Comparing Film to Red , Tests and Example images

[Kyle Dones]

For what it's worth, since there has been a lot of scientific talk, I have asked my girlfriend to pick out the film and the RED image. She thought the film still was RED because the actual RED still looked better therefore couldn't be RED. Just wanted to toss in the opinion of a regular non cinematic mind. We may be the target market for these companies products but the opinion of the general public matters too. Even if they couldn't care less.

As for my opinion, I stopped liking film's aesthetic after Man of Steel so it's irrelevant what I think. (That doesn't reflect my opinion of the film itself)

 

[Marc Wielage]

Marc,
The scans you see were from a custom area array scanner, using a 4.8K Kodak sensor. It was also scanned on a Northlight at 6k, but they looked pretty much the same.... no further detail was showing up.

The Northlight 2, to me, is the best film scanner in the world in my experience. I haven't used the later Arriscanners, but I've used everything else. Not a fan of the Imagica scanners.

The StEM test material comparing digital cameras and film, shot by members of the ASC, is available from SMPTE in both 2K and 4K files:

http://store.smpte.org/category_s/31.htm

If you're seriously trying compare digital and film, this is the way to do it. I've seen this projected on the Academy's screen at their Wilshire building, and it's very impressive. I think many, many DPs in the audience were kind of floored, because they thought there'd be a huge visible difference, and there was not. And this is a relatively old test. I believe there was an issue of American Cinematographer that had more details on how the tests were done, as well as interviewing Alan Daviau and others on their specific lighting techniques for each sequence.

And again: my joke for years about what "log scans" are is that if you take the same piece of 35mm negative and send it to ILM, eFilm, Technicolor, Company3, and Imageworks, you'll get five different pictures back. There is no real standard of what a raw scan is, because it's a moving target. I always insisted on getting a balanced log DPX scan based on the specific piece of film as it was shot... meaning, it requires human judgement, experience, and interaction with the device. Still, a good colorist can make pictures out of damn near anything, but it's more a question of where the starting point is set.

Digital pictures are no different, but I think you can get closer provided all the onset dailies monitors are calibrated and the procedure is coordinated with the editing facility and the facility doing the final color timing. If they're all using different monitors pulled at random out of the box, all bets are off.

 

[David Mullen ASC]

In some ways, StEM was more about creating high-quality film images with movement, contrast, etc. that could be used as a reference for judging things like digital projectors, etc.

The most recent PGA/ASC test assessing the state of digital cameras, film, and 4K workflows is probably available from the PGA by now. It was finished using ACES workflow in 4K and then recently some of it was redone in non-ACES mode to see if any cameras like the Epic benefited from that (one person described the results of the second test as a giant step sideways.) I saw the ACES version projected in 4K at Sony. Even on a big screen, the differences in resolution and quality were not dramatic. As you'd expect, the F65 seemed to have the most resolution followed closely by the Epic, then 35mm and the Alexa were tied after that, but you had to spend some time staring at fine detail like signage in the frame to determine that.

 

[Freya]

This test was done using 50D, the slowest and finest-grained stock out there, but it's not grainless.

Am I right in thinking that 50D is not the sharpest tho? I seem to remember someone saying that 200T or maybe even the older 100T was sharper but I might be a false memory. I think it's a bit strange to talk about the best film stock, but then I kind of feel the same about video cameras almost too. It depends on what you are trying to do and what you want it to look like.

I also don't think 50D is that uncommon, I certainly see people shooting with it quite often in American Cinematographer but then it is only really going to get used for exterior scenes and there often seem to be shoots that just like to compromise on a single film stock, even for outdoors.

Lastly I like the grain! I know some red users like to add some noise or fake grain or the like to try and make things more like film so I can't be alone in liking the grain.

I loved the film, but was surprised at just how well the Red MX did in these tests. Aside from the skin tones problem it did really well.

