Comparing Film to Red , Tests and Example images

[Les Dittert]

I am starting this thread to place test results in.

I recently did a shoot with some Kodak 5203 ( ISO 50 daylight Vision III ) the best filmstock out there right now.
The purpose of the test shoot was not to compare the Red with film, but I took the opportunity to set up my Scarlet next too the Arri to have some fun.
The Arri was using Zeiss super speed glass ( 35mm and 50mm ) and the Red was using Sigma 35mm 1.4 A series and am old Zeiss 50mm 1.4 .
The apertures were kept away from being too wide open, as to maximize sharpness. ( usually f 5.6 or 8 , sometimes 11 )

The Kodak negative was scanned at 4K and 6K. I am using the 4K scan here.

These frames were to stress the dynamic range. Unfortunately the framing is a bit different between the film and digital, but it is still useful to see what these look like.


HINT : you may want to right click the image and open in new window to see it full sized .

All Images posted by me here in this thread are COPYRIGHT 2013 and may not be posted on any other site, please.

First: The Film frame, sized to 1280 :

Now the RED frame, sized from 4K to 1280 :

Now a gamma corrected series to let you see the highlites and shadows better ( these are 1:1 pixels from the 4K ) :

First the Film ( Three images , different color corrections )

 

[Les Dittert]

Now this post has the RED cropped images ( 1:1 from 4K no resize , same color corrections series as above )

I will also repost the other frames I did from the other thread soon.

 

[Les Dittert]

These are from the same day, but for looking at details.

First the Film full frame :

Now the RED frame , resized from 4K :

And the Crops from the frames , first the Film frame :

And now the RED frame, cropped from 4K :

Again, Hint : right click and open in new window to see the image bigger than here in the thread.

 

[Bruce Allen]

Oh - that's awesome - Thank you Les.

I'm having the usual problem: when you color correct the first two images and punch them up... the film still has way better skin tones than the RED still.

Is it just me?

I'm presuming that your next step would be to nicely color balance them and make them match, so if you have some setting that makes the RED skin tones match the film ones, please share. Thanks!

Bruce Allen
www.boacinema.com

 

[Karim D. Ghantous]

Oh - that's awesome - Thank you Les.

Thanks from me, too!

I'm having the usual problem: when you color correct the first two images and punch them up... the film still has way better skin tones than the RED still.

Is it just me?

No. The RED shots of the boy show hideous green fluffy chrominance noise. Perhaps if that can be easily eliminated, you're as close to film's skin rendering as you're going to get. The film shots show lovely tonality in the boy's face. It's a shame you posted JPEGs 'cos they make the shadow detail in the RED image look really shit. OTOH, the RED shows fine detail, such as that in the baseball cap.

[Edit: have the films scans been sharpened to industry standard (whatever that might be)?]

As for the shots of the woman, film is suspiciously lacking in sharpness. Hmm. I don't know what to make of that one.

I would love to see a comparison between the normal Scarlet and your custom one. Film loses out as far as fine detail is concerned, however, you aren't going to get moire with film.

And damn doesn't film look beautiful. Digital cameras are great, even for film purists, because they make more obvious what film really looks like. If that makes any sense. I'm watching the Harry Potter movies with a friend, one a week, and it's so damn obvious that they were shot on film. I don't know what words would describe it, but 'depth' and 'richness' come to mind.

I am biased towards film. Having said that, I took over 350 frames with my iPhone last night at a friend's 40th. Being biased towards one thing doesn't make the other thing bad or wrong by default. Just preferences, that's all! I see myself as a future RED customer for sure. But it makes no sense right now. Sorry, just thinking aloud...

 

[Vithya Sharma]

Woow..RED Really awesome thanz for post

 

[Marc Wielage]

The contrast is not the same in the film scan, so it's not a fair test. If you color matched the clips and was very careful on scanning levels, you'd get different results. Note that scanners have to be very precisely focused, too; I know of cases where negatives were scanned with soft focus because of different emulsions being used, and the scanner operator didn't check it, wreaking much havoc. (In fact, I just remembered one where this happened with a film that wound up making $300M... but I won't name it or the facility.) And different scanners yield different results; Northlight, Arriscanners, DFT scanners, and Imagica scanners all have a different look, in my opinion.

5203 was not and is not a 35mm film stock used very often; I think 5213 or 5219 would be more realistic in terms of what's actually used on contemporary big-budget feature films.

The ASC has done extensive tests between color negative and Red and Alexa and so on, and I think the end result was, the differences are not gigantic either way, and each has its pros and cons. Speaking as a former Technicolor guy, the biggest difference I felt I saw in film vs. 2K digital were in wide shots, not in close-ups. Show me (say) a football stadium filled with 50,000 people, and let's look at them in film vs. looking at them in digital. That's a real torture test.

