Super35

[Guest Daniel J. Ashley-Smith]

I take it with Super35 you shoot the 4:3 ratio and then take the 16:9 out of it for widescreens?

 

The thing I'm a bit confused on, is that a 35mm neg is closer to a 4:3 ratio picture than a 16:9. So, why take the 16:9 picture, and take the 4:3 out of that? You could get a much larger 4:3 picture if you took it straight from the neg, not the 16:9 picture.

 

EDIT_

 

Oh and.. Ok you can only stretch 35mm so far before it starts getting grainy, and it varies with different film. So it's quite hard working it out perfectly digitalised. So, what would be a standard digital image spec if you were to transfer 35mm to digital? Must be quite a bit, at LEAST 10 mega pixels large.

 

EDIT2_

 

Oh and, I don't suppose you can interpolate images AFTER they?ve been shot can you? It's just that when stretching an image the pixels grow, interpolation does it a different way (I think). They could do that in cinemas.

 

And is there a good way to remove grain on film, digitally?

Edited September 28, 2004 by ashleysmithd

[Mike Panczenko]

With Super 35, you shoot an image of exactly 1.33:1, or 4:3 what often happens is that you take the 2.40 image out for the theatrical release, and, for the 4:3 transfer, you reframe a little bit using the parts of the image outside the 2.4:1 image, but you don't use the whole original 4:3 image, because when you shoot Super 35, you don't usually frame for the whole 4:3 image, as it would compromise the framing for the 2.40 image. You can usually only compose well for one aspect ratio. People are starting to compromise, though, to compose with several aspect ratios in mind to make both the theatrical and video release have good composition, though. This page may explain it better: http://home1.gte.net/res0mrb7/widescreen/film.html

[David Mullen ASC]

Super-35 is the same thing as Full Aperture photography, meaning that you expose the image all the way across the negative from sprocket row to sprocket row on each side, rather than leave a small area for a soundtrack.

 

4-perf 35mm Full Aperture, used in the Silent Era, is 1.33 : 1 (4x3). 3-perf 35mm Full Aperture is close to 1.78 : 1 (16x9).

 

The reason why TV shows compose 16x9 and 4x3 with the same top & bottom frameline, as opposed to shoot in 4-perf 35mm and expose 4x3 as large as possible and crop top & bottom to achieve 16x9 (similar to how we crop 1.37 Academy top & bottom to achieve 1.85 : 1 for theatrical films) is:

 

The transfer and post work are done using a 16x9 transfer (often HD these days) and THEN a 4x3 version is center-extracted from that.

 

If you exposed 4x3 to be taller than 16x9, you'd either have to make separate transfers and do separate posts, or you'd have to do the post in 4x3 and crop that later to create the 16x9 version, which looks worse than cropping the 16x9 version to create 4x3 -- especially if the post work was done in HD, which is naturally 16x9 and thus there is no loss in cropping to achieve 4x3 for a standard def downconversion. There IS some loss of you do a 16x9 standard def post and then crop to create 4x3, since 16x9 and 4x3 use the same number of pixels in standard def.

 

In the case of shooting 1.37 Academy and framing for theatrical cropping to 1.85 during projection, you DO make separate transfers for 4x3 full-frame TV versions versus widescreen versions since you might as well take advantage of the taller area not used for the theatrical version. But should digital intermediates become the norm, I can see post work being done for 1.85 features where only a 16x9 area of the film is scanned and used, and 4x3 versions are extracted from that.

 

Film can be scanned at whatever pixel resolution you want. Ideally it would be high enough to not compromise the image visibly. Right now, the industry standard has been to work at 2K, although some companies now scan at 4K and then downrez to 2K. But ideally they would work at 4K throughout (4K in a 4x3 aspect ratio is 12 MP).

[John Sprung]

If you exposed 4x3 to be taller than 16x9, you'd either have to make separate transfers and do separate posts, or you'd have to do the post in 4x3 and crop that later to create the 16x9 version, which looks worse than cropping the 16x9 version to create 4x3

 

It's really personal preference as to which way of dealing with the two aspect ratios looks worse. My own opinion is that if you tell a story in 4x3, slapping on whatever happened to be alongside the intended frame to make 16x9 results in a very loose, empty feeling show. Cropping the top and bottom off of 4x3 to make 16x9 usually looks better, provided that you can "tilt" -- adjust the headroom -- on a shot by shot basis. The only time that gets troublesome is if you want tight closeups in 4x3, they get way too tight in 16x9. Other than that, shooting 4x3 so it can be cropped to 16x9 is easier than keeping those sides empty.

