Jump to content


Photo

2.35 anamorphic color timing problems


  • Please log in to reply
11 replies to this topic

#1 John Harcourt

John Harcourt

    New

  • Basic Members
  • Pip
  • 1 posts
  • Other

Posted 19 June 2005 - 03:01 PM

I am currently in the final stages of post production on a feature that was shot 2.35 anamorphic. The primary film work has been successfully completed and we are at the stage of effecting the various transfers from IP to HD (2.35/1.78 and 1.33). My question has to do with a color timing anomaly I'm experiencing in this process. As I'm sure everyone has experienced, in answer printing there will occasionally be a partial flash of varying color or density in the first frame of a new shot. It's an artefact caused by the timing valves not quite fully adjusting from the previous shots' values. In my past experience working with flat negatives (mostly 1.85) this anomaly is largely masked by the relatively generous frame line. In other words, the color and density error takes place in non-image parts of the frame. In this 2.35 anamorphic picture there is virtually no frame line beyond the splice itself so, although you may get the benefit of maximizing film real estate, all the evils of the splice and the aforementioned color timing flashes are painfully apparent.

In a projection environment at real time these problems are hard to see. You really have to know what you're looking for. However, on video, and in particular during the non-real time critical analysis that video masters are now subject to by international distributors, these problem jump out by the dozen. I?m wondering if anyone else has had to deal with these peculiar anamorphic issues of quality control. Are there particular types of registered optical printers that would eliminate these color flashes? Am I simply applying 21st century video standards to a 20th century film technology? Any constructive thoughts are welcomed.
  • 0

#2 Max Jacoby

Max Jacoby
  • Sustaining Members
  • 2955 posts
  • Other

Posted 19 June 2005 - 03:39 PM

This sounds more like an issue with your lab than with anamorphic itself. I've never experienced this problem while timing anamorphic.
  • 0

#3 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 19759 posts
  • Cinematographer
  • Los Angeles

Posted 19 June 2005 - 08:12 PM

This sounds more like an issue with your lab than with anamorphic itself. I've never experienced this problem while timing anamorphic.

<{POST_SNAPBACK}>


We had a major problem with the IP for a couple of reels of "Northfork" with this problem -- the timing light changes were a COUPLE of frames late, not just one frame late. Trouble is that we had our transfer sessions booked, so we color-corrected the movie with the screwed-up IP and then had to wait for the lab to fix the printer they used so they could make a new IP and hope that the color hadn't drifted.

We couldn't move the printing of the IP to another printer altogether because then the color WOULD be different and we couldn't use our programmed corrections. So a week later, they fixed the printer, we finally got a corrected IP from the lab, ran it in the telecine with our programmed color-corrections, and it was fine, it all tracked.

In other words, get your lab to do it right.
  • 0

#4 Dominic Case

Dominic Case
  • Basic Members
  • PipPipPipPip
  • 1357 posts
  • Other
  • Sydney Australia

Posted 19 June 2005 - 09:50 PM

In other words, get your lab to do it right.

More easily said than done, David.

The problem that you describe (2 frames late) is definitely a lab error: a matter of adjusting the delay between the light change cue being triggered by the frame counter, and the actual change. The solution you found, although inelegant, was sensible, and obviously the lab had little difficulty in fixing that problem.

However, the slight "drag" described by Harcourt is a different matter. Unfortunately, it takes a discrete amount of time for the light valves to move from thesetting for one shot to the next, and on a continuous printer that means a discrete length of film. It's also limited by the width of the printing aperture.

With flat negs, this isn't an issue. Nor is it an issue on 'scope pictures (with the narrow frame line) when the light changes from shot to shot are relatively small. (In other words, if exposure has been consistent). It is when you start making 10 (or more point) light changes that the problem sometimes becomes visible. This can also arise, of course, if you are intercutting Fuji and Eastman neg, with their quite different printer lights, or if you are cutting dupe neg stock (eg digital or optical shots) into camera neg.

This is nothing new, but as Harcourt says, it was never a problem when projecting the film in real time (which is how audiences see the film). It's only when you inch through frame by frame (and sometimes even compare one frame with the next using a frame store) that you can pick the problem - if that's what it is.

