Jump to content


Photo

Difference between Contrast and Latitude


  • Please log in to reply
20 replies to this topic

#1 Sean Conaty

Sean Conaty
  • Basic Members
  • PipPip
  • 48 posts
  • Cinematographer
  • Los Angeles

Posted 04 March 2009 - 10:44 PM

I was asked this question and realized that, although I can recognize contrast vs. latitude, i can't actually define it. I was hoping someone could succinctly clarify the differences between the two.

here's where my confusion lies, if a film stock is considered contrasty, i understand this to mean the rate at which the image falls off to black or white, regardless of its latitude. what confuses me is how it's theoretically possible to have a film stock with increased latitude while also being very contrasty; if it falls off faster, wouldn't it then mean it can see less, and therefore have less latitude? or is this fall off (and my understanding of contrast) strictly in the shoulder and the toe, while the mid range is unaffected and able to see more?

I apologize for being so pedantic, but I recognize my ignorance due to years of misinformation and confusion.
  • 0

#2 Mike Lary

Mike Lary
  • Basic Members
  • PipPipPipPip
  • 472 posts
  • Other

Posted 05 March 2009 - 12:29 AM

In an instance where two characteristic curves are the same length, it would be the shape of the curve that defines whether one is high or low contrast relative to the other. A more linear curve would be lower contrast than one with more shape to it, because the latter would have areas where the tonal range is compressed. Theoretically (photochemistry of our stocks aside) a curve could fall off gradually on both ends but have compression in the center, resulting in higher contrast in the midtone region.
  • 0

#3 Sean Conaty

Sean Conaty
  • Basic Members
  • PipPip
  • 48 posts
  • Cinematographer
  • Los Angeles

Posted 05 March 2009 - 12:52 AM

A more linear curve would be lower contrast than one with more shape to it, because the latter would have areas where the tonal range is compressed.


So here's where I get confused. Assuming the latitude is the constant between 2 stocks. If the slope is steeper in the stock with more contrast (A), then what happens in the shoulder and the toe if they are now essentially elongated? In other words, let's assume the slope on the less contrasty stock (B) begins at 2 stops under 0 and ends at 2 stops above 0. If we compare this to the more contrasty stock (A) whose slope begins at 1 stop under 0 and ends at 1 stop above 0, what then happens in the region between -2 and -1 & 1 and 2 of the more contrasty stock (A), as it compares to the stock with less contrast (B)? Does this simply mean that the difference between A and B is that more contrasty stocks (A) have a longer shoulder and toe than less contrasty stocks (B) in order to compensate for their steeper slope? And what does this mean in terms of the image if the shoulder and toe are longer? Does have anyone have any examples?

Thanks!
  • 0

#4 Mike Lary

Mike Lary
  • Basic Members
  • PipPipPipPip
  • 472 posts
  • Other

Posted 05 March 2009 - 01:26 AM

Does this simply mean that the difference between A and B is that more contrasty stocks (A) have a longer shoulder and toe than less contrasty stocks

I think a simple answer might be that the slope is greater in higher contrast stocks, hence the shoulder and/or (not 'and') toe would be longer.
  • 0

#5 Simon Wyss

Simon Wyss
  • Basic Members
  • PipPipPipPip
  • 1404 posts
  • Other
  • Basel, Switzerland

Posted 05 March 2009 - 01:29 AM

I perfectly understand your confusion. It derives from the fact that any given film can be processed in various ways.

Let us stay with black and white, because all this does not apply to colour stock where (almost) everything is standardised.

Now, as you certainly know, there are different films. One is a highly sensitive rather low-contrasty emulsion, an other one reacts more harshly, and so on. A developing formula given, different stocks may be processed with the identical chemistry and then characteristic photographic curves established. Yet, we must not forget that we can develop a film to a certain contrast while not developing it out to full density. So, a film's contrast range will be defined only when developed out.

To change contrast at full density we shall have to employ other chemistry. That is why there are so many formulas, and sometimes somewhat mysterious ones.

I have to say, as a lab manager, that the cinematographer needs to develop her or his taste in order to cook or to let others cook. Of course, there is only so and so much contrast to be displayed on the average cinema screen. Starting from the cinema conditions and calculating backwards over the printing process, the camera original and its treatment we finally arrive at certain light(ing) levels for the shoot. That is what I have been teaching in courses since years.

