Understanding Exposure Latitude of Red

[Mike Welle]

I understand that Red has 11.3 stops of exposure latitude. What I'm wondering about is a test that I saw on cinematography.net about the Red latitude:

http://www.cinematography.net/red-exposure.html

It appears, in my opinion, and what I believe would be the opinion of a rational viewer, that based upon these tests, the camera over and underexposes rather easily. My opinion is buttressed, I feel, by studying this other exposure latitute test on Kodak Vision2 and Fuji Eterna film:

http://www.cinematography.net/Pages%20GB/5218%20-%208573%20comparison.htm

The aforementioned tests appear to show that Kodak Vision2 has at least 8 stops of latitude. The Fuji has about 6 stops of latitude, I feel.

Could someone clarify whether Red has more latitude than motion picture film and why these tests seem to show the opposite?

[Gopher77]

Mna thats an old test and been disscussed at length, do a search to find much more updated material here.

[Mark Linthicum]

There is no way the Red camera has more latiude than film!
You must be misunderstanding the tests.

Mark

[Graeme Nattress]

No. RED sensor has 11.3 stops of dynamic range between hard clip at one end and when the information drops below the noise floor at the other end. Of that 11.3 stops, over 10 stops is very linear, the remaining portion less linear as it descends into the noise floor with a bit of a curve. This is because the sensor has >66db signal to noise ratio.

Digital measures very differently to film and works very differently to film.

Graeme

[Gopher77]

Well I think there is a lot more room on both sides of the film tested. Red has about the same lattitude as film, but we're comparing Apples top oranges. Me presonally I think red has plenty of lattitude on a grey scale to accomplish everything I want to do, and lattitude in color we'll wait and see but everything is looking very impressive. It has a 12 bit color depth and contrast seems better then any video camera I've ever used.

Film on the other hand depends on stock, processing and many other factors, you can get 12 to 13 stops out of film if necessary. The one thing I've always liked about film is the gradaiton of color, I think red has that beat. Look at the FAQ in this forum for some comparisions Evin accomplished.

[David Mullen ASC]

There is also a difference between "latitude" and "dynamic range" -- latitude really means the ability to correct a misexposed image, so obviously the latitude will be lower than the dynamic range.

Often when you look at simple over and underexposure tests, you are seeing the practical latitude for shooting, which will be less wide than the actual range that can be captured. You often work practically in a narrower range just for safety's sake, plus the gamma of the final presentation format (like a film print) usually means some loss of exposure information on the original, just so that you can get a deep black and pure white on the screen.

The advantage of camera negative film is that the response to light is not linear, so it gets more low-contrast at the extreme ends of exposure, causing a more gentle roll-off in detail. The ideal digital camera someday would actually have a much wider exposure range than film so that some compression can be applied to the extreme ends to better simulate that gentle roll-off.

[Gopher77]

Good point David, I really shouldn't have been saying Lattitude, instead Dynamic range.

[Mike Welle]

Thank you, David and Graeme, for clarifying the difference between dynamic range and latitude. Since there is 11.3 stops of dynamic range in Red how many stops of latitude are available?

[Evin Grant]

Look at the FAQ in this forum for some comparisions Evin accomplished.

Although I appreciate the kudos it's actually my esteemed colleuge Brook Willard that created the FAQ you are reffering too.

[David Mullen ASC]

Whenever you talk about practical latitude, it is somewhat indeterminate. There are a lot of factors that come into play. Best thing would be to test it yourself and see what works for the way you like to light, expose, and color-correct.

In film, for example, if you preferred the contrast characteristics of the Vision Premier print stock, you'd modify what range of exposures you wanted to work in, so even though a shadow-side of a caucasian face may go black at 5 stops under on the negative, it may go black at 4-stops under on the print, and you may decide that you'll want the shadows to be no darker than 3 1/2 stops under if you want some detail in there. Also, what ASA rating you give the negative affects the contrast to some degree, and also your latitude in over or underexposing by accident.

