Epic and Scarlet release in 2010

[S. Matthews]

I believe new technologies like this done properly would be hard to categorize with the techniques we have now hence the new technology, cause if those other methods worked don't you think we would have seen HDR video done on a single sensor before now?

 

[Razz16mm]

No one knows yet what Red has come up with, but it is not two exposures in the conventional sense of the term like one would do with at still camera. That much has been confirmed. They have previously announced a dual ISO range feature that allows sensitivity to be optimized for either a 25 to 250 ISO range or an 200 to 2000 ISO range. It may be related to that feature in some way which could give a single exposure with a 5 stop increase over the normal linear 13+ stops by modifying relative highlight and shadow exposure sensitivity on the fly. But this is just speculation on my part.

Properly tone mapped, HDR compresses a wider dynamic range linear digital capture into a logarithmic gamma curve that more closely approximates human visual perception in a narrow dynamic range presentation medium, whether that is a print or electronic display.
The result should be better detail and tonal range in the shadows without noise, and better detail and tonal range in highlights without burnout. A lot of tone mapped HDR images take it too far and end up looking very cartoonish and unnatural, but done well it can make an image really pop.

 

[Tzedekh]

I believe new technologies like this done properly would be hard to categorize with the techniques we have now hence the new technology, cause if those other methods worked don't you think we would have seen HDR video done on a single sensor before now?

Maybe we have -- see here.

 

[Otis Grapsas]

I believe new technologies like this done properly would be hard to categorize with the techniques we have now hence the new technology, cause if those other methods worked don't you think we would have seen HDR video done on a single sensor before now?

In these affordable electronics markets, we have seen multi-slope response solutions that gets the HDR, but it looks bad due to mapping and the discontinuities creates by the slope changing points. We have also seen bracketing tecniques that work impressively well for subjects without motion. These are actually far better than any single exposure technique because they can look natural on HDR. High enough frame rates and enough processing power for artifact reduction such as deghosting can make these usable in video.

There are some extremely high cost HDR sensor technologies with log response, but they all hit the mapping problem. Would you like a 28 stop sensor that maps the typical photographically useful dynamic range into a very small (low resolution) part of the digital output? There are 22 stop SD sensors of this type for special applications. Most of the times only computers see the images.

We will see commercially available HDR video from the big manufacturers when it becomes practical, affordable and competitive in image quality. Until that time comes, most photographers would rather light to a lower contrast ratio or sacrifice the highlights than compromise motion or the quality of the core photographical dynamic range.

 

[Razz16mm]

I could add another 20 methods to the list, but all have artifacts. The methods that do work involve either huge multisensor setups (limited for resolution by allignment), pixels with different sensitivities (fuji), and sensor implementations inherently capable of high dynamic range (multi slope linear, currently problematic).

Arri Alexa is increasing dynamic range captured by using dual A/D paths and simultaneous readouts for each pixel at different gain levels.

 

[Otis Grapsas]

Maybe we have -- see here.

I won't do technical or implementation comments on this since it is in development and there is intellectual property involved:

http://vimeo.com/13513876

 

[Otis Grapsas]

Arri Alexa is increasing dynamic range captured by using dual A/D paths and simultaneous readouts for each pixel at different gain levels.

Based on their own tests, it doesn't look like it's working! The pixel size is very large and the dynamic range they achieve is quite logical for the pixel size. Much larger pixel than Red MX but not much better in dynamic range.

 

[Razz16mm]

Based on their own tests, it doesn't look like it's working! The pixel size is very large and the dynamic range they achieve is quite logical for the pixel size. Much larger pixel than Red MX but not much better in dynamic range.

I think the Alexa is still very much a work in progress too, even though it has been released.

 

[lawriejaffa]

Hey Otis, yep just what you said about the interest the defence industry's interest in an 18stop sensor (alone) had me giggling.

I mean I would simply guage the development of technology with the competition. I too am wary of the boasts made in some styles of marketing and how it can distort not only attitudes towards products, but the development of products too.

For example, if the marketers sell on pixel numbers, suddenly everything is sold to chase pixels instead of the very technology that would provide the genuine aesthetic to aid digital cinematography.

So I appreciate Otis providing his point of view in relation to these marketing claims. If not discussed they become 'fact' to an audience that shall we say, would benefit from a sense of deeper scrutiny.

 

[Jason Ramsey]

I'm sure they will supply images once the HDRx is more fully baked... This was the first run of it ever... Jim started to post an image of their first test, but thought better of it, recognizing people's tendency to over-scrutinize images from a prototype camera using a function that it has never used before...