Back in the Fourties a director of photography would have called for more or less light with stop 1:5.6 already set. Admittedly there was nothing around like hand held cameras, cinéma vérité, available light filming or more than 200 ISO.
  • 0

#6 Brian Pritchard

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

Posted 08 March 2009 - 04:40 AM

The characteristic curve of a film stock is ‘S’’ shaped. The bottom part of the curve is called the toe and the top part the shoulder. Between the two lies the ‘straight-line’ portion. The ‘straight-line’ portion is not always completely straight, it varies from stock to stock.

There are a number of ways of measuring contrast but basically the contrast is the slope of the straight line portion; the steeper the line the higher the contrast.

The latitude of a film can be considered as the distance between the beginning of the toe and the end of the shoulder. It describes the brightness range that a film will accommodate. A normal subject on a correctly exposed negative will occupy the toe and some distance up the straight-line of the curve.

If you increase the exposure it will move up the curve until eventually the highlights of your scene will reach the shoulder. Increasing the exposure further will cause crushing of your highlights entailing a loss of detail. The difference between normal exposure and the exposure when you reach the shoulder is the latitude of the film; the higher the contrast of the film the less the latitude because the shoulder is closer to the toe of the film.

The latitude will also depend on the brightness range of the original scene; if you are shooting a black cat on a snow field you will have a large brightness range and much less latitude than if you were shooting in a fog where the brightness range is very small.
  • 0

#7 Karel Bata

Karel Bata
  • Basic Members
  • PipPipPipPip
  • 487 posts
  • Director
  • London - a rather posh bit

Posted 08 March 2009 - 07:35 AM

My understanding is that latitude is the total range the film neg can capture. Contrast, as stated above, is the steepness of the characteristic curve. So a more contrasty stock will have a steeper curve and will exhibit less overall latitude. Forced processing will compress all this (compress horizontally if you imagine the traditional graphical representation of a characteristic curve) and give you increased contrast with decreased latitude.

I don't think anyone's making the confusion here (but for other readers' benefit ) it's worth noting that the latitude of a stock is not the same as the tonal range in the final print. The neg captures more information than is seen in the print - usually a stop's worth or more in the shadows and highlights (can anyone here put a figure on this?). Hence you can print up or down to get more detail at either end, but not both at the same time. Unless you do some clever stuff in DI.

I've had a lot of fun in the past using freddie Francis' invention the Varicon. This does something very similar to flashing by effectively lifting the toe of the curve, with the advantage over traditional flashing in that you can see the effect through the viewfinder. Maintaining continuity between shots was a challenge though! All sadly gone with DI...

I'm now doing some experiments trying to create HDR tone mapping effects in moving pictures. Currently HDR filming (video actually) is limited to time-lapse. example One limitation is the latitude of film stock. So I'd like to ask: does anyone know of any tricks for extending a stock's latitude? Or is there some esoteric stock that has that characteristic?

Edited by Karel Bata, 08 March 2009 - 07:36 AM.

  • 0

#8 Brian Pritchard

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

Posted 08 March 2009 - 08:39 AM

My understanding is that latitude is the total range the film neg can capture. Contrast, as stated above, is the steepness of the characteristic curve. So a more contrasty stock will have a steeper curve and will exhibit less overall latitude. Forced processing will compress all this (compress horizontally if you imagine the traditional graphical representation of a characteristic curve) and give you increased contrast with decreased latitude.

I don't think anyone's making the confusion here (but for other readers' benefit ) it's worth noting that the latitude of a stock is not the same as the tonal range in the final print. The neg captures more information than is seen in the print - usually a stop's worth or more in the shadows and highlights (can anyone here put a figure on this?). Hence you can print up or down to get more detail at either end, but not both at the same time. Unless you do some clever stuff in DI.

I've had a lot of fun in the past using freddie Francis' invention the Varicon. This does something very similar to flashing by effectively lifting the toe of the curve, with the advantage over traditional flashing in that you can see the effect through the viewfinder. Maintaining continuity between shots was a challenge though! All sadly gone with DI...

I'm now doing some experiments trying to create HDR tone mapping effects in moving pictures. Currently HDR filming (video actually) is limited to time-lapse. example One limitation is the latitude of film stock. So I'd like to ask: does anyone know of any tricks for extending a stock's latitude? Or is there some esoteric stock that has that characteristic?


Print stocks are capable of recording all the information captured in a negative. You are more likely to lose information if you under or overexpose the negative, particularly if you are recording a scene with a large brightness range.

