Jump to content


Photo

What is the actual resolution of 35mm negative?


  • Please log in to reply
19 replies to this topic

#1 Werner Klipsch

Werner Klipsch
  • Basic Members
  • PipPip
  • 86 posts
  • Other

Posted 14 December 2006 - 05:18 AM

I never seem to be able to get an absolute answer to this question.
I realize there possibly isn't one, but what I read is very confusing.

I have heard the figure of '4000 lines' used but other deny this.
I presume what this is saying is that if the camera was precision framed up to a chart of 2000 north-south black lines on a white background, that is, so that 2000 black lines are focussed across the negative area, the lines should be visible on the processed negative. How visible will of course depend on the efficiency of the focus puller and the quality of the lens used. :D

But given 100 ASA film approximation, the lines should be visible at least on the negative.

Is this correct, and if it is, would you expect to also see the lines on a positive print, made by wet-gate perhaps?

My other question is, I have heard that HD cameras like the Cine Alta can only resolve about 500 black lines on white background. Is this right? It sounds unlikely to me.
  • 0

#2 Stephen Williams

Stephen Williams
  • Sustaining Members
  • 4708 posts
  • Cinematographer
  • Europe

Posted 14 December 2006 - 06:48 AM

I never seem to be able to get an absolute answer to this question.
I realize there possibly isn't one, but what I read is very confusing.

I have heard the figure of '4000 lines' used but other deny this.
I presume what this is saying is that if the camera was precision framed up to a chart of 2000 north-south black lines on a white background, that is, so that 2000 black lines are focussed across the negative area, the lines should be visible on the processed negative. How visible will of course depend on the efficiency of the focus puller and the quality of the lens used. :D

But given 100 ASA film approximation, the lines should be visible at least on the negative.

Is this correct, and if it is, would you expect to also see the lines on a positive print, made by wet-gate perhaps?

My other question is, I have heard that HD cameras like the Cine Alta can only resolve about 500 black lines on white background. Is this right? It sounds unlikely to me.


Hi,

People scan 35mm at 6K and more sometimes, I guess they are not stupid.

What ends up on the screen after duplication will be less. I understand in the 3-4K range. Anamorphic may be more.

Stephen
  • 0

#3 Patrick Cooper

Patrick Cooper
  • Basic Members
  • PipPipPipPip
  • 868 posts
  • Other

Posted 14 December 2006 - 08:08 AM

"But given 100 ASA film approximation, the lines should be visible at least on the negative."

Why limit yourself to a 100asa speed film? You could find out the resolution of Kodak Vision 50D, Fuji 64D, Kodak Vision 100T or even a 500asa speed film all within the 35mm format.
  • 0

#4 David Venhaus

David Venhaus
  • Basic Members
  • PipPipPip
  • 129 posts
  • Other
  • Wiseburn, CA

Posted 14 December 2006 - 09:41 AM

A direct comparison to resolution, in the digital sense, isn't really valid. Grain in film is not square and not all a uniform size and is also not 2 dimensional so it can't be measured in a grid fashion like pixels. In fine grained B+W films, typically 4 or 5 silver halide atoms form a grain (about .01 microns or 10nm, but is varible) and the shapes of the crystals can be spheriod, cubic, needle like, and many more.
  • 0

#5 David Mullen ASC

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

Posted 14 December 2006 - 12:14 PM

Film does not have pixels so it doesn't have a pixel resolution. Film resolution, like with lenses, are usually measured in MTF.

The real question is what pixel resolution does 35mm color negative need to be scanned at to preserve every particle of grain without loss of high fequency detail/edges, understanding that the grains ARE the image with film.

Ideally, you'd just overscan at 16K or something ridiculous just to be sure, but most post people will say that 35mm color neg needs to be scanned at 4K to 6K (approx. 4000 to 6000 horizontal pixels -- the vertical depends on the aspect ratio) in order to not lose information in the grains. Below 4K, there starts to be a "smoothing" effect that comes from rounding off the sharper edges of the grains.

