took into account when filming.
On the other hand, many modern filmakers (and some not so modern), regardless of the advances in film stocks to minimize grain, use techniques to increase grain for a certain cinematic effect. This too should be respected when transfering film to a video format.
It's also true that some may not like the look of a cinematographer's increasing the grain (as in the case of Minority Report), which is also legitimate. There are many artist's painting styles I don't care for, but I would never attempt to change them to my own tastes. That's what the artist intended it to look like and I respect that, even if I don't like it. This is merely a personal style, and to like it (or not) is neither right or wrong.
Anyway, my 2 cents. And while we're on the subject, I've always thought Universal's DVD of Bride of Frankenstein showed excessive grain. Anybody care to comment on it specifically?

 

[MatthewA]

I refused to watch the ABC airing of "Sound of Music" for three reasons:

1. Endless commercials

2. Modified aspect ratio

3. I'm going to see the new 70mm print at the Academy screening on February 3rd (I'm going to be in L.A. visiting my grandparents).