Freya

 

[David Mullen ASC]

50D is not quite as sharp as 100T was (now discontinued) but I think it's a bit hair-splitty... Some of it is just visual perception, visible grain gives the impression of sharpness because our eye has something to focus on. But for some reason, 100T had a bit more sharpness, maybe the grain had more acutance, I don't know.

 

[Freya]

Maybe there was just something about tungten as opposed to daylight balance that made it easier in that context.
I just remember someone directing me to those special Kodak graphs with the curves and going on about the l/pm etc and being slightly blinded by the science.
I remember it took Kodak a long time to come out with a vision 50D, so I got the impression it must have been difficult for them. To be honest I wished they hadn't as I LOVED the old EXR 50D. That was a magical stock!

Anyway...

Freya

 

[PatrickFaith]

Modern T-grain is more uniform, tighter, and more even -- older film stocks had more random, coarser grain. But older film stocks were slower so they weren't too grainy unless underexposed and/or push-processed. But many remasters of old movies aren't able to use the original negative, so you also have some grain from the duplicate elements used, whether color interpositives or dupe negatives, or b&w separations. And whether those dupe elements were originally made on an optical printer or contact-printed will effect the graininess. Plus older movies often contain an optical printer dupe element for an effect (fades, dissolves, titles, visual efx) and that shot will be grainier than the rest of the footage.

This test was done using 50D, the slowest and finest-grained stock out there, but it's not grainless.

I have been having a lot of problems on distribution tests of my monochrome epic footage because of 8 bit compression artifacts being created where I have gradated shadows(blu-ray and dvd compression is driving me crazy). Graeme suggested I dither the pixels a bit, and I am finding randomly dithering across red/blue/green (each color is random on the dither) ... is helping out a lot. I have also noticed this looks similar to how some of the older movies that were shot in mixed black&white(BW) and color, I think the intermediate was mixed BW-color-saturation filter(?), then the distribution was often in color or BW. Question is why was this was done and/or some of the ascetics in grain when doing these types of processes?

 

[David Mullen ASC]

I have also noticed this looks similar to how some of the older movies that were shot in mixed black&white(BW) and color, I think the intermediate was mixed BW-color-saturation filter(?), then the distribution was often in color or BW. Question is why was this was done and/or some of the ascetics in grain when doing these types of processes?

Not sure what you are describing... b&w separations of old color negative movies? The b&w negatives used in 3-strip Technicolor cameras? Movies with b&w shots and color shots printed on color stock versus splicing b&w print stock and color print stock?

 

[PatrickFaith]

Not sure what you are describing... b&w separations of old color negative movies? The b&w negatives used in 3-strip Technicolor cameras? Movies with b&w shots and color shots printed on color stock versus splicing b&w print stock and color print stock?

The effect as scene by the movie audience was a "black and white" film, but you could see a full "color" range in the grain when projected on the screen. When I go back and look at the DVD's and Blu-rays of these movies i don't see this effect though. Only major movie where it kind of looks that way is in some of the shots on "2001", where it almost looks like it was shot in b/w with a small tint. Thanks for reviewing, I'm not exactly sure how this was done (and i don't understand how color was desaturated in the film days).

 

[David Mullen ASC]

Well, a b&w section in a color movie is usually printed on the same color print stock as the rest of the roll. There are a few exceptions, b&w movies with small color inserts where they spent the money to hand-splice the color print shots into every b&w reel shipped out, as was done for "Rumblefish", "Kafka", "Schindler's List". Otherwise you have to print the b&w image onto color print stock and it almost always picks up a tint.

In the video transfer, you just have to turn off the color when transferring the b&w sections so it is a purely b&w image.

 

[PatrickFaith]

Well, a b&w section in a color movie is usually printed on the same color print stock as the rest of the roll. There are a few exceptions, b&w movies with small color inserts where they spent the money to hand-splice the color print shots into every b&w reel shipped out, as was done for "Rumblefish", "Kafka", "Schindler's List". Otherwise you have to print the b&w image onto color print stock and it almost always picks up a tint.