Same can be true with digital cinematography. For example, in "Collateral" they found that if they exposed faces to look correctly balanced with the nighttime view outside the cab windows, the faces had too much noise in them because of the overall underexposure and gain levels. So they lit the faces a little brighter than they wanted, compared to the backgrounds, and then used Power Windows in color-correction to darken the faces again, thus keeping the noise down. So while there was detail when underexposing the faces more, it was not desirable to expose a face that far down because of noise.

If you play the detail you want to see right out to the far ends of what the system can capture and record, you not only make it harder to move the image up or down in brightness in post (unless you alter the gamma), but a minor mistake in exposure pushes detail at one end or the other over the edge.

This is one reason why every DP will test the contrast range of the format to see what works best for the way they will be exposing, color-correcting, and then presenting the image. The initial image capture is only part of the imagemaking equation, just as in photography, it's really a negative-positive system designed to work together, not just a negative system. You don't just look at a negative; it always gets transformed in some way into a viewable positive with normal contrast.

I've shot thirty features and one TV series, all in film except for eight HD features. I've never really tested the exact dynamic range of the stocks I'm using -- numbers like "11.3 stops of dynamic range" don't really mean much to me when I'm lighting and shooting a scene. I test a format or stock to simply see how it looks, I do some real-world under and exposure tests to see where usable detail starts to drop off, I do some lighting contrast ratio tests to see what I prefer, and then I shoot the movie, usually only metering one area (usually the face) and setting everything else by eye. In some ways, it's the only way to shoot as quickly as needed on most features. You can't take the time to get a hundred spot-meter readings on the scene so you rely, to some extent, for the format to do what you know it will probably do to the extreme ends while you concentrate on the middle range.

[Yash Keough]

There is also a difference between "latitude" and "dynamic range" -- latitude really means the ability to correct a misexposed image, so obviously the latitude will be lower than the dynamic range.

Often when you look at simple over and underexposure tests, you are seeing the practical latitude for shooting, which will be less wide than the actual range that can be captured. You often work practically in a narrower range just for safety's sake, plus the gamma of the final presentation format (like a film print) usually means some loss of exposure information on the original, just so that you can get a deep black and pure white on the screen.

The advantage of camera negative film is that the response to light is not linear, so it gets more low-contrast at the extreme ends of exposure, causing a more gentle roll-off in detail. The ideal digital camera someday would actually have a much wider exposure range than film so that some compression can be applied to the extreme ends to better simulate that gentle roll-off.

Wow. As somebody who is really just a beginner with very little practical experience, that's extremely informative. The part about the difference between film and digital in terms of linear and non-linear curves was particularly interesting. Thanks Mr. Mullen!

[Barry Gregg]

Excellent information Mr. Mullen. Testing and a keen eye. Expose for the most important portion of the image, which is often the face. In my world it is difficult to get the time and budget for testing, but it is critical.
Thank you.

[Greg Cokes]

Hello,

I find this very enlightening as well. To give a bit more prospective for those of us not familiar with film, could someone compare Red's Lat and DR to those of several popular video camera systems? For instance:

Panasonic
HVX200
Varicam

Sony
Z1U
XDCAM 350
F900

Thanks!
Greg
F950

[Andrew M.]

In film, for example, if you preferred the contrast characteristics of the Vision Premier print stock, you'd modify what range of exposures you wanted to work in, so even though a shadow-side of a caucasian face may go black at 5 stops under on the negative, it may go black at 4-stops under on the print, and you may decide that you'll want the shadows to be no darker than 3 1/2 stops under if you want some detail in there. Also, what ASA rating you give the negative affects the contrast to some degree, and also your latitude in over or underexposing by accident.

"it may go black at 4-stops under on the print"

Dave, it mean that when shooting such crucial scenes you have to know up front what print or transfer to copy method you will be using and film it will be printed on?

So you correct not only for stock you are currently using but also for final copy method and stock?
Do we have different stock of final copy we do print on?