A step printer would make the problem go away immediately. Labs usually have a higher charge for step printing (usually it's an optical step printer) as it's a time-intensive process and often a labour-intensive one as well. And there is no guarantee that another, potentially worse, problem - jumping splices - might pop in to replace the one you've just eliminated. This could arise if the splices are not absolutely perfectly aligned ( a greater risk with the very narrow splices required for anamorphic frames). Optical printers are not really designed to accomodate cement splices in the gate, whereas continuous printerts are much more tolerant of them.

So in a way you are indeed applying 21st century requirements to a 20th century technology - but I always question the reality of a problem where you need to step-frame through and use framestores for comparison to detect any change, and where there is no noticeable difficulty on normal projection.
  • 0

#5 John Pytlak RIP

John Pytlak RIP

    (deceased)

  • Sustaining Members
  • 3499 posts
  • Industry Rep
  • Rochester, NY 14650-1922

Posted 20 June 2005 - 02:01 PM

To some extent, the problem may be lessened if the lab uses a slower printing speed to make the answer print or master positive. If the printer light valves take a certain amount of time to make a change, the distance on the film during the light change is lessened if you print at 120 feet per minute rather than 240 feet per minute. But a lab might not be able to print slower, or charge a premium due to the added time to print the job.

Electronic light valves are used on some additive printers, and are said to have a somewhat faster response time than electromechancial valves.

http://wwwdb.oscars....y_in=Laboratory

1997
(70th) Academy Award of Merit Laboratory GUNNAR MICHELSON  for the engineering and development of an improved, electronic, high-speed, precision light valve for use in motion picture printing machines 



United States Patent  4,594,539 
Michelson  June 10, 1986 

--------------------------------------------------------------------------------
Light valve


Abstract
A light valve has a pair of rotating vanes for controlling the cross section of a light beam passing between the vanes. The vanes rotate on corresponding shafts, each controlled by its own integral motor. Separate feedback control systems produce electrical control signals to each motor for rotating each shaft independently of the other to control the desired phase angles of the vanes. In one embodiment, both motors are permanent magnet direct current servo motors in which the wound armature is an integral part of each shaft. A stationary magnetic field generated by permanent magnets is common to both motors. Power is supplied to each armature by a pair of thin, parallel, flexible, electrically conductive strips extending from an end of each shaft. The conductive strips reduce the torque on the shaft when twisted during operation. In one embodiment, the control system produces a calibrated digital control signal from a PROM representing the desired phase angles of the rotating blades. The digital signal is converted to analog control signals compared with position feedback signals from the rotating blades and damping signals produced by blade velocity feedback signals. Error signals fed to variable gain amplifiers control the positions of the vanes. Whenever new commands are input to the variable gain amplifiers, amplifier gain is temporarily switched to a high gain mode, and after the phase angle of the shaft has stabilized, the amplifier gain is switched back to a low gain mode. The separate motors and their servo systems speed up response time and substantially reduce any tendency toward oscillations.


--------------------------------------------------------------------------------
Inventors:  Michelson; Gunnar P. (505 Sea Ranch Dr., Santa Barbara, CA 93109) 
Appl. No.:  529945
Filed:  September 6, 1983


  • 0

#6 dan_taylor

dan_taylor

    New

  • Basic Members
  • Pip
  • 2 posts
  • Producer

Posted 21 June 2005 - 01:49 AM

I am currently in the final stages of post production on a feature that was shot 2.35 anamorphic.  The primary film work has been successfully completed and we are at the stage of effecting the various transfers from IP to HD (2.35/1.78 and 1.33).  My question has to do with a color timing anomaly I'm experiencing in this process.  As I'm sure everyone has experienced, in answer printing there will occasionally be a partial flash of varying color or density in the first frame of a new shot.  It's an artefact caused by the timing valves not quite fully adjusting from the previous shots' values.  In my past experience working with flat negatives (mostly 1.85) this anomaly is largely masked by the relatively generous frame line.  In other words, the color and density error takes place in non-image parts of the frame.  In this 2.35 anamorphic picture there is virtually no frame line beyond the splice itself so, although you may get the benefit of maximizing film real estate, all the evils of the splice and the aforementioned color timing flashes are painfully apparent.