In the video transfer, you just have to turn off the color when transferring the b&w sections so it is a purely b&w image.

Thanks David, that makes sense.

 

[Charles_Z]

The Northlight 2, to me, is the best film scanner in the world in my experience. I haven't used the later Arriscanners, but I've used everything else. Not a fan of the Imagica scanners.

What don't you like about the Imagica scanners? Wasn't the LOTR trilogy DI done on those things?

Also, what's your opinion of the DFT Scanity head-to-head with the Northlight 2?

 

[Phil Holland]

Heh. I'll just say I agree with Marc here and have a similar "use of everything else". The Northlights are a pleasure to use as well.

 

[Marc Wielage]

What don't you like about the Imagica scanners? Wasn't the LOTR trilogy DI done on those things?

Honestly not sure what Park Road Post was using. I think every scanner has a characteristic "look," and there's something I can't put my finger on that I don't like about the Imagica's. Purely my opinion.

Also, what's your opinion of the DFT Scanity head-to-head with the Northlight 2?

The Northlight 2 to me has a better transport, can handle problematic film better, and I think the stability and overall color rendition is better. But the DFT's (formerly Thomson Spirit 4K's) are infinitely faster and less costly to run. If I had a low-budget project that I had to crank out very quickly -- trailers would be a good example -- I'd scan all the selects on the Spirits. If it was a feature where quality was of the utmost importance, I'd lean towards the Northlights, cost and time no object. I might do a "shoot out" with the Arriscanners, which I have the least experience with.

The important point here is that saying "here's a film scan, and it looks like X, and here's a digital image of the same scene, and it looks like Y," is an incredibly misleading statement, because there are so many variables. In truth, I think so much depends on workflow, LUTs, and signal path (and final color), I think you can degrade anything in the process, and you can enhance anything in the same process, to get almost whatever results you want. Note also that lenses and contrast have a big effect on perceived sharpness.

There are huge differences in 8-bit cameras compared to 10-bit cameras, and those can't be disguised by technical tricks or color-correction. I have seen cases where DPs have thrown in Canon 7D's in the middle of scenes shot with (say) Red Cameras, and we've had to degrade the Red footage in order to match it to the Canon's. No amount of correction could save the Canons. In an Apples-to-Apples situation, like a Red thrown in with an Alexa, I'd expect the two to both look very good, assuming good exposure and good lenses. I bet we could make them look identical. I've never had the chance to match an Alexa to a Red, but I did once work on a show that used the Arri D21 and a Red One, and those weren't too far off. If anything, I think the Red was slightly worse, but I think there were issues with the crew on that project [which is still on the air, and I shouldn't name it].

 

[dio Zafi]

TEST IS NICE..BUT EVERYONE HAS TO TEST THE BEST CAMERA FOR THE JOB...FILM OR DIGITAL...THEN GET THE BEST LIGHTING PACKAGE...GRIP..ETC...(camera is only a part of it)

 

[Phil Holland]

What don't you like about the Imagica scanners? Wasn't the LOTR trilogy DI done on those things?

Honestly not sure what Park Road Post was using.

From what I recall they were using Imagicas. Oddly never toured the scanning facility on my trips to Wellington.


On LOTR 1 and 3 we had a few small shots we scanned on an older 4K scanner that was eventually replaced by a Northlight. Over time and a great deal of testing between houses Northlights became rather popular out here in LA.

 

[Marc Wielage]

The effect as scene by the movie audience was a "black and white" film, but you could see a full "color" range in the grain when projected on the screen. When I go back and look at the DVD's and Blu-rays of these movies i don't see this effect though. Only major movie where it kind of looks that way is in some of the shots on "2001", where it almost looks like it was shot in b/w with a small tint.