Although you have all the information on the print stock the amount of information you see will depend on the projection method and its contrast range. 5393 Vision print stock is capable of densities up to 5.0, a contrast range of around 64,000 to 1. Some LCD TV's are only capable of a contrast range of 1500:1 similarly digital projectors.
  • 0

#9 Karel Bata

Karel Bata
  • Basic Members
  • PipPipPipPip
  • 487 posts
  • Director
  • London - a rather posh bit

Posted 08 March 2009 - 09:28 AM

Print stocks are capable of recording all the information captured in a negative.

Now you have me worried :( but I do think you're wrong there. Low contrast print stocks can capture it, but anything meant for projection will need more oomph.

"print film has less latitude and higher contrast than negative film and a gamma of 2.6 as opposed to .6 for negative." American Cinematographer, April 2005 'The color Space conundrum'
  • 0

#10 Brian Pritchard

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

Posted 08 March 2009 - 12:36 PM

Now you have me worried :( but I do think you're wrong there. Low contrast print stocks can capture it, but anything meant for projection will need more oomph.

"print film has less latitude and higher contrast than negative film and a gamma of 2.6 as opposed to .6 for negative." American Cinematographer, April 2005 'The color Space conundrum'


Yes, colour print has more contrast than negative; it has to have. In order to get the gamma of reproduction to the figure of around 1.5 you have to have a contrast of 2.6. 2.6 X 0.6 = 1.56 (theoretically the gamma of reproduction should be 1.0 but because of flare, light within the theatre etc. the contrast is usually 1.5). With colour print you can fix the exposure exactly to use the entire length of the curve, you very rarely use the entire length of the negative curve. If you did there would be no latitude at all, if you under or over exposed you would lose detail in the highlights or shadows.

I have every confidence that print stocks are capable of reproducing every thing that is on the negative. You can calculate this by taking the density range of the neg stock and equating it to the exposure range of the print stock.

Can you seriously imagine that the stock manufacturers would make a print stock that is incapable of reproducing what is on the negative?

Brian
  • 0

#11 Karel Bata

Karel Bata
  • Basic Members
  • PipPipPipPip
  • 487 posts
  • Director
  • London - a rather posh bit

Posted 08 March 2009 - 04:53 PM

Looking at your link Brian it would seem that you have to know what you're talking about. But it just doesn't tally with my experience.

For instance, if you print a scene up from neg then some detail in the shadows will appear. Likewise printing down will bring out some detail in, say, clouds. I always regarded that 'hidden' stuff as part of what is meant by latent image. Am I wrong? It seems to me that if there is no such latent image then changing the printer lights would be a bit like adjusting the 'brightness' knob on a domestic TV - turn it up and the black level just comes up and there's no extra detail to reveal.

Recently I've been trying to explore just how much (of what I regard as) latent image might be buried in the highlights and shadows (for potential use in HDR). Here's a few results using an old fairly random, deliberately unimpressive, stills neg. taken on Kodak stock you'll need password 55555 Stills neg stock behaves much the same as movie stock, or have I got that wrong..?

The first pic is what the default setting on the scanner gave, and looks very like what the photo print service provided. But turn the light on the scanner down and some detail from the clouds appears. Turn it up and you see a little more detail from the shadows. Not a lot, but it's there. The final print is a Photomatix HDR composite wiith a little tweaking of contrast and saturation (and mending the scratches!).

Now, I would say that the first print is very much the kind of image you would expect to get if that had been shot on film - the sky burns out. But what you seem to be saying is that all that info that was brought out in the second and third images (which is what I'd also expect to see with film) was actually there in the first one because everything that is on the negative is reproduced.

So, where am I going wrong here? :(
  • 0

#12 Sean Conaty

Sean Conaty
  • Basic Members
  • PipPip
  • 48 posts
  • Cinematographer
  • Los Angeles

Posted 09 March 2009 - 01:58 AM

the higher the contrast of the film the less the latitude because the shoulder is closer to the toe of the film.


This answers it perfectly. I guess what I was trying to quantify was the exact relationship between the latitude and the contrast. If I understand it correctly, it is therefore theoretically impossible to have a high contrast stock with much latitude as well.

in terms of negative vs positive, i agree with Karel. While a negative may be able to capture up to 13 stops, print stocks don't have nearly that range.
  • 0

#13 Simon Wyss

Simon Wyss
  • Basic Members
  • PipPipPipPip
  • 1404 posts
  • Other
  • Basel, Switzerland

Posted 09 March 2009 - 02:10 AM

So a more contrasty stock will have a steeper curve and will exhibit less overall latitude.

Or is there some esoteric stock that has that characteristic?

First point: wrong. The more contrasty stocks in fact can digest more latitude, up to the contrast of 1 to 10,000 with sound negatives (log D 4 to 5).