 

We did it that way for three years, using an "anamorphic" telecine so as not to waste any of the pixels on the 1080p/24 tape. The tape to tape tilt and scan session cost an extra $1k per hour show. Associate producers hated having to do the extra work and keep track of two versions. Producers, directors, executives, and even DP's had a hell of a hard time even grasping the concept of common sides and tilt and scan. With the probability that the networks will soon want single tape delivery, we dropped the idea. We'll still let a show do it that way if they want to.

 

Bill Bennett who does the car commercials told me that this is sort of similar to the way he works. His operators use a special ground glass that makes them compose for a smaller 4x3 with lots of extra image around it. Then he makes the final compositions in telecine, much the way a still photographer shoots loose and makes the final composition in the darkroom. Interesting things become possible when you're spending millions of bucks on something no more than maybe a dozen shots long.

 

 

 

-- J.S.

[David Mullen ASC]

I only meant "worse" technically, not artistically, in that people who watch 16x9 broadcasts tend to have a better monitor than people who watch 4x3 broadcasts, so it makes sense to lose pixel resolution (if working entirely in standard def for post) by getting 4x3 from 16x9 rather than 16x9 from 4x3.

 

Compositionally, I tend to believe in a "one aspect ratio for all viewing systems" -- i.e. shoot 16x9 and letterbox for 4x3 rather than panning & scanning. But practically, the best compromise would be something like a cross-shape where 4x3 is slightly taller and slightly less wide than 16x9 (sort of like the TV/1.85 framelines in cameras shooting for print projection.)

 

But like I said, most producers only want to do one post and color-correction and then extract other versions from that, hence why in 16x9 HD posts, 4x3 is center-extracted. I suppose if we could move away from video systems and work in pure data for all post work, then you can extract both 16x9 and 4x3 from one master that could, in theory, be oversized like Bill Bennet's "Big TV" format. However, I'm more for pushing for a single aspect ratio, the one orginally chosen and composed for, for each presentation (but not overall a single aspect ratio for all production, unlike Storaro.)

[Phil Rhodes]

Hi,

 

I find it's also better to rescale horizontally for TV output. CRT displays at least produce a continuous display in the horizontal direction - vertically it's just a series of stripes, which I find makes any aliasing and lack of resolution much more visible. Stretching horizontally - to do a centre extraction from 16:9 - seems to be a much lesser evil than trying to scale the 4:3 up to make 16:9.

 

Having said that I don't have a problem with scaling 16:9 down to fit inside 4:3 - it's such a large resample that there really doesn't seem to be any aliasing.

 

Phil

[Guest Daniel J. Ashley-Smith]

reframe a little bit using the parts of the image outside the 2.4:1 image, but you don't use the whole original 4:3 image, because when you shoot Super 35, you don't usually frame for the whole 4:3 image, as it would compromise the framing for the 2.40 image.

 

It may compromise the 2.40 image slightly, unless your shooting with 2.40 guidelines, but, in doing that it's a huge compromise the the 4:3 image.

 

If you shoot the film in 4:3, you can still get the full negative plus a 2.40 image. When taking the 4:3 image out of the 2.40 film you lose a lot of the negative.

 

I personally would want to shoot the film with a 2.40 guideline on the viewfinder (if they exist), and then just use the full frame image for the 4:3. Therefore making the most out of the negative, gaining the maximum amount of "picture" that I can get.

 

Oh and, you shoot exactly 2.40 with Super35, what's the ratio with normal 35mm film?

[Keith Mottram]

"normal" is 1.33:1 or 4:3, there are markers for 2.40:1, on the groundglass. What you want and what you can do are incompatable, as a test make yourself a piece of card with both aspect ratios cut out of them, then draw a picture of a mid shot of a person (head to waist) now put your card over the top. You'll quickly see it is not possible to frame for both, try this test over pictures of landcapes etc. This is why you have to pan and scan - if for example you look at a film in 2.40:1 with two people on either side of frame then work out where your 4:3 image would be you'll quickly see the problem.