Anyways, I'm sure eventually that there will be images supplied for people's scrutiny and subjective opinions, etc... But, HDRx is still a ways off as it is not likely to be an enabled feature when the first production EPIC's begin to ship, according to RED...

 

[Ryan E. Walters]

I believe we should all wait and see, the proof will be in the images when the working man gets his hands on these cams.

You mean like this footage? http://reduser.net/forum/showthread.php?t=48362
(The footage is from the scarlet, and it is shooting the epic.)

However, as we all now neither of these cameras are available just yet ...

 

[Otis Grapsas]

What I find strange is how Jim Jannard approaches the issue of competing with film: "1) resolution 2) dynamic range 3) "feel".

The resolution of Red One is already better than 35mm. Even Scarlet will be generally better than 35mm IMHO. But Jim keeps pushing to 28K.

The dynamic range of MX is adequate. Even though negative gets additional dynamic range that leaves some room for correcting exposure, there is always a sacrifice in image quality. And once it ends up on a film print, the print shapes the dynamic range so hard that the highlight and shadow range is crushed into a lot less than 10 stops.

Feel is last on the list?! The feel is where it's at IMHO. The overexposure character, the skin tones, the texture, the look film MTF provides. These are issues that require an approach by aesthetics, you can't solve them by engineering alone. No amount of dynamic range will give you smooth overexposure and the right skin tones. And some technical improvements, like increasing resolution, producing even higher MTF and producing even cleaner output can even move in the opposite direction, producing a sterile image that is far from being filmic and inviting to the viewer. Red is moving closer to a computer graphics rendering look with every new model. There is magic in film, magic that is the work of highly skilled engineers that also were/are people with great aesthetics. We can't treat that by numbers, we can only observe and learn and try to make digital better using whatever experience and knowledge we can extract from our observation. Film is the golden standard, and it's not because of resolution and dynamic range.

The german HDR camera is a good example. I will ignore the optical implementation. The samples are supposed to demonstrate high dynamic range. They even demonstrate how you can correct lighting ratio in post. That will make shooting easier, not to mention less challenging and artistic from a photography point of view. But the output looks like a cctv camera in terms of skin tones and feel. Dead, flat, unsaturated. When the customer notices that, and mentions the HVX shot he loved, what am I going to say? I know it looks bad but I had great room to move in post? You can pay less because you we didn't need a photographer or any lights on set? In order to work with this material you need to be an expert in camera design! You need to know gamma curve design techniques, saturation correction techniques, very good color correction in general etc. The samples IMHO simply demonstrate that one can get superior results with a 7stop camcorder where the manufacturer handles everything in camera.

 

[Michael Olsen]

The resolution of Red One is already better than 35mm. Even Scarlet will be generally better than 35mm IMHO. But Jim keeps pushing to 28K.

Well, I'd say that depends on who you ask. There seem to be a number of people who are well respected in the community (e.g., ASC, BSC folks) who find the RED to lack in resolution in relation to 35mm film.

Then, consider the fact that some of the same people are quite enjoying the FF35mm format of the 5D. And then the fact that some folks (Nolan) are once again looking at 65mm. And then, of course, there are the people that have actually made their desire for a 28K 3:1 camera well-known - lots of visual effects, landscape, architectural, and stock footage folks would love to have such a camera.

Not to mention that Jim likely has some unique personal motives for 28K. I seriously doubt he sees that becoming the standard for motion picture acquisition. Rather, I think he understands the artistic abilities of Medium Format and Large Format cameras and would like to have something which exists in that market. Remember that Jim is a pretty big camera buff - he owns dozens and dozens and dozens of cameras, including two that went to the moon. I think a part of the reason 28K is on the roadmap is simply so that Jim can have one.

The dynamic range of MX is adequate.

Personally, I agree. Looking at greater public concensus, I disagree. Again, there are plenty of ASC and BSC folks (along with many others without such credentials) who would like to see more. In fact, virtually every test I've read of RED and film together states that the RED produces a good image, but doesn't yet match the DR of film. People want more.

Feel is last on the list?!

Well, to be fair, Jim didn't say this was a list in order of importance. Placing something last on a list can mean one of about three things.

1. I placed this last because it came to mind last and I feel it is genuinely less important than the other two.

2. I placed this last because it is more important than the others and I wanted to suggest that it is more than a simple specification (i.e., one can note a "4K Camera" and "N amount of DR" in a spec sheet. They can't note "Film-like look".)