Second point: Yes, Gigabitfilm. You will be able capture nine stops, a mathematical contrast of 1 to 1024, with it at useful gamma values. But beware, who can take such contrasts to the screen?
  • 0

#14 Brian Pritchard

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

Posted 09 March 2009 - 04:38 AM

First point: wrong. The more contrasty stocks in fact can digest more latitude, up to the contrast of 1 to 10,000 with sound negatives (log D 4 to 5).

Second point: Yes, Gigabitfilm. You will be able capture nine stops, a mathematical contrast of 1 to 1024, with it at useful gamma values. But beware, who can take such contrasts to the screen?


The definition of latitude according to 'Fundamentals of Photographic Theory' by James and Higgins of Eastman Kodak is:
'The projection of the straight line portion of the characteristic curve on the Log E axis determines the log exposure range over which direct proportionality exists between D and Log E. This Log E interval is called exposure latitude and is expressed in exposure units.'

From this it can be seen that the higher the contrast of a film the lower the exposure latitude. A sound negative has a very small exposure latitude even though it is capable of a very high maximium density.

The article goes on to say that 'exposure latitude' should not be confused with total scale which is the exposure range within which the material is capable of rendering differences in object luminance by density differences. The total scale of most negative materials is considerably larger than the luminance range of the majority of scenes.

As I have previously mentioned the amount of latitude you can obtain will be dependant on the brightness range of your subject. The bigger the range the less latitude you will have.

Brian
  • 0

#15 Karel Bata

Karel Bata
  • Basic Members
  • PipPipPipPip
  • 487 posts
  • Director
  • London - a rather posh bit

Posted 09 March 2009 - 05:58 AM

The article goes on to say that 'exposure latitude' should not be confused with total scale which is the exposure range within which the material is capable of rendering differences in object luminance by density differences.

Ah... I've always thought that latitude was the total scale of the neg stock.

Interestingly that definition of latitude excludes the toe and shoulder.

Edited by Karel Bata, 09 March 2009 - 06:01 AM.

  • 0

#16 Simon Wyss

Simon Wyss
  • Basic Members
  • PipPipPipPip
  • 1404 posts
  • Other
  • Basel, Switzerland

Posted 09 March 2009 - 09:03 AM

Brian, I fully agree with that but things change, though. James & Higgins of EKC are fine, so is The Theory of the Photographic Process edited by Charles Edwin Kenneth Mees, if you know it.

The question was: Is there a film which might offer more exposure latitude and react with a useful steepness? One can ignore Gigabitfilm. My answer remains positive. Gigabitfilm is processed with its own very special chemistry. Any gamma between 3 and 0.3 is feasible. The emulsion will always be completely developed out to full density (around log 2.4). So classic theory doesn't hold true any longer.

By the way, Gigabitfilm curves have a very short toe portion and then a very long straight one until Dmax.

www.gigabitfilm.de
  • 0

#17 Karel Bata

Karel Bata
  • Basic Members
  • PipPipPipPip
  • 487 posts
  • Director
  • London - a rather posh bit

Posted 09 March 2009 - 01:28 PM

I don't suppose you do a color version? :D

It does look very interesting, but all those chemicals...

I was wondering if shooting with some low contrast interneg stock might get the increased latitude I'm looking for. Or is that completely nuts? I've seen Adrian Biddle use titles stock! Bear in mind I'm after the HDR effects shown above. Though that test I did seems to indicate it's actually possible with regular color stock by extracting info around the heel and shoulder that's outside of the 'latitude'.
  • 0

#18 Gregory Middleton

Gregory Middleton
  • Basic Members
  • PipPip
  • 78 posts
  • Cinematographer
  • Vancouver BC

Posted 09 March 2009 - 02:21 PM

I thought I would add something not related in technical terms.
If you are evaluating a shot you will more easily notice contrast if it was high than a very wide latitude.
How your eye sees contrast is all within the image. You can see when the shadows on the dark side of a face become very intense, how fine details stand out against a lighter color etc. Think of a Bleach By Pass print for example.
Latitude is better noticed if you were present when the shot was taken. You then have a reference by eye of all the relative brightnesses in the scene.

Now factor that in to the physical limitations of the Print you want to create. Super low contrast stocks can give you detail everywhere but the image will have no true Black or White. Balancing those 2 things is the tricky part.

If you have the exposure, Pull Processing will reduce contrast and saturation.

In the Digital realm you now have almost complete control of the 'Curves' so you can really experiment.