 

Keith

[John Sprung]

Compositionally, I tend to believe in a "one aspect ratio for all viewing systems" -- i.e. shoot 16x9 and letterbox for 4x3 rather than panning & scanning.  But practically, the best compromise would be something like a cross-shape where 4x3 is slightly taller and slightly less wide than 16x9

<{POST_SNAPBACK}>

 

Yes, absolutely, one aspect ratio per show is the best idea. We have a few that are letterboxing for NTSC, and it's really a pleasure to go that way. You get better compositions when the DP has only one goal in mind. You have less stuff to do in post if you only make and title one version. And it's a lot easier for the rest of the life of the show to have only one version on the shelf.

 

Interestingly, we tested the "cross-shape" idea back in 1996-97. I drew up over a dozen different possible relationships between 4x3 and 16x9 in AutoCad, all with equal amounts of "extra" 16x9 room on the sides, but more 4x3 slop below than above. I showed them to some DP's, and Bob Primes and I selected two of them for testing. Panavision made up some ground glasses, and Jonathan West shot some tests for us on the DS9 set. Those tests blew the idea out of the water. That's what convinced me that the only right solution was to give up on a fixed relationship between the two frames, and argue for the extra work and expense of an online repositioning session to get the best possible compromise for the secondary composition. I gave the test ground glasses to Marianne in the camera department, and she made earrings out of them.

 

Later we did tests of pan and scan vs. tilt and scan on Voyager, and chose the tilt and scan approach.

 

-- J.S.

[David Mullen ASC]

You can shoot Super-35 and "protect" all of 4x3 for TV but compose for 2.40 -- but even if you managed to keep all junk out of 4x3 (near impossible), you'll find some zooming in and reframing unavoidable for artistic reasons. 2.40 and 1.33 is such a big difference that if you simply showed the whole Super-35 negative frame, your film would seem to have no close-ups if it were composed to be cropped to 2.40.

 

You look at Cameron's Super-35 films in the 4x3 TV pan & scan version and he uses more of the width of the original Super-35 frame when something is important at both ends of the frame, but zooms in more when it's just a close-up and he's not losing too much info in the sides of the frame. So shot-by-shot, the image is recomposed for 4x3 TV, which is absolutely necessary if you are working with something composed for 2.40.

[John Sprung]

.... you'll find some zooming in and reframing unavoidable for artistic reasons.  2.40 and 1.33 is such a big difference ....

<{POST_SNAPBACK}>

 

Come to think of it, it goes even farther than that. I'd expect that a reasonable director would even choose to block a scene differently for those two aspect ratios. For instance, the three shot walking towards camera on a long lens that is done so often in scope -- you'd probably want to arrange the actors differently for 1.33. Instead of just re-composing shots, you'd have all different coverage.

 

 

 

-- J.S.

[Mark Sasahara]

Come to think of it, it goes even farther than that.  I'd expect that a reasonable director would even choose to block a scene differently for those two aspect ratios.  For instance, the three shot walking towards camera on a long lens that is done so often in scope -- you'd probably want to arrange the actors differently for 1.33.  Instead of just re-composing shots, you'd have all different coverage.

-- J.S.

<{POST_SNAPBACK}>

 

 

If the Director agrees, would the Producers and 1st AD let us? It's a difficult question that inevitably runs into: "how much?" and "how long?". I'd rather get it right, in the camera, but 2.4 would require almost every shot to be recomposed for 4:3. I've been able letterbox sometimes. Most times it's a center extraction, but I know before hand so I "respect the pumpkin". I've argued about formats a time or two.

[John Sprung]

If the Director agrees, would the Producers and 1st AD let us?

<{POST_SNAPBACK}>

 

In episodic, they're balking at an extra $1k for a repo pass. To really do it right, we'd almost be shooting the show twice. If I have a strong diagonal shadow in the BG, I'd probably want to move a light for the other version. The economics would be out of the question on anything but a big time TV commercial.

 

 

 

-- J.S.

[Timo Haapasaari]

So what´s the difference and/or history between 2.35 and 2.40(2.39)? Some sources talk about 2.33:1. I´ve noticed that in the AC magazine, when they talk about anamorphic it´s 2.40, otherwise 2.35. In the American Cinematographer Manual though it says that the 2.35 hasn´t existed for a long time!

 

I´m a bit confused here..

[David Mullen ASC]

Anamorphic lenses have a 2X squeeze, so on a Full Aperture 4-perf 35mm neg and print, it would be 2.66 : 1 when unsqueezed (since Full Aperture is 1.33 : 1). This was the original intent with CinemaScope, to use Full Aperture (since its rival, Cinerama, was 2.66 : 1). The sound would be run in interlock using 35mm fullcoat mg, as with Cinerama.

