Super 8 film on CNN ... Thursday Aug 1

[Tommy]

I caught the very tail end of a commercial-promo on CNN about a Richard Nixon special being broadcast 9pm ET on Thursday...if I interpreted correctly, it appeared to use extensive Super 8 filmed by someone in Nixon's circle. It will be interesting to see the scan quality....if anyone knows what lab did it.. please post.

[super8man]

Prediction - it will be roughly presented and perhaps even purposely poorly transferred (I was going to say even sublet to an intern to transfer to digital but an intern might actually add value to the process).

I hope I am proved wrong.

[carllooper]

I saw some of that footage last year when a doco was being proposed around it. Not sure where. It wasn't a particularly good transfer but it wasn't deliberately done poorly either. They appreciated the value of the footage, but at the time didn't quite have the expertise to do the best job of it. Perhaps what is going to air might be better than what I saw. I don't think it will be worse.

There has been a lot better work done with that material since I last saw it ... having just now done a search on it ...



I found the following - from a 3K transfer, and it's not half bad at all - it just goes to show what Super8 can look like if you treat it with some respect rather than the alternative gung ho approach that is more often used (not that gung ho transfers are a bad thing):

https://www.youtube.com/watch?v=NZi8N5Y1AFk

I've seen some beautifully restored Standard 8 on some World War II documentaries where they've put considerable effort into doing something half decent. They sometimes frame it as "The War In Colour" as if those shooting the original film (back when) were using some sort of secret new technology (colour film) that has only recently come to light. Ha ha. But despite that rather fascinating spin they nevertheless put some considerable love and care into the actual restoration of such films.

The evolution of digital restoration will continue to get better. Of that I have no doubt - if only because there is a rich archive of material that deserves it.

Carl

[Nicholas Kovats]

The Super 8 footage was scanned onsite by Jeff Krienes with his sprocktless KInetta scanner at 3296 x 2472 x 12 bits . Your original assessment of it's quality is interesting Carl, i.e. poor.

Here is the relevant section form Jeff's website, i.e. http://kinetta.com/news-nixon.html and here is the Our Nixon web page, i.e. http://ournixon.com/

Nicholas

---------- Forwarded message ----------
From: Jeff Kreines <jeffkreines@mindspring.com>
Date: Thu, Mar 21, 2013 at 6:35 PM
Subject: [Frameworks] OUR NIXON and small-format, high-resolution film scanning
To: Experimental Film Discussion List <frameworks@jonasmekasfilms.com>

The new film OUR NIXON, by Penny Lane and Brian Frye, consists only of archival film, and about 80% of it is the Super-8 home movies shot by HR Haldeman, Dwight Chapin, and Bob Erlichman (all of Watergate fame).

Rather than accept the conventional wisdom -- that HD is more than adequate for such a small format -- Penny and Brian decided to scan the original Super-8 film at 3296 x 2472 x 12 bits -- over five times the resolution of pillarboxed HD. It made a huge difference. With lower resolution scans, there are not enough pixels to properly reproduce the grain, and the grain is essentially beating against the pixels, which turns grain into an ugly video-like noise with none of the organic feeling real grain has.

There was a problem with scanning this original footage -- it couldn't leave the Nixon Presidential Library in Yorba Linda, California. But the archivist, Ryan Pettigrew, realized that researchers were permitted to scan original documents with their own scanners in the reference room -- and the rule did not prohibit film scanners.

So, fascinated by the project, I drove out to Yorba Linda with a Kinetta Archival Scanner in the trunk of my car, and scanned over 22 hours of Super-8 in the reference room (where we had to whisper!). We were able to capture the full dynamic range of the reversal original without compromising either shadows or highlights, which makes a huge difference.

A portion of the footage only existed as very contrasty Super-8 contact prints, which obviously stand apart from scans from the original footage. Also, a lot of the shooting is, in Variety-ese, subpar. But given those limitations I was very pleased when I finally saw it last week at a press screening at MoMA.

Our Nixon closes the New York Film Festival New Films/New Directors festival on March 31 at Lincoln Center. Below is a list of other showings.
SCREENINGS

Hot Docs, Toronto Canada (April 2013)
Full Frame Documentary Film Festival, Durham NC (April 2013)
Independent Film Festival Boston, Boston MA (April 2013)
New Directors/New Films, NYC (March 2013)
Atlanta Film Festival, Atlanta GA (March 2013)
Ann Arbor Film Festival, Ann Arbor MI (March 2013)
SXSW Film Festiva, Austin TX (March 2013)
International Film Festival Rotterdam, The Netherlands (January 2013)

There's more information at the Our Nixon and Kinetta websites. Let me know if you have any other questions.