This doesn't take into account any issues regarding the sharpness of the photography.
  • 0

#6 Dan Goulder

Dan Goulder
  • Sustaining Members
  • 1259 posts
  • Cinematographer

Posted 14 December 2006 - 12:53 PM

As film is a chemical medium, its "true" resolution would have to be measured at the atomic level. A digital representation of this resolution in terms of pixels would be limited to the resolving power of the measurement device itself.
  • 0

#7 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11936 posts
  • Other

Posted 14 December 2006 - 01:24 PM

Hi,

By the time it's been duped about four times, not very much - 2K digital projection, from a 2K electronic origination, is most certainly sharper than any 35mm projection I've ever seen, whatever the other faults may be. This might be because it hasn't had its guts ripped out by the duplication process, but that's the upside of digital. If the DCI people manage to make 4K happen, that'll be great, but people do need to realise that it'll make the current 35mm release process look very feeble.

It's a rather unspecific question, really. It's certainly not "atomic level" because you can only make things developable on a per-grain basis; the grains are very much macroscopic.

As has been hinted at already, it depends whether you want to know either -

- What digital resolution you need to scan film at to get every last detail, which seems to be about 6k, or

- What end-to-end digital resolution you need to use to meet or exceed the traditional 35mm production path, which is about 2K.

Some films, such as those shot super-35 on fast stocks and then de-anamorphosed through cheap projection lenses, can have a screen resolution of much less than 2K, regardless of whether they've been through a DI.

Phil
  • 0

#8 Dan Goulder

Dan Goulder
  • Sustaining Members
  • 1259 posts
  • Cinematographer

Posted 14 December 2006 - 01:38 PM

It's a rather unspecific question, really. It's certainly not "atomic level" because you can only make things developable on a per-grain basis; the grains are very much macroscopic.

A single grain can hold "complex" information, thus the atomic reference. Grains also vary in size. It is true that a "grain" is a more useable unit of measurement than the atomic building blocks of which the grain is comprised. I was using the atomic reference merely to make a point that it is difficult to ultimately compare a chemical process with one that is typically defined with a finite number of 1s and 0s.
  • 0

#9 Henri Titchen

Henri Titchen
  • Basic Members
  • PipPipPip
  • 125 posts
  • Other

Posted 14 December 2006 - 03:52 PM

Continuous contact printers are used for most release prints. There are resolution limits due to the physics of the duplication process. It looks like figures of 50 line pairs/mm are achieved with some printers. See specs of a typical printer at http://www.rtico.com/bhp/6127specs.pdf

Henry.

I never seem to be able to get an absolute answer to this question.
I realize there possibly isn't one, but what I read is very confusing.

I have heard the figure of '4000 lines' used but other deny this.
I presume what this is saying is that if the camera was precision framed up to a chart of 2000 north-south black lines on a white background, that is, so that 2000 black lines are focussed across the negative area, the lines should be visible on the processed negative. How visible will of course depend on the efficiency of the focus puller and the quality of the lens used. :D

But given 100 ASA film approximation, the lines should be visible at least on the negative.

Is this correct, and if it is, would you expect to also see the lines on a positive print, made by wet-gate perhaps?

My other question is, I have heard that HD cameras like the Cine Alta can only resolve about 500 black lines on white background. Is this right? It sounds unlikely to me.


  • 0

#10 Werner Klipsch

Werner Klipsch
  • Basic Members
  • PipPip
  • 86 posts
  • Other

Posted 14 December 2006 - 06:34 PM

What end-to-end digital resolution you need to use to meet or exceed the traditional 35mm production path, which is about 2K.


Phil


When you say '2K' do mean a projector with 2000 horizontal projecting elements, which indicates 1000 actaul lines of resolution freed from antialiasing, or one which projects the actual 2000 north-south lines? Or in other words a '4K' projector?




"But given 100 ASA film approximation, the lines should be visible at least on the negative."

Why limit yourself to a 100asa speed film? You could find out the resolution of Kodak Vision 50D, Fuji 64D, Kodak Vision 100T or even a 500asa speed film all within the 35mm format.


I felt 100ASA is the most finegrained film that is easly available and practical to use. I know finegrained slower stocks exist , but they may be harder to get, and need quality lenses and cameras an great skill in use

________________________________________________________________________________
_________

Perhaps I might ask my original questions in another way.

IF you had a test chart with numerous east-west strips of different amounts of north-south lines, so that the top strip had 4000 lines, the next one down had 3500 lines, the next one had 3000 and so on, at which strip of lines would the projected images therafter simply become gray?

This is with 100 ASA film (and of course good lenses and operators with the shooting and the processing :D )

Also, where would this happen for the variety of digital cameras avaijable at them moment (CineAlta, Genesis, Origin, RED and so on)

And for comparison, domestic MiniDV and VHS :P

Edited by Werner Klipsch, 14 December 2006 - 06:35 PM.