This is also known as "saving the best for last".

3. I placed this last because the list is in random order and it drew the short straw.

That being said, and once again attempting to represent the cohort of what I've read and viewed - people would like to see RED take an even more "filmic" look and feel.

 

[Ryan Patrick O'Hara]

The Red Vs Film Resolution argument has so many variables, it's sometimes not worth venturing into.

Does each photosite on a bayer pattern chip count as a pixel, or a full pixel of image? Some say yes, some say no, because it takes many surrounding photo sites to make complete 'pixel'.

It also depends how both are displayed. Most 35mm motion picture film is scanned to 2k, 3k, or 4k for projection... and digitally projected. Not a fair comparison either. Also if you are lucky enough to see a film print... it depends on quality. Sometimes, after 5 generations, and depending on the quality the studio deems the prints receive, a 35mm print can be exhausted. Film negative, inter-positive, inter-negative, and then prints. The inter-negative is where the color timing is usually done, AND, each inter-negative is only good to be struck into so many prints. So if the studio, in an effort to save money, prints more prints from each inter negative, those copies of the film will have quality issues. Those are probably the ones sent to Kansas or Oklahoma.

I've seen Baraka on 70mm film and it was off the inter-negative :shocked:. Something people almost never do, since it's up there right off the negative! This is why dailies look so nice. They are right off the negative!

Anyhow, I am off track... then you have to discuss 'apparent' resolution. Things like noise or film grain, and how they are structured can effect a person's perception of resolution.

Film also hold greater *ahem* dynamic range (thanks guys!), which can effect resolution as the appearance of greater range, smoother tonal transitions, and a greater amount of information in the screen can look like more resolution... but it's just more information. Is that the same? You decide.

I know that I've written to David Mullen about how resolution effects the appearance of dynamic range and how in return, dynamic range effects the appearance of resolution.

Hello, David Mullen.

We've met briefly on a few occasions, such as the ASC open house and etc. I always try to save my questions for the good ones, and today I think I've found a good one for ya.

I was recently reading the July 1998 issue of American Cinematographer, and I came upon the article for the best film ever, Armageddon... haha! Despite the film, the article by David E. Williams was quite good and a wonderful read! In the article, Micheal Bay and DP John Schwartzman, ASC discussed using Super 35 on The Rock, and how they were not happy with the process, liking true anamorphic widescreen better. This made me very happy as I am in love with true anamorphic origination.

However, in their brief dissertation on the anamorphic medium, Schwartzman made a comment regarding an advantage of anamorphic which I was unaware. According to Schwartzman, the larger negative improves latitude in the shadow detail. I am a very young cinematographer, so I apologize if this is a well known fact, but for me, I was completely unaware, and confused on how this happens.

Besides the differences in lens characteristics, I thought shooting the same cut of film stock whether it be in 8mm, 16mm or 35mm had the same characteristics such as latitude, speed ASA/ISO, color rendition, grain structure, etc. The only major difference being, when projected the resolution and apparent size of the grain structure is obviously lacking, the smaller the negative.

To make a long winded thought process come to an end, does a larger negative size increase latitude of a film stock? The only thing I can think of is that Anamorphic is known to have monster flares... perhaps it is the bouncing and refracting of light inside anamorphic lenses which cause a slight "flash" effect (around 10%?), raising the latitude in shadow detail?

Here is the quote:

Quote:
From American Cinematographer, When Worlds Collide by David E. Williams (July 1998)

Though Schwartzman and Bay expressed enthusiasm for the Super 35 process while shooting The Rock, the theatrical prints were a bit of a letdown for both men.

"The drag about Super 35 is the grain and its 'optical' feeling. We did about 30 ENR-treated prints on The Rock to keep some of the contrast. Those were shown in major cities, but the other prints lost a lot of snap. The film looked good for Super 35, but we were still working with this tiny negative."

- Micheal Bay

"What became very apparent to me was that Super 35 is not just an optical process that makes the grain more apparent; the grain is also bigger because it's enlarged so much during projection. You're getting boned on both ends. The beauty of anamorphic is that there is no intermediate optical process. If you like your dailies, you're going to love your release print. The larger negative also gives you greater shadow detail and greater latitude, so even though I was shooting deeper stops in 'Scope, I felt I was using [relatively] less light to get more image.