Hmmm, I don't remember any B&W shots in 2001. I can tell you that on Soderbergh's The Good German, they shot it on color Kodak negative, but then completely desaturated it and timed the whole film in B&W, then output to Kodak 2302 B&W print film for theatrical release. I'm not sure what they did for Good Night and Good Luck -- also shot in color, but desaturated and timed in B&W for release -- but my memory is that this also went on B&W stock. These movies shot on color negative mainly because of availability and wanting to get the dailies out fast, and (at that time) the LA labs couldn't turn B&W negative over very quickly.

Mr. Mullen is absolutely right that printing B&W images on color stock is problematic at best. Even when we can get it 100% perfect in an initial Answer Print, the release prints are all over the place just because of variabilities in the developer. In a good sense, with an all-digital world, that's not nearly as big a problem, and the digital releases should be pretty much 100% correct if the projector is set up correctly. The digital showings of The Artist that I saw were damned near perfect (and had really good grayscale performance). I'm not sure how they handled release printing, since I think that was done in France.

It'd be an interesting test to shoot some B&W negative and B&W digital and see how those compare when precisely matched in a timing theater. There's clearly a different "feel" to it, and I've come upon situations where even with custom curves and special LUTs, the end results don't quite have the same impact. Again, the film scanner characteristics comes into play as a big factor.

 

[David Mullen ASC]

I think Patrick is referring to the Stargate sequence with the multi-colored landscapes going by, where they made b&w separations from the color helicopter footage and mis-combined them through filters to create false colors. I think Patrick thought that it was b&w footage that had been printed with a tint, which is semi-correct, it's just that it was color footage separated to RGB b&w elements.

Also, a number of spaceship shots, particularly during the Blue Danube sequence, are b&w still photos cut-out and animated on an Oxberry, but I don't think Patrick was referring to that.

The lack of grain in digital movies converted to b&w in post can look a bit odd to me, especially if combined with clippy digital highlights. To me, it's a new b&w look, not a classic b&w look. Generally I like it just because I like monochrome and I like b&w lighting techniques, but sometimes it lacks the heart of true b&w photography (or even color film converted to b&w). But a lot of this is due to whether they are willing to make the blacks really black in color-correction.

 

[Gabriele Turchi]

The StEM test material comparing digital cameras and film, shot by members of the ASC, is available from SMPTE in both 2K and 4K files:

http://store.smpte.org/category_s/31.htm

Hello Mark ,
which one is it?

i can't see it ..

thanks!

g

 

[Randall Weatherington]

I'm not sure what they did for Good Night and Good Luck -- also shot in color, but desaturated and timed in B&W for release -- but my memory is that this also went on B&W stock. These movies shot on color negative mainly because of availability and wanting to get the dailies out fast, and (at that time) the LA labs couldn't turn B&W negative over very quickly.

I went to a union screening of "Good Night and Good Luck" where both the DP and George Clooney took questions after the screening.
The movie was shot on color stock because European release absolutely, positively required the film to be able to be released in Europe in color. Of course, it was only released in B&W, printed on 5302.

I have used several scanners over my career with interesting results.
One of the most unique jobs was wet gate scanning black & white work print of Frank Sinatra's 1957 ish TV shows.
Because of the dynamic range of B&W print, there were one or two songs that I scanned twice - once for shadows, and once for highlights.
That was done on an Oxberry 6400 with software that was already 11 years old at the time. Needless to say, I had to do some work-arounds to get the scan range to fit the dynamic range of the film.
One of the good features of that scanner was the fact that it used fixed pin movements. We could scan several elements of a shot, as long as the movement's pin was in the correct Mitchell position in relationship to the film, and combine them digitally without any chattering between elements. As originally equipped, the Oxberry 6400 was only a 2K scanner. Put a 4K camera on it, let's say a RED One, and it instantly becomes a 4K scanner.

I never used Spirit or Imagica scanners, but on both, I noticed that they were just drum scanners with no frame registration.

I did use a Northlight in combination with Arri recorders.

So how did all of these scanners compare? All scanners were able to output at least 10 bit. As long as the image was sharp, and the scans were within the dynamic range, there were very few problems getting the image to look good, and we could still see the original film grain.