I should note we also did scans (with As'Image in Paris) of the Super-8 footage in Jane Weiner's recent film Ricky on Leacock. Resolution and bit depth and dynamic range are important and shouldn't be sacrificed! HD is a delivery format, not a good capture format, IMHO.

Jeff Kreines
Kinetta
kinetta.com
jeff@kinetta.com


Disclaimer: I designed and built the Kinetta Archival Scanner, and it reflects my philosophy of how film should be scanned. Your opinion may differ.

[carllooper]

Your original assessment of it's quality is interesting Carl, i.e. poor.

No the Kinetta scans were superb.

What I was referring to, was some earlier scans done on the same film, which were poor. They weren't the Kinetta scans. The Kinetta scans are gorgeous. And prove my point that the more pixels you put in the better it looks. Basically film to video theorists, of the 70s and 80s, putting pixel figures on the nature of film were completely wrong. They were looking at 16mm film vs Super8 on their SD video monitors and theorising film completely incorrectly.

There is a very subtle kind of interference going on between film and video/digital, one that is only just starting to be understood. What small gauge films need in a transfer is not less pixels than 16 or 35, but a lot more!

I mentioned this over on cinematography.com in relation to a film grain question for Super8. Basically it's that film encodes an image in the spatial frequency domain, where there is a statistical correlation between film and the image it encodes that goes all the way down, even if we can't see it. With video or digital the information is in the form of an amplitude modulated signal where there is a cut off frequency beyond which it can't see the signal. What happens at this frequency boundary is that the signal in the film (which we can't see) gets converted into actual noise on the digital side, precisely because the video/digital can't see the signal. The noise doesn't belong to the digital side, but nor does it belong to the film side. It is noise created - at the interface between film and digital. It is noise created in the difference between two entirely different ways of encoding an image.

Carl

[Nicholas Kovats]

Thank for the excellent clarification and I very much appreciate your consistency. Onward, my friend! We have alot of gorgeous film to expose. Possibly 4,800 ft of expired 35mm Ilford HP4 b/w stock in a few weeks. My first.

[carllooper]

Thank for the excellent clarification and I very much appreciate your consistency. Onward, my friend! We have alot of gorgeous film to expose. Possibly 4,800 ft of expired 35mm Ilford HP4 b/w stock in a few weeks. My first.

No problem. The more I look at film the more in love with it I become.

What is noise? From a mathematical perspective there is no such as thing as noise. But from an emperical perspective there is, so what is it? Noise is a form of energy. Heat. It is the result of a conversion between useful information into useless information. But what causes it? Apart from a noise generator? It occurs where information hits an obstacle and scatters. The increase in noise that occurs, between film and digital means that somewhere, useful information is being converted into useless information. If there is an increase in noise it means that something is not getting through. Heat is being created. Energy is being dissipated. Useful information (as distinct from useless information) is being converted into heat (into useless information).

Basically very subtle signals, statistically encoded in the film, are banging up against the mosaic of pixels and not getting through. The signal has no choice but to be converted into noise: all of the energy is being absorbed by the sensor. In other words the information has nowhere else to go. The result is an increase in noise.

Something like that.

C

[JeremyC]

Basically it's that film encodes an image in the spatial frequency domain, where there is a statistical correlation between film and the image it encodes that goes all the way down, even if we can't see it. With video or digital the information is in the form of an amplitude modulated signal where there is a cut off frequency beyond which it can't see the signal. What happens at this frequency boundary is that the signal in the film (which we can't see) gets converted into actual noise on the digital side, precisely because the video/digital can't see the signal. The noise doesn't belong to the digital side, but nor does it belong to the film side. It is noise created - at the interface between film and digital. It is noise created in the difference between two entirely different ways of encoding an image.

Carl

Carl,

Is this due to the physical nature of film grain i.e. it is randomly distributed across an emulsion with the shape and orientation of each grain also being mostly random so that light falling across grains can be captured in an almost infinite variation where as with a sensor, CCD or CMOS, the repeating structure of photosites, which is one form of bandlimiting, misses a lot of this random interaction?

[Nicholas Kovats]

I really like you explanation of noise. Your on a cosmic elemental journey seeking fundamental film particle nirvana.