 

But before the first CinemaScope film was finished, it was decided to put the soundtrack on the print. Kodak made a special print stock with smaller perfs ("CS" or "CinemaScope" perfs) and mag stripes were added to the print on both sides of the frame, shortening the width to 2.55 : 1. This was how early CinemaScope films were shot and projected. This is also why when you see a new print of "20 Thousand Leagues under the Sea", the titles of off-centered, because it was meant to be trimmed on both sides equally by mag stripes, not just on the left by an optical soundtrack.

 

Then when mag-striped CS perf prints were abandoned and normal prints with an optical track were made, the image was shifted to one side of the print (as will all sound formats like Academy or 1.85) and the width was further shortened to 2.35 : 1. Picture extended the whole height of the negative and print, with very thin frame lines between each picture.

 

Due to complaints about flashes appearing onscreen at every splice, SMPTE / ANSI in the early 1970's decided that the projector gate should be shortened slightly for anamorphic projection, to produce a thicker frameline and hide bad splicing better. This reduced the height of the image, making it a wider aspect ratio, 2.39 : 1.

 

In the early 1980's, the changed the size of the projector aperture matte again, making it slightly smaller overall, but the proportions remained 2.39 : 1 (actually 2.3913042 : 1).

 

So it hasn't been 2.35 since the early 1970's but old habits die hard. Some people round 2.39 up to 2.40. Some people still say 2.35. Some projectors still have to older 2.35 anamorphic masks cut.

 

Oddly enough, "Northfork" had a special screening at Kodak in Hollywood, and I noticed some frameline flashing, so I talked to the projectionist, who said "But I just had some new 2.35 anamorphic mattes cut for this projector gate." "But, it hasn't been 2.35 since 1971" I said. "Are you sure?" he said, "I always thought it was 2.35."

[Dominic Case]

Here's anamusing site that effectively demonstrates why there's NO solution to multiple aspect ratios.

 

http://www.widescreenmuseum.com/flikfx/loatest1.htm

[John Pytlak RIP]

Oddly enough, "Northfork" had a special screening at Kodak in Hollywood, and I noticed some frameline flashing, so I talked to the projectionist, who said "But I just had some new 2.35 anamorphic mattes cut for this projector gate."  "But, it hasn't been 2.35 since 1971" I said. "Are you sure?" he said, "I always thought it was 2.35."

<{POST_SNAPBACK}>

 

Sounds like I need to do a bit of training the next time I visit our office in Hollywood. ;)

 

Standard SMPTE 195 specifies the current projectable image areas for 35mm print formats. The image area for "scope" is 0.690 x 0.825 inches, or 2.39:1 with a 2X anamorphic lens. Here's the story:

 

http://www.widescreenmuseum.com/widescreen/apertures.htm

 

From 1957 through today, there have been a number of small changes to the anamorphic projector aperture recommended standards. John P. Pytlak of Eastman Kodak dug through countless back issues of the SMPTE Journal in order to provide the following chronology of the changes to the standards:

 

 

 

The March 1957 SMPTE Journal has PH22.104-1957, the standard for 2.55:1 anamorphic (no optical track), with an aperture size of 0.912 X 0.715 inches. Notice of withdrawal of this standard was in the January 1964 Journal.

 

The December 1957 SMPTE Journal has PH22.106-1957 for 2.35:1 anamorphic, with an aperture of 0.839 X 0.715 inches. It was unchanged in the September 1964 Journal. The November 1965 SMPTE Journal published PH22.106-1965 still with the 0.839 X 0.715 aperture size.

 

In the September 1970 SMPTE Journal, a new draft of PH22.106 was proposed, with an aperture size of 0.838 X 0.700 inches, to minimize the flashes at splices. This was republished as standard PH22.106-1971 in the October 1971 issue.

 

In the June 1976 SMPTE Journal, the two (flat and scope) projectable image area standards (PH22.58 and PH22.106) were consolidated into one standard and renamed PH22.195. The publication of PH22.195-1984 in the October 1984 Journal still had the scope area as 0.700 X 0.838 inches.

 

The June 1992 SMPTE Journal published a proposed revision, with a scope area of 0.690 X 0.825 inches. In August 1993, the standard was published as SMPTE 195-1993, with the current area of 0.690 X 0.825 inches. So August 1993 is when the two formats became the same width of 0.825 inches.