[David Mullen ASC]

Generally most D.I. color-correction systems use LUT's to emulate print characteristics when viewing the footage while color-correction -- hence why if you turned the LUT off, you see this milky 10-bit Log image, sort of the digital version of a film negative.

Most D.I. facilities are balanced for a final print on standard Kodak Vision 2383. Now if you want additional snap and saturation, there are better stocks. Either you can just take your film-out and print onto those stocks (which are a little more contrasty) knowing that there will be some slight loss of shadow detail compared to the Vision print, or adjust the color-correction for the higher contrast print stock characteristics (which may screw you if the distributor opts for the cheaper Vision stock in the end.)

For example, "Superman Returns" (shot on the Genesis) was color-corrected for release prints on Vision Premier 2393, a more high-contrast, higher saturation print stock. But the IMAX version had to be printed on regular Vision 2383 because the slightly thicker Vision Premier meant that the movie wouldn't fit on the standard IMAX reels. This upset the DP because there was some noise problems now visible in some shots that were not visible in the Vision Premier prints.

The simple answer is if you are shooting for cinema release in 35mm prints, you always test the entire chain of post from shooting through color-correction and then laser recording and printing, during prep, just to make sure you know how things will look. And most D.I. facilities will do a 1-minute test pretty cheaply for you.

The best example of this is the use of diffusion - many people advocate using diffusion on video camera because they think that digital "looks too sharp", confusing resolution (typical HD camera has less of that than 35mm) with edge-enhancement. So they slap a #1/2 Black ProMist on the camera, but when they do the film-out to 35mm and project the result on a cinema screen, they go "hey, it looks soft!" It's because they were judging sharpness on some 23" HD CRT monitor, which tends to exaggerate edge detail and make everything seem super-sharp. They weren't accounting for the look of a film-out and the resolution needs of projection on a large screen, which all act to some degree as a softening step.

[S. Um]

I've never really tested the exact dynamic range of the stocks I'm using -- numbers like "11.3 stops of dynamic range" don't really mean much to me when I'm lighting and shooting a scene. I test a format or stock to simply see how it looks, I do some real-world under and exposure tests to see where usable detail starts to drop off, I do some lighting contrast ratio tests to see what I prefer, and then I shoot the movie, usually only metering one area (usually the face) and setting everything else by eye. In some ways, it's the only way to shoot as quickly as needed on most features. You can't take the time to get a hundred spot-meter readings on the scene so you rely, to some extent, for the format to do what you know it will probably do to the extreme ends while you concentrate on the middle range.

David,

I'm glad you've become a regular poster on this board. We really appreciate your depth of experience and your willingness to share. And it's nice to see that even on big movies there's time pressure and sometimes you just go with your gut feelings.

[Gopher77]

Although I appreciate the kudos it's actually my esteemed colleuge Brook Willard that created the FAQ you are reffering too.

Sorry Brook, I was tired and only had one eye open when I was posting. But it is a very nice comparison of film to red.

[Jonathan Payne]

I don't have a lot of experience Sumfun, but I think the irony of your statement is that it should really read: "Especially on big movies there's time pressure and you just go with your gut feelings."

[Phil Bates]

This discussion remindes me of problems I have with the way video clips the highlights. I could accept a little less dynamic range if the roll-offs at either end were more like film. For example, when shooting into a sunset, somethings going to clip, even with the best dynamic range. With video, the reds clip first leaving a posterized ugly neon yellow to handle the brightest parts of the image. This is the biggest issue I have with the F900, even when using the "knee". Digital SLRs seem to handle this better, and I hope the Red follows suit.

Another problem I have with clipped in video is "ringing", faint echos or ghost images, again, especially on high contrast edges.

If anyone "in the know" would comment on these issues, I would appreciate it.