I don't suppose you do a color version? :D

It does look very interesting, but all those chemicals...

I was wondering if shooting with some low contrast interneg stock might get the increased latitude I'm looking for. Or is that completely nuts? I've seen Adrian Biddle use titles stock! Bear in mind I'm after the HDR effects shown above. Though that test I did seems to indicate it's actually possible with regular color stock by extracting info around the heel and shoulder that's outside of the 'latitude'.


  • 0

#19 Simon Wyss

Simon Wyss
  • Basic Members
  • PipPipPipPip
  • 1404 posts
  • Other
  • Basel, Switzerland

Posted 11 March 2009 - 02:37 AM

In the Digital realm you now have almost complete control of the 'Curves' so you can really experiment.


Can you?

Consider how the image is finally displayed. Ambient light with an electr(on)ic monitor, beamer image, LED arrays, front and back projection.

We have a consistent environment with darkness alone. We can install a white surface with almost identical properties in any corner of the world with limewash. Our discussion on contrast revolves around the (film) positive. Copies can be made on a variety of stocks.

For instance, Orwo LF 2 is a fine grain stock, orthochromatic, on a colourless base. Maximum density around log 5. Or Gigabitfilm HDR 32 which can be brought to any gamma with densities up to log 4.5. One has to know what one wants to show.
  • 0

#20 Dominic Case

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

Posted 11 March 2009 - 05:20 PM

Sorry to chime in so late on this, but I have to have a word or two on this one.

Contrast and latitude are obviously related, but you can't compare them any more than you can compare apples and trees.

Start with "dynamic range" or "sensitivity range". That is the distance between the start of the toe and the end of the shoulder (measured along the exposure axis) on the characteristic curve of the emulsion. It's also called "useful exposure range". The "start of the toe" etc was defined mathematically and very strictly, but nowadays tends to be based on subjective ideas of "acceptable images" etc. It's still a vital concept though.

Then you have the "brightness range" of the scene. That's the range of all tones you want to record and distinguish from adjacent tones. Blown highlights and deep shadows don't count.

LATITUDE is "useful exposure range" minus "brightness range". So id does depend on the scen you are shooting. If it's a traditional bride & groom in strong light and no fill, you've got wide brightness range and little or no latitude (or even negative latitude on a dodgy prosumer video format). If it's Birmingham on a foggy day, it's all shades of mid grey: minimal brightness range and heaps of latitude: you could overexpose or underexpose several stops and get away with it.

High contrast stocks have less latitude, because the steeper curve (higher gamma) goes from toe to shoulder in less exposure range. Shooting an average scene on hicon stock (titles stock or sound negative for example) you will inevitably miss out on shadow or highlight detail. Same goes for bleach bypassing - it steepens the curve.

Finally, I would argue with Brian about print stocks. Certainly they have higher contrast (the gamma rule sees to that) and certainly they have a much greater density range - going up beyond a density of 4.0 in the case of Kodak Premier print stock. (Though densities much greater than around 3.5 are a waste, as tthey are totally fogged out be the ambient light in a cinema).

But just as a negative can often not capture the entire brightness range of a scene (except in foggy Birmingham see above), the sensitometry of the emulsions will show that a print won't necessarily reproduce the entire range of tones in a negative. That is where film grading come in - making a lighter or darker print means that you favour one part or another of the negative tonal range.

Put another way, negative stocks can typically capture more range than you are shooting for, or expect to see on the screen. In making a print, you would normally discard shadow or highlight detail (tones that are recorded on the neg even if you didn't shoot for them), or you have make a low contrast print, (which itself has more useful exposure range to the negative), which results in a flat and lifeless image.
  • 0


Gamma Ray Digital Inc

Abel Cine

Opal

Media Blackout - Custom Cables and AKS

Ritter Battery

Rig Wheels Passport

Paralinx LLC

FJS International, LLC

CineLab

Glidecam

Visual Products

rebotnix Technologies

Willys Widgets

Aerial Filmworks

Technodolly

Tai Audio

Broadcast Solutions Inc

CineTape

The Slider

Wooden Camera

Metropolis Post

Glidecam

Metropolis Post

Technodolly

Ritter Battery

The Slider

Visual Products

Paralinx LLC

Opal

CineLab

Willys Widgets

Aerial Filmworks

FJS International, LLC

Rig Wheels Passport

Media Blackout - Custom Cables and AKS

CineTape

Gamma Ray Digital Inc

Broadcast Solutions Inc

Wooden Camera

Abel Cine

Tai Audio

rebotnix Technologies