On Conspiracy Theory, I was doing very large night exteriors in New York City, and I needed to be working at least a T4 or 4.5 for them to look good. But that didn't mean I had to light everything to that exposure. If I could get the lenses to that range, I found that the level of shadow detail I could get in the darker areas was quite extraordinary. One of the things I explained to Michael on Armageddon was that for shuttle interior scenes, I was going to be shooting at a T4.5. I might only have a T2.8 on the actors' faces, but he'd be able to read them beautifully even though they would be underexposed by a stop-and-a-half. The faces wouldn't be muddy, just dark. I was able to do that simply because of the resolving power you get with anamorphic's big negative."

- John Schwartzman, ASC

So can shooting 5219 on anamorphic yield greater shadow latitude then shooting the same scene with the same stats (shutter, lighting, T-stop, etc) on Spherical lenses? Does shooting 5219 yield greater shadow latitude then shooting 7219?

Thanks for your time!

Best,
- Ryan


The Answer by David Mullen, ASC:

"Depends on how you define "latitude" -- larger negatives capture more fine detail with less grain, so you have more flexibility in making corrections before grain kicks in, and in softer muddier light, the larger negative still resolves detail. Plus a larger negative will differentiate between an area made up of tiny bits of color, like a field of flowers or even fleshtones -- smaller negatives, because they capture less information, tend to reduce fine gradations of color into a single color. I remember seeing a 70mm print of "Far and Away" (shot in 65mm) and there were subtle shades of pinks and golds in warm colors, colors I hadn't quite seen before in 35mm.

But in terms of actual dynamic range and contrast, technically the size of the negative doesn't matter, but again, by dint of capturing more fine detail and subtle variations, you preserve information with a larger negative that may drop away or blur in a smaller negative, giving the impression of more depth to the shadows. Plus if you shot in Super-16 or Super-35 and did the conversion to a 35mm sound format using an optical printer, as was done until recently, you did have contrast build-up from duping, which did lose dynamic range.

So it's sort of a yes and no answer -- if your recording format has more resolution, it picks up more texture and variation in detail, contrast, and color, then dark and bright areas near going black or white will appear to have more "life" to them and thus give the impression of containing more dynamic range.

Plus you have to factor in that both film and digital pick up shadow information in the bottom end, the noise floor or murky grainy level, and if that noise/grain is reduced because you used a larger negative (or a sensor with less noise) then more of that low-end information is usable in the final color-correction. So less noise/grain can give the impression of creating more dynamic range because more of the range is actually usable."

- David Mullen, ASC

Of course, you've also got to take into consideration what is the film versus Red arguement counting? 4k? versus what? Super35? Regular 35? Anamorphic?

From my standpoint, Film is still king... because of the following:

-as of now, S35mm has more grains that pixels.

-It has three layers of emulsion in a depth configuration which make up the image.... not side by side dependent photosites, taking up surface area and creating an artificial, yet pretty accurate, guestimations based on neighboring sites.

- If photochemically shot, processed, released and on a good release print (or interpositive) it could kick some ass.

As interest in preserving film through the pipeline and especially for presentation fades, it will be hard for the little superiority film still has to be appreciated or witnessed. Dumping film to 4k for DI, then to 2k for digital projection/distribution, it will be difficult, besides aesthetic on-screen differences, to make an arguement for superior resolution.

 

[dadoboy]

I would be careful when visiting reduser.net right now. There's a google advisory against malicious software at the site. My XP box was infected badly last week. I thought it was just me, but it looks like google sees the problem.

 

[Ryan E. Walters]

The dynamic range of MX is adequate. Even though negative gets additional dynamic range that leaves some room for correcting exposure, there is always a sacrifice in image quality. And once it ends up on a film print, the print shapes the dynamic range so hard that the highlight and shadow range is crushed into a lot less than 10 stops.

Having worked with the Red, and the MX chip for a long time now, personally the DR on the old sensor and the new sensor is not adequate for me. The MX is a GREAT improvement over the old chip, but it still has a way to go before it is as good as, or better then film. Hopefully, with the 16 bit capture and processing in the Epic, and the HDR mode allowing 18 stops of DR, it will give Film a serious run for it's money.

Just because a delivery format may have a limited range, that doesn't mean you want the capture device to be limited to that same range. Shooting with higher DR and resolution then delivery allows for higher quality results if handled well in post - making for a high quality end product.

Just my $0.02 anyway ...

 

[Ryan Patrick O'Hara]

Having worked with the Red, and the MX chip for a long time now, personally the DR on the old sensor and the new sensor is not adequate for me. The MX is a GREAT improvement over the old chip, but it still has a way to go before it is as good as, or better then film. Hopefully, with the 16 bit capture and processing in the Epic, and the HDR mode allowing 18 stops of DR, it will give Film a serious run for it's money.