Phil
www.artbeats.com

[Mike Welle]

On page 149 of The Filmmaker's Handbook by Steven Ascher and Edward Pincus it is stated:

"Confusingly, the word latitude is used to mean both the total exposure range and the degree of acceptable error (in the latter case, it is sometimes called exposure latitude). If someone says, "This stock has a latitude of ten stops," they mean the total exposure range the stock is capable of. If they say, "You have two stops of latitude on the overexposure side," they're talking about exposure error. As we have discussed, latitude as a meaure of acceptabel exposure error depends not only on the exposure range of the stock, but on the contrast of the particular scene."

According to this definition, it seems to me that I was correct in saying that Red has 11.3 stops of exposure latitude, and it seems to me that, based upon this definition, latitude "can be used used to mean both the total exposure range and the degree of acceptable error." Therefore, I believe, that I was not incorrect in saying that that Red has 11.3 stops of exposure latitude (unless "total exposure range" means something different than "dynamic range.) Furthermore, the link I provided in cinematography.net was entitled "Red exposure latitude":
http://www.cinematography.net/compare.htm
For these reasons, I don't think I was using the wrong term in saying "latitude."

[Stephen Williams]

On page 149 of The Filmmaker's Handbook by Steven Ascher and Edward Pincus it is stated:

"Confusingly, the word latitude is used to mean both the total exposure range and the degree of acceptable error (in the latter case, it is sometimes called exposure latitude).

Hi,

IMHO the book is wrong.

Stephen

[Graeme Nattress]

That's why we prefer to say dynamic range or SNR, rather than use the word lattitude, which is open to confusion.

Graeme

[Mike Welle]

According to the definition I found in The Filmmaker's Handbook I believe I was incorrect in saying "Red has an exposure latitude of 11.3 stops," I would have been correct in saying "Red has a latitude of 11.3 stops."

[Graeme Nattress]

Just say Dynamic Range instead, and then that removes the confusion. Next we get into differences of opinion on how to measure DR, but that's another story....

Graeme

[David Mullen ASC]

You're right, it's not really a wrong or right thing -- we all tend to toss the term "latitude" around freely (causing some confusion) but I think there is a distinction to be made between a technical number of exposure stops that the camera records versus a practical range that the cinematographer will work within.

I just think that sometimes people can get too anal-retentive about these sorts of things; ultimately what matters is how it looks and how you can deal with it practically on the set. Whether the RED camera can capture 10.9, 11.3, or 12.2 stops of exposures doesn't necessarily affect how you might light and expose a scene because you would never actually place an object right on the edge of being recorded, either at the bottom or the high end. And if you wanted something to go pure black or pure white, you'd probably overshoot the ends just to be sure (or count on post color-correction.)

I think some of these dynamic range figures are used too much as gospel as an attempt to beat-up on another format or camera, that's all. Honestly, I doubt the RED camera has the same dynamic range as color negative, at least not the wide-latitude stocks available like Fuji Eterna 400T or Kodak Expression 500T, in particularly how bright highlights roll-off, just based on the fact that I've never seen digital still photography match that range, and based on my experience with digital cinematography. But I won't push that belief too hard without shooting my own tests.

However, there is simply the practical issue of whether it is close enough to be workable and film-like, and I think the highest-end digital cinema cameras are starting to achieve that, so in my mind, while I would like manufacturers of digital cameras to keep working on that issue, improving dynamic range, it is almost a non-issue for most shooting situations except for saving grossly overexposed shots, which hopefully you won't be doing too often.

[Mike Welle]

That's why we prefer to say dynamic range or SNR, rather than use the word lattitude, which is open to confusion.

Graeme

I see, thank you for the clarification, these respones have been very helpful.:biggrin:

[Stephen Williams]

According to the definition I found in The Filmmaker's Handbook I believe I was incorrect in saying "Red has an exposure latitude of 11.3 stops," I would have been correct in saying "Red has a latitude of 11.3 stops."

Hi Mike,

IMHO the camera has a quoted dynamic range of 11.3 stops, the author of the book is incorrect.

Assuming your scene has a dynamic range of 7.3 stops you could then assume that you have a latitude of +/- 2 stops.

Stephen

[Vincent Rice]

Great input Mr. Mullen, thank you.