[Tscan]

I saw it tonight. When i first tuned in, i thought it had to be 16mm footage boxed in home movie style.. until i saw the title info. It looked really good. I didn't see much grain at all and the colors were pretty rich. The difference from colorful film shots to video reminds me why film is so awesome.

[super8man]

That's great the did a fabulous job on it. Good to hear that they treated it properly.

[carllooper]

Is this due to the physical nature of film grain i.e. it is randomly distributed across an emulsion with the shape and orientation of each grain also being mostly random so that light falling across grains can be captured in an almost infinite variation where as with a sensor, CCD or CMOS, the repeating structure of photosites, which is one form of bandlimiting, misses a lot of this random interaction?

Yes.

At very high spatial frequencies it is not the case that there is just noise in film. There is a signal. But the higher the frequency the weaker that signal becomes. There isn't any point beyond which it just stops. It just gets increasingly weaker. In theory it goes all the way down into the quantum mechanical domain, because in practice it would be difficult to demonstrate. Now the weaker it becomes the more unable it is to punch through the mosaic of a particular definition sensor: the signal gets converted into noise instead. This has obviously led to the erroneous belief that there is no signal beyond a certain frequency - that there is only noise: leading to the erroneous idea that film could actually be quantified in terms of pixels/mm. But this is simply not true. If 16mm can punch through the barrier of a particular definition sensor it is because it has more energy (information) than Super8 with which to do that. More signal means less noise created. But all this means is that Super8 requires less of a barrier than 16mm when doing a transfer: and by "less of a barrier" that means using sensors with more pixels. The complete opposite strategy of conventional wisdom !!

By way of analogy we're all familar with the interference patterns that happen when shooting old TV sets on film. Roll bars etc. Well the reverse also occurs - but it's a completely different kind of interference and is much more subtle. It is only with the evolution of high definition sensors that we can start to actually appreciate the interference that was otherwise happening with low definition sensors. In the past we just assumed it had something to do with the film despite the obvious fact that when looking at the film projected on a wall it was always way better than the video/digital transfer. We almost started to believe we were just imagining the film looked better ...

Note that increasing the capture definition doesn't require increasing the definition of the delivery channel. After capturing Super8 at a higher definition you then downscale it to the delivery definition. Strangely enough there will be a huge difference despite the delivery definition remaining unchanged. Its all to do with the beautiful way film encodes an image. But also the equally beautiful way that digital rationalises an image.

A conventional response to this is that if Super8 requires 3K then surely 16mm would require 6K, and 35mm would require 12K - but this is just not true. It is based on the false metric that px/mm (or bits/mm) can actually quantify the information in film. But the more information you have in a particular film the less the number of extra capture pixels you will need to satisfy a particular delivery definition. Of course, the higher the definition of the capture the better it will be, but we're speaking in comparative terms here. Super8 needs more capture pixels than 16mm (or 35 etc) precisely because it's signal is weaker. The important term here is "weaker" rather than non-existant. Non-existence happens in digital - not in film.

None of this is to suggest that Super8 has more information than 16mm. It is to say that Super8 has more information (or requires more pixels) than digital transfer setups have, for decades, insisted on assuming. It's nothing to do with the definition of film vs that of digital. It's do with the very different way each encodes an image. It is an assymetrical relationship and can require some counter-intuitive theorisation.

Or put it another way you can't use digital theory to measure the difference between film gauges. You will get it wrong. You have to use analog theory. And when it comes to film-to-digital systems you have to use both!

C

[Tommy]

I thought the scans were impressive, from the sections I saw not a single scratch was visible. The only thing I thought was dumb, was presenting super 8 in a boxed in shrunken format....the quality was good enough for full screen, ...why not use it to full advantage?
Its like going to a screening with a 72 inch screen and projectionist zooms the picture down to 60 inches wide. I know the producers thought they would create more of a "home movie " look and matte it in...... would you have done the same?

[Maurizio Di Cintio]

Carl, I really like your explanations.

[Will2]

I've seen a few Super 8 negative scans, especially from the new 50D that really seemed closer to 16mm than Super 8; but rarely does Super 8 reversal look this good. I'm sure it has much to do with the colorist but with reversal film the scan itself seems like it would be very tricky and this scanner handled it great.

Unfortunately that means I have a ton of footage that will need to be rescanned. But at least I can do it; glad I didn't shoot this stuff on VHS or 8mm video.