In a projection environment at real time these problems are hard to see.  You really have to know what you're looking for.  However, on video, and in particular during the non-real time critical analysis that video masters are now subject to by international distributors, these problem jump out by the dozen.  I?m wondering if anyone else has had to deal with these peculiar anamorphic issues of quality control.  Are there particular types of registered optical printers that would eliminate these color flashes?  Am I simply applying 21st century video standards to a 20th century film technology?  Any constructive thoughts are welcomed.

<{POST_SNAPBACK}>


Man, do I ever hear you on this. Perhaps you know more about this stuff than I do, but I remain convinced that the flashes are in fact the result of neg splicing cement, which is not being as successfully masked as it would in 1.85.

I have no idea how to remedy the problem. We were on a limited budget, so during our transfer we just ended up adjusting the mask to conceal as much as possible without ruining the composition. But there are still remnants, and it's really, REALLY quite disheartening.

My colorist said that producers on larger-budget projects will sometimes do the transfer, then literally go edit out the final frame of shot A (replacing it with the second to last frame) and first frame of shot B (replacing it with the second frame).

If it's of any consolation: I recently saw Wong Kar Wai's new movie "2046" (shot in anamorphic by Christopher Doyle) and nearly 1/3 of the film seemed to suffer this problem. And this was a $12 million dollar movie. Sigh.
  • 0

#7 Max Jacoby

Max Jacoby
  • Sustaining Members
  • 2955 posts
  • Other

Posted 21 June 2005 - 06:27 AM

If it's of any consolation: I  recently saw Wong Kar Wai's new movie "2046" (shot in anamorphic by Christopher Doyle) and nearly 1/3 of the film seemed to suffer this problem.


That's interesting, since that film went through a DI.
  • 0

#8 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 19759 posts
  • Cinematographer
  • Los Angeles

Posted 21 June 2005 - 10:47 AM

Frameline flashes are a different issue that printer light changes that are a frame late. With the printer light problem, in the video transfer, you could just color-correct that frame to match the following scene -- tedious but it beats cutting out the frame.

Cropping the shot makes no sense because then you'd be cropping the whole shot even where there was no flash or color problem.

Frameline flashes occur when the neg cutter did not use a proper CinemaScope splicer.

Honestly, you're overstating this problem of shooting in anamorphic. The solution is not to just give up and use a wasteful format like 1.85 to hide a lab or neg splicer's mistakes. The solution is to get the lab or neg splicer to do a better job.
  • 0

#9 John Pytlak RIP

John Pytlak RIP

    (deceased)

  • Sustaining Members
  • 3499 posts
  • Industry Rep
  • Rochester, NY 14650-1922

Posted 21 June 2005 - 01:01 PM

Frameline flashes are a different issue that printer light changes that are a frame late. With the printer light problem, in the video transfer, you could just color-correct that frame to match the following scene -- tedious but it beats cutting out the frame.

Cropping the shot makes no sense because then you'd be cropping the whole shot even where there was no flash or color problem.

Frameline flashes occur when the neg cutter did not use a proper CinemaScope splicer.

Honestly, you're overstating this problem of shooting in anamorphic. The solution is not to just give up and use a wasteful format like 1.85 to hide a lab or neg splicer's mistakes. The solution is to get the lab or neg splicer to do a better job.

<{POST_SNAPBACK}>



SMPTE Recommended Practice RP111-1999 specifies a narrower overlap for splices used on 35mm anamorphic films:


1 Scope
1.1 This practice specifies the significant dimensions
of splices for 70- , 65-, and 35-mm motionpic
t ure f ilm i n tended for project ion and
exhibition or for laboratory printing.


1) Notice that the splices on anamorphic films will fall within the projected area, and extra care must be
taken in making a clean splice.
2) To minimize projection of splices, the width of the laboratory splice should be no greater than 0.040 in
(1.02 mm).


Just as with 16mm splices, A/B roll "checkerboard" cutting can also be used for 35mm "hide" the splice overlap:

http://www.acvl.org/handbook/3c.htm
  • 0

#10 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11936 posts
  • Other

Posted 21 June 2005 - 05:26 PM

Hi,

> The solution is to get the lab or neg splicer to do a better job.