[Mike Welle]

The real reason why I posted doesn't have to do with splitting hairs about the word latitude but rather the two tests in the original post. According to the tests, my eye tells me that you must be very accurate with your exposures when using Red in order to record as much usable information as possible. However, when reading other posts in other red user forums, it seems that you will be able to pull back information if you mis-exposed (is that a word?) your Red footage. However, through looking at those tests on cinematography.net, I just can't believe that. Not to follow in the folly of our so-called "great" commander in chief, but my gut (and more importantly my eye) tells me that that white "blown-out" data is irretrievable. My gut and eye also tells me that the data on the film, didn't blow out, and is therefore workable. Shakespeare this isn't. Please, pardon my wordiness.

[Sanjin Jukic]

Guys this discussion is not making sense.

Why?

Because the film is having other latitude or dynamic range!?? FILM>>It is totally another and an old technology than digital. The RED is the best you can get today and even tomorrow from the digital acquisition technology. Don't forget this. Just because the RED take care about all you need now and in the future. It is about 100% 4K digital acquisition easy to get work.

I would say all those discussions about the RED latitude or dynamic range are not making sense.

And just a couple days ago Jim said:

" Major Announcement...
We are very happy to announce that we have a significant upgrade to the RED ONE camera coming in the form of an update to one of the major boards in the camera, effectively further increasing dynamic range and reducing noise in the image. Our original goal was to complete this upgrade later in the year, but we now plan to implement this improvement prior to shipping any cameras. It will cause a minor delay, but we are absolutely certain that every customer will feel the improvement was worth the wait. It also eliminates a future board change down the road.

We will publish a new shipping schedule that reflects this update in a couple of weeks.

Jim"
LINK>>http://www.reduser.net/forum/showthread.php?t=2027

I do not understand what people are asking for? If it is not enough for them OR they should give back or look for another camera.

[Graeme Nattress]

With Digital you ETTR - or Expose to the Right, protecting highlights - then you get your DR by how deep you can go into the shadows. All very similar to digital stills photography - this is a good site for learning about it: http://www.luminous-landscape.com/

Graeme

[Mike Welle]

I must admit that I do take offense to the comments that "remarks are not making sense" (please save me the, chastising comments about "guys this is not a mud-slinging contest." I'm not slinging mud, I'm simply pointing out, that I don't think saying "remarks are not making sense" is rude.)

I, personally, would not say such a thing to anyone, regardless if I though it, because I know it would hurt their feelings. I am simply putting forth a "dichotomy" between the film test and the Red test I saw on cinematography.net. It is appropriate, I think, that both tests were on cinematography.net, and they both appeared with a similar layout (frames on top of frames). If the fact that I'm presenting a comparision between two tests (one in film the other with Red) on cinematography.net doesn't make sense than I am truly puzzled as to why? I understand that you said two formats. I understand that digital works differently than film. But what I'm asking is does Red have less latitude, than film. I think the answer is yes as the first responder roughly said through admonishing me (also, please save me the directives to previous forums). I think I'll take a rest from this site for a while while I grieve the barbs I have taken. Thanks for being so tolerant.



Guys this discussion is not making sense.

Why?

Because the film is having other latitude or dynamic range!?? FILM>>It is totally another and an old technology than digital. The RED is the best you can get today and even tomorrow from the digital acquisition technology. Don't forget this. Just because the RED take care about all you need now and in the future. It is about 100% 4K digital acquisition easy to get work.

I would say all those discussions about the RED latitude or dynamic range are not making sense.

And just a couple days ago Jim said:

" Major Announcement...
We are very happy to announce that we have a significant upgrade to the RED ONE camera coming in the form of an update to one of the major boards in the camera, effectively further increasing dynamic range and reducing noise in the image. Our original goal was to complete this upgrade later in the year, but we now plan to implement this improvement prior to shipping any cameras. It will cause a minor delay, but we are absolutely certain that every customer will feel the improvement was worth the wait. It also eliminates a future board change down the road.