  • 0

#11 Brian Pritchard

Brian Pritchard
  • Basic Members
  • PipPipPipPip
  • 328 posts
  • Other
  • Stoke-on-Trent, UK

Posted 15 December 2006 - 07:28 AM

When you say '2K' do mean a projector with 2000 horizontal projecting elements, which indicates 1000 actaul lines of resolution freed from antialiasing, or one which projects the actual 2000 north-south lines? Or in other words a '4K' projector?
I felt 100ASA is the most finegrained film that is easly available and practical to use. I know finegrained slower stocks exist , but they may be harder to get, and need quality lenses and cameras an great skill in use

________________________________________________________________________________
_________

Perhaps I might ask my original questions in another way.

IF you had a test chart with numerous east-west strips of different amounts of north-south lines, so that the top strip had 4000 lines, the next one down had 3500 lines, the next one had 3000 and so on, at which strip of lines would the projected images therafter simply become gray?

This is with 100 ASA film (and of course good lenses and operators with the shooting and the processing :D )

Also, where would this happen for the variety of digital cameras avaijable at them moment (CineAlta, Genesis, Origin, RED and so on)

And for comparison, domestic MiniDV and VHS :P

The usual way to measure the resolution of film and also film chains (printers, lens. projectors etc) is using MTF, modulation transfer function. You start with a pattern of sinewaves of increasing frequency with 100 % modulation, that is the black to white ratio. You then examine the pattern after, for example, photographing it with a camera, by using a micro densitometer and finding out the modulation for each frequency. As the resolution of the system drops the difference between black and white gets smaller until the pattern turns into just a grey area. You can plot modulation against frequency.

Kodak film datasheets always include MTF curves. The system has the advantage that you can calculate the result when you have a chain.

For example if your lens has an MTF of 50% at 200 cycles per mm and you use it to shoot on film which has an MTf of 25% at 200 cycles per mm the resultant image on the film would be 12.5% modulation. When you print your negative the MTF of the printer would come into play as would the MTF of the print film and then when you project the film the MTF of the projector/lens combination comes into the equation. At which point probably the MTF of our 200 cycles per mm pattern would have an MTF of 0%; you would not see the pattern.

5222 has an MTF of around 25% at 100 cycles per mm.

One of the reasons for using MTF is that resolution varies with object contrast; 5222 has a resolution of 25 lines/mm at a contrast of 1.6 to 1 and a resolution of 80 lines per mm at a contrast of 1000:1.

Brian
  • 0

#12 Paul Bruening

Paul Bruening

    (deceased)

  • Sustaining Members
  • 2858 posts
  • Producer
  • Oxford, Mississippi

Posted 15 December 2006 - 11:40 AM

Hello Werner,

You opened a can of worms that cannot easily be closed. Film res and digital res are not easily compared. The quick answers are roughly this: Negative films in the ASA100 range deliver (and I empahsize, roughly) about 5K in equivalent digital, horizontal resolution. How much of that ends up on screen through a back-to-film strategy has much to do with cost. However, many people feel that 2K of the negatives' resolution can be projected and recognized by the human eye. This is (again, roughly) better than an all optical post path that can deliver something around 1.5K to the screen (depending on the quality of the post equipment). Where things are changing and still up in the air is what will become of digital projection. With continued improvements and cost reductions we could eventually see all 5K of the negative's resolution on the screen. I wouldn't mind this at all since it would increase the merit of film as a taking medium. 5K of the aesthetically lovely film, projected forty feet wide might come farely close to what Imax currently delivers.
  • 0

#13 Werner Klipsch

Werner Klipsch
  • Basic Members
  • PipPip
  • 86 posts
  • Other

Posted 15 December 2006 - 05:55 PM

Film res and digital res are not easily compared. The quick answers are roughly this: Negative films in the ASA100 range deliver (and I empahsize, roughly) about 5K in equivalent digital, horizontal resolution.

This is the problem I have. People seem to liberally inter-mix terms for example '4K digital' with '4000 lines film'
However I am led to believe they are not the same thing. As I understand it, a digital projector needs FOUR pixels to define a unitary pixel of film. By that I mean a point on the film which is defined by the crossing of two of the thinnest lines which the emulsion can capture. If the emulsion can capture 5000 lines horizontal for instance, that means 2500 black lines on a white background. A 'unitary pixel' would mean the area of the spot defined by the crossing of two such black lines.
The illusion of the correct position on the unitary pixel is generated by maniopulating the brightness of a group of four actual pixels on the projector. If this is not done, the precise pixel position cannot be so correctly defined and the result is jagged edges and aliasing.