Just because a delivery format may have a limited range, that doesn't mean you want the capture device to be limited to that same range. Shooting with higher DR and resolution then delivery allows for higher quality results if handled well in post - making for a high quality end product.

Just my $0.02 anyway ...

18 stops of DR is unarguably past film. The latitude will have yet to be seen... but there is not ONE Kodak filmstock that boasts an 18+ stop range. 13-15 maybe. Reading the sensitometry curve of VisionIII 5207, I count 14 stops at most. Will that be enough? I find with proper tools, such as lighting and grip to control lighting... 13 stops is enough for me. I can bring highlights down or fill shadows well enough to get my desired range. I'll take 18, but not sure I'd be jumping up and down for more. I mean, I doubt exterior contrasts for exposure reasons outrange 15 stops or more.

 

[dadoboy]

Whether its film or the RED HDR mode or the next camera by whomever, all claims to latitude and DR should be tested by yourself in real world conditions, and to your standard as to what is acceptable image and what is not. They're usually woefully short on what is claimed.

If in fact any technology can shoot 18 or more stops of latitude, and that technology / camera becomes prevalent in the industry, you're going to find that even more (than it already is) of the "look" (color, contrast, mood, etc). will not be determined so much by the DP anymore as the colorist. Not that the DP is not absolutely needed, but much of the creative control will be transferred far later down the workflow.

 

[Ryan E. Walters]

18 stops of DR is unarguably past film. The latitude will have yet to be seen... but there is not ONE Kodak filmstock that boasts an 18+ stop range. 13-15 maybe. Reading the sensitometry curve of VisionIII 5207, I count 14 stops at most. Will that be enough? I find with proper tools, such as lighting and grip to control lighting... 13 stops is enough for me. I can bring highlights down or fill shadows well enough to get my desired range. I'll take 18, but not sure I'd be jumping up and down for more. I mean, I doubt exterior contrasts for exposure reasons outrange 15 stops or more.

Whether its film or the RED HDR mode or the next camera by whomever, all claims to latitude and DR should be tested by yourself in real world conditions, and to your standard as to what is acceptable image and what is not. They're usually woefully short on what is claimed.

If in fact any technology can shoot 18 or more stops of latitude, and that technology / camera becomes prevalent in the industry, you're going to find that even more (than it already is) of the "look" (color, contrast, mood, etc). will not be determined so much by the DP anymore as the colorist. Not that the DP is not absolutely needed, but much of the creative control will be transferred far later down the workflow.

Yep- I'd be happy with an actual 13 - 15 stop range. (The MX I've found to be around 11 stops.) So I'm hoping that the Epic will be a published 18 - 20 with a real 15 - 17. That would be killer for me as it would be at the same to above what film can do. But the proof will be in the details when the camera ships. (BTW- the HDR mode will not be enabled when the camera first ships.)

 

[Otis Grapsas]

The dynamic range of the camera has nothing to do with how the camera treats highlights (false color, look of overexposure clipping) or the actual dynamic range in the shot on the ouput. It can just provide room for setting exposure with more freedom. The overexposure artifacts are due to inferior gamma curves.

In order to produce a good looking output you need a certain contrast in the midrange and you need to choose a point that will map 18% grey to 50% video. This is the limiting factor, not the sensor dynamic range. Most camera manufacturers put that point 2.5 to 4 stops below clipping because of this limitation. In the majority of dSLRs it's 3.5 stops. Even the 15 stop dSLRs do only 4stops over and 11 stops under mid grey (with only 6 or 7 visible). They get cleaner shadows and that's it. If you try to put more than 3 stops above the 18% point, the extra stops are heavily shaped into something that is nearly white. You get no dynamic range advantage on the output and you start wasting saturation as you approach the highlights.

The key to a rich high dynamic range look is the right gamma curves. If you don't have that, you can have just one of the two, a rich saturation/contrast with low dynamic range look or a poor saturation/contrast with a high dynamic range look. This is not something that can be realistically handled shot by shot in post. The camera manufacturer must build the advantage in the camera gamma curves. This is the way to the film look. If a camera has a good set of gamma curves, any sensor will look very filmic in the way it handles exposure and overexposure, the only difference from one sensor to the other is how clean the shadows will be. From personal tests with HDR, the difference in the shadow noise once the sensor is capable of 11 linear stops is not that important and further improvement does nothing for the highlight part of the dynamic range problem. Even the most accurate film models that match the behavior of film, cannot use more than 11 stops.