We will publish a new shipping schedule that reflects this update in a couple of weeks.

Jim"
LINK>>http://www.reduser.net/forum/showthread.php?t=2027

I do not understand what people are asking for? If it is not enough for them OR they should give back or look for another camera.

[Graeme Nattress]

RED is as much about education as it is cool technology. And there's a lot of us here, both with RED and in the industry who love to help and educate.

Graeme

[istvanttt]

I also learned a lot in this threat, so to me it made sense to read all the written comments. Thank you who started it and to all who partecipated.

[Jannard]

#1- the old posts on C.net were short on dynamic range due to my error. That has been discussed forever on other threads at what a goof-ball I was.

#2- shortly after, without an error on my part, we demonstrated 11.3 stops on Frankie.

#3- We have announced a break-through in our program that will cause a short delay in shipping, but one that will please our customers.

#4- RED is not film. And film is not RED. There are advantages to each. We did not enter this market to replace film. We entered this market to provide a "filmic" digital alternative to film.

Jim

[Shawn Nelson]

Jim, so the breakthrough, #2, is going to push it beyond 11.3 stops? I know this seems like a stupid obvious question, but I have been puzzling about the upgrade for a while. You've said you can't go above 12 stops due to the 12 bit system and since you're already at 11.3, there doesn't seem to be room for a "breakthrough". I'm not challenging you, it's just been confusing me.

[Jannard]

We are just getting started (15 months). There is room to improve in this area... and we are on it.

Jim

[Alvise Tedesco]

15 months = ...! stops:biggrin: Yesss, you're on it! We love you

[Michael Brennan]

Another way to look at ccd/cmos vs film is that given a sunny outdoor scene both achieve their maximum dynamic range at a precise exposure level.


But given that film neg exceeds the dynamic range of display systems (including interneg/pos for film release) you can over or under expose neg and still have an acceptable image.
This is called "exposure latitiude".


But the dynamic range of existing cmos/CCD and recording systems is closer to that of display systems and so has less scope to be pulled around in post.

We could use a term like "post latitude" to describe the ability of a particular camera/recorder to be manipulated in post.

This phrase factors in the sensors dynamic range and noise, recording bit depth and how much the resultant image can be manipilated in post before noise or banding becomes an issue.


Mike

[Sanjin Jukic]

I must admit that I do take offense to the comments that "remarks are not making sense" (please save me the, chastising comments about "guys this is not a mud-slinging contest." I'm not slinging mud, I'm simply pointing out, that I don't think saying "remarks are not making sense" is rude.)

I, personally, would not say such a thing to anyone, regardless if I though it, because I know it would hurt their feelings. I am simply putting forth a "dichotomy" between the film test and the Red test I saw on cinematography.net. It is appropriate, I think, that both tests were on cinematography.net, and they both appeared with a similar layout (frames on top of frames). If the fact that I'm presenting a comparision between two tests (one in film the other with Red) on cinematography.net doesn't make sense than I am truly puzzled as to why? I understand that you said two formats. I understand that digital works differently than film. But what I'm asking is does Red have less latitude, than film. I think the answer is yes as the first responder roughly said through admonishing me (also, please save me the directives to previous forums). I think I'll take a rest from this site for a while while I grieve the barbs I have taken. Thanks for being so tolerant.

I would say that nobody here on this forum is "rude'. Nothing personal or even "rude' was involved in my opinion and writings here. It was about the facts and evidences only. The Reduser.com should be a strong and valuable discussion forum. If you are too "soft" for a discussions like this then don't enter the high-tech GLADIATORS ARENA! And the Gladiators are also gentle and polite.

[Graeme Nattress]

Mike, Dynamic range curent CMOS / CCD also extends beyond what current display technology is capable of....

Graeme

[Stephen Williams]

#2- shortly after, without an error on my part, we demonstrated 11.3 stops on Frankie.

Jim

Hi Jim,

Any possibility to release that frame to people who weren't there at the time.

Stephen