However, many people feel that 2K of the negatives' resolution can be projected and recognized by the human eye. This is (again, roughly) better than an all optical post path that can deliver something around 1.5K to the screen

But by this, do you mean a 3K (or 4K) projector is required to show 'around 1.5K' or do you mean a '2K' projector?
  • 0

#14 James Steven Beverly

James Steven Beverly
  • Basic Members
  • PipPipPipPip
  • 4199 posts
  • Director
  • El Paso, Texas

Posted 16 December 2006 - 01:04 AM

I never seem to be able to get an absolute answer to this question.
I realize there possibly isn't one, but what I read is very confusing.

I have heard the figure of '4000 lines' used but other deny this.
I presume what this is saying is that if the camera was precision framed up to a chart of 2000 north-south black lines on a white background, that is, so that 2000 black lines are focussed across the negative area, the lines should be visible on the processed negative. How visible will of course depend on the efficiency of the focus puller and the quality of the lens used. :D

But given 100 ASA film approximation, the lines should be visible at least on the negative.

Is this correct, and if it is, would you expect to also see the lines on a positive print, made by wet-gate perhaps?

My other question is, I have heard that HD cameras like the Cine Alta can only resolve about 500 black lines on white background. Is this right? It sounds unlikely to me.


What I had read in Videography (I think that's were I read the article) a 35mm answer print has the equivelant of 4000 lines of resolution but once a distribution print has be projected a few times that number drops down to the equivilent of 2500 lines of resolution. Of course this purely academic as film can never really be measured in video terms and this is more of an astetic evaluation rather than a scientific one.
  • 0

#15 Paul Bruening

Paul Bruening

    (deceased)

  • Sustaining Members
  • 2858 posts
  • Producer
  • Oxford, Mississippi

Posted 18 December 2006 - 10:36 AM

Sorry, I'll try to clarify: An all optical approach, given the ragged state of projectors and all, is considered to be capable of delivering a dependable 1.5K equivalence to digital. How that figure was arrived at is abit elusive, but many people still traffic on those figures. Of course, if everything were perfect in the optical post production/distribution path, that number would rise. Some prints make it to the screen with about 2-2.5K. I don't think any look that good at my local theaters.

DI skips some of the generation losses of an optical post path. Yet, its print-out resolution depends on how much money the production had for post and what equipment they used to move the data out to the final prints. On the whole, a 4K render on computers in post, recorded out to film on good equipment can deliver somewhere around 2K to your neighborhood cineplex. I've heard people claim as much as 3K. That sounds like a stretch. Maybe, with the latest gear... I don't know. IF the scans are done at 2K, then the prints come closer to a mid-budg optical path at 1.5K.

The thing is, DI is about as good, overall, as all optical, except for the price. However, costs are coming down and storage/resolution issues are dimenishing to the point that whether you go optical or DI may be more of a concern of working habits or aesthetic choices.

I wish I could give you more concrete numbers. I'm not sure anyone can other than salesmen trying to unload an expensive scanner or recorder on you. There are some lab folks who post here that might be able to give you numbers from their personal experience with whatever equipment they have used. Dominic Case comes to mind. David Mullen never states things vaguely and posesses more knowledge in his pinky than most of the rest of us put together. Try their opinions.

The dig/film res thing just doesn't convert well. Not just because of the physics but also the aesthetic reactions.
  • 0

#16 Patrick Cooper

Patrick Cooper
  • Basic Members
  • PipPipPipPip
  • 868 posts
  • Other

Posted 28 December 2006 - 04:06 AM

There's someone who operates a local telecine business over here who claims that 16mm film has the same resolution as VHS tape!
  • 0

#17 Jonathan Bowerbank

Jonathan Bowerbank
  • Basic Members
  • PipPipPipPip
  • 2815 posts
  • 1st Assistant Camera
  • San Francisco, CA

Posted 28 December 2006 - 04:09 AM

There's someone who operates a local telecine business over here who claims that 16mm film has the same resolution as VHS tape!


Dude's got issues.
  • 0

#18 Dennis Kisilyov

Dennis Kisilyov
  • Basic Members
  • PipPipPip
  • 152 posts
  • Other

Posted 28 December 2006 - 09:45 AM

I never seem to be able to get an absolute answer to this question.
I realize there possibly isn't one, but what I read is very confusing.