Quite. I'm aghast at this problem, of which I had previously had no knowledge, although I had wondered how they managed to move the valves so fast. But hang on - the equipment causes what are effectively single mistimed frames to occur because it isn't fast enough, then they want to charge you more to do it slower? That's a complete con; making excuses about how it's twentieth century technology and we're asking too much of it is just playing into the hands of the anti-film crowd and a piss-poor excuse for faults on the equipment.

LED light sources with variable intensity would be faster.

Phil
  • 0

#11 Dominic Case

Dominic Case
  • Basic Members
  • PipPipPipPip
  • 1357 posts
  • Other
  • Sydney Australia

Posted 22 June 2005 - 12:09 AM

OK you guys. Read this carefully. This problem is NOT about the speed of the light valves. It is about (1)the width of the printing aperture, and (2) the magnitude of the light changes. And by the way, I've had this conversation with Mike Michelson, who should know about high speed light valves!

Oh, and it has absolutely nothing to do with splices. Seeing the film overlap or cement spread into the frame is simply a matter of using the wrong splicer or bad cement or technique. But it's not going to affect the position of a light change.

(1) - During printing, the neg and the rawstock move continuously past a printing aperture or slit that is about 4mm wide. The colour of light coming through that aperture is set by the R G & B light valves. When they change, the colour of the light changes. It doesn't matter how slowly you run the printer, or how quickly the light valves move, the beam of light hitting the film is still 4mm wide, and so the colour cannot change along the length of the film in less than 4mm. If the vanes are very slow, or the printer runs very fast, it can be a lot more than 4mm - but that's the bottom limit.

(2) - If the light changes are only a few points - the normal extent of shot-to-shot colour balancing - then the "problem" isn't ever noticeable. If they are big changes, you can pick them more easily in frame-to-frame analysis on telecine, but they have to be very dramatic changes to see them on film projection.

Phil complains about charging more for using slower printers. In fact, many years ago, it was the norm for labs to make IPs on step printers - it was a world where there was a bit more time to do the job the way you thought best. However, there was (as always) price pressure from producers, and labs found they could do the job on continuous printers a little more quickly and a little more cheaply. With good and consistent exposures in the original negative, there is never any problem. Now the continuous printer has become the accepted norm for making IPs - at the lower cost.

So the labs aren't "charging more to do it slower". Very often they won't charge more to do the step print if there's a problem, they will have to absorb the cost themselves. And if they do charge for step printing, then what they are doing is charging the correct price for doing the job the correct way, instead of cutting corners as demanded by most producers.

And while LEDs would indeed respond a lot faster than a mechanical light valve, this would not address the problem (see (1) above), and there seems to be little appetite for building this type of light head. There are only a couple of printer manufacturers in the world, and they resolutely build light sources and lightheads the same way.

Except in extreme cases (often compounded by the slightly different timing error described by David Mullen), this IS NOT a problem except in non-real time critical analysis. In other words you DON'T SEE IT. The fault is in the level of inspection being provided by transfer and mastering houses which far exceeds the reasonable requirements of even the most critical audience. The original poster (Harcourt) said as much in the first email in this thread.
  • 0

#12 John Pytlak RIP

John Pytlak RIP

    (deceased)

  • Sustaining Members
  • 3499 posts
  • Industry Rep
  • Rochester, NY 14650-1922

Posted 22 June 2005 - 05:22 AM

Dominic: Thank you for your "real world" perspective.
  • 0


CineTape

Opal

CineLab

Glidecam

Abel Cine

The Slider

Wooden Camera

FJS International, LLC

Aerial Filmworks

Ritter Battery

Rig Wheels Passport

Technodolly

Visual Products

rebotnix Technologies

Metropolis Post

Tai Audio

Paralinx LLC

Willys Widgets

Broadcast Solutions Inc

Media Blackout - Custom Cables and AKS

Gamma Ray Digital Inc

Broadcast Solutions Inc

Rig Wheels Passport

Gamma Ray Digital Inc

Opal

Willys Widgets

Technodolly

FJS International, LLC

Wooden Camera

Abel Cine

Metropolis Post

Glidecam

Ritter Battery

Aerial Filmworks

CineLab

rebotnix Technologies

The Slider

CineTape

Paralinx LLC

Tai Audio

Visual Products

Media Blackout - Custom Cables and AKS