Here is a dated, however very insightful article, this refers to still photography, and film resolving power with primes on a tripod. But typically resolving power is measured in lpmm (Lines Per Millimeter).

It goes on and on about sharpness of lenses and etc....

In theory the resolving power of film is limited by the size of the crystals in the emulsion, however it's always precieved that you'll get far less, esp at 24fps flying through the camera.

the 50lpmm limits

Edited by Dennis Kisilyov, 28 December 2006 - 09:46 AM.

  • 0

#19 Mark Dunn

Mark Dunn
  • Basic Members
  • PipPipPipPip
  • 2420 posts
  • Other
  • London

Posted 29 December 2006 - 12:50 PM

There's someone who operates a local telecine business over here who claims that 16mm film has the same resolution as VHS tape!

Perhaps it HAS when he's finished with it.
  • 0

#20 Phil Savoie

Phil Savoie
  • Sustaining Members
  • 94 posts
  • Cinematographer
  • Montana / Wales

Posted 29 December 2006 - 02:38 PM

Dear Werner,

If I read you correctly you are looking at comparing film to HD. Your questions are something I?m sure we all ponder as DP?s during the rise of HD origination in today?s production.

I spent months researching formats during pre-production of the BBC Natural History Unit?s PLANET EARTH series. This was over three years ago I should note. We were looking for a real world comparison of HD v film. We utilized resolution target charts with line pairs per mm, latitude tests, studio tabletop macro, and outdoor wides for comparison. We shot Super 16 and Super 35 (3 Perf) on 50, 64, 250 and 500 asa stock (both Eastman & Fuji, daylight & tungsten)vs. the Sony 900 and Panasonic Varicam. For the resolution and tabletop macro we used the same lens, a 60mm Zeiss macro that had been tested, projected and was always shot at its optimum t-stop with lighting varied to facilitate this. For other tests we used Zeiss Ultra primes and Zeiss Digi primes for the HD cameras. Film stocks were rated at 40,50,200 and 400 respectively. I used my own 16 & 35mm Arris that were serviced and registration tested prior. The film was scanned via a Spirit at 2K and transferred to D5 HD. Super 35mm was the clear image quality winner with all asa stocks, although the 500 asa had noticeable grain. Super 16 looked better than HD at 50,64 and 250asa (Eastman 250 not Fuji ? the T-grain made a big difference in resolution), while the HD looked an even draw and at times better than 16mm on 500asa. The tests were subjective ? I don?t have direct numbers for comparison, but of the twenty or so individuals who screened the results we all were in agreement. I should also add the tests were conducted at Arri Media in London with the assistance of both Sony and Panasonic reps and engineers, Zeiss primes from Arri Media and donated stock from both Eastman and Fuji.

In the end PLANET EARTH was shot multi format, 16mm, 35mm and HD ? with producers choosing the right origination format for the given shoot. This is for the main TV series delivered in HD. At present the Beeb is cutting PLANET EARTH the movie, which I suspect will have more 35mm origination.

As I say this was three years ago or so. I know that the HD cameras are getting better but for the type of location run and gun required in nature film I?m personally still favouring film for most shoots for a number of reasons; films picture quality and latitude, robust cameras, low power consumption, optical viewfinders, etc. It would however be interesting and illuminating to re-test HD vs film including the SI and upcoming Red (and others) and a 4/6K scan of the latest film stocks, I?m just waiting for someone to ask me. ; )
  • 0


CineTape

Broadcast Solutions Inc

Rig Wheels Passport

FJS International, LLC

Wooden Camera

Opal

Gamma Ray Digital Inc

Technodolly

Metropolis Post

Ritter Battery

Visual Products

Paralinx LLC

CineLab

Media Blackout - Custom Cables and AKS

The Slider

rebotnix Technologies

Tai Audio

Willys Widgets

Glidecam

Abel Cine

Aerial Filmworks

Metropolis Post

Technodolly

Tai Audio

Glidecam

Wooden Camera

rebotnix Technologies

Abel Cine

Media Blackout - Custom Cables and AKS

CineTape

CineLab

Paralinx LLC

Gamma Ray Digital Inc

Aerial Filmworks

Broadcast Solutions Inc

The Slider

Visual Products

Ritter Battery

Opal

FJS International, LLC

Rig Wheels Passport

Willys Widgets