In truth, I think the lighting, the lenses, and the skill of the cinematographer are the biggest determining factors in picture quality -- not choice of camera. I think in 2013, arguing about production formats is kind of like two world-class writers arguing about which word processor and monitor to use for novels.

 

[M Most]

Show me (say) a football stadium filled with 50,000 people, and let's look at them in film vs. looking at them in digital. That's a real torture test.

Except that you'd get vastly different results depending on the film format - 35mm, 35mm anamorphic, 65mm, etc. These things have never been "film vs. digital." They've been a specific film format vs. a specific digital format. And that's part of why the only sensible comparisons are with the formats that one can actually afford and use, put through whatever process is optimal for each - a test that is rarely done in a general purpose "shoot out" scenario, but often done in specific tests for specific projects.

 

[Björn Benckert]

These are really nice. But as stated above, they do not say much as they are graded. But it would be nice to play with the actual frames. How was the film digitised?

 

[Les Dittert]

Marc,
I'm not sure what you mean about the contrast. There is a slight kink in curve in the top whites of the female that I did not notice before, but other than that it's pretty darned close.
I did notice the skin tones came out without much effort on the film version of the boy. That shot is a tough one, as the skin tones are way at the bottom of the histogram. I'm sure I could get better skin tones on him, but I would probably have to key it. Curves did not get there, I tried.
The 3 different versions of the boy were to see into the brights and lows. They were CCed then saved as 8 bits, not CCed from 8bits which would never work.

I don't care about 5203 not being the most popular film stock. The goal was to use the best image quality film stock.
The film was scanned on two different scanners with special attention on focusing the the grain on the actual film used. Minimal 'aperture correction' ( term coined by Kodak, but is sharpening ). I do know a thing or two about motion picture scanners, since 1986 actually

I have wide shots as well, the sharpness difference is still dramatic. But my Scarlet E is sharper than normal...
I'd like to repeat the test. Next time more attention to framing , and including a couple of digital cameras. Maybe a 65mm film camera.
But I will definitely use low speed vision III. Why handicap film? I want to see the best results from each medium.

 

[Les Dittert]

Except that you'd get vastly different results depending on the film format - 35mm, 35mm anamorphic, 65mm, etc. These things have never been "film vs. digital." They've been a specific film format vs. a specific digital format. And that's part of why the only sensible comparisons are with the formats that one can actually afford and use, put through whatever process is optimal for each - a test that is rarely done in a general purpose "shoot out" scenario, but often done in specific tests for specific projects.

I want to do 65mm next time.
I got tired of all the speculative chit chat about film being so great, and decided to actually post images on the subject. This was not a perfect test, but there it is.
Next time, more effort on the scientific method for the test.

 

[Robert Hart]

Les. It would be interesting to see what you get with say a graded blend of three or four frames of a static subject for film and for RED origination. Motion film gains in "apparent" sharpness, a subjective thing our dumb eyes see when it is rolling.

 

[Les Dittert]

Les. It would be interesting to see what you get with say a graded blend of three or four frames of a static subject for film and for RED origination. Motion film gains in "apparent" sharpness, a subjective thing our dumb eyes see when it is rolling.

Robert,
That's true. Our eyes-brain are essentially doing something like a degraining process, in a perceptual sense.
I could simulate that by running a good degrainer to see how clean the film image comes through the other side.
A temporal degrainer does the same thing our brain does and sifts out the real detail from the noise.

 

[mikeburton]

The soft rolloff in the film file and the many tests I've done in this same regard against epic, alexa, etc along with the natural color rendition of the film lends itself to my personal taste between the two for narrative work. That said, we always tend to judge these images in still frames as if we are still photographers, and I think that's not quite a fair test. In most cases, stills will always impress due to the clarity of the picture as your eye has time to scan the frame and see all the details but in motion is what these images are intended to be viewed and often film feels more natural to my eye when you press play.
I'd love to see 35mm anamorphic versus RED Epic shooting 5K anamorphic. I have an Arri 435es 4 perf that i'd be willing to donate for that test if you want to rent some lenses Anamorphic would yield the most resolution and best 35mm film has to offer which would be a worthwhile test IMO.

 

[Marc Wielage]

Except that you'd get vastly different results depending on the film format - 35mm, 35mm anamorphic, 65mm, etc. These things have never been "film vs. digital."

Also very true. I have no argument there, but for purposes of my original statement, let's assume I mean "35mm 4-perf." I would hope you'd agree that wide shots are tough, scanning is variable, and the end results depend more on the skill of the DP, the lighting, and the lenses than the format itself.

I remember a project some years back where (for various reasons) the filmmaker opted to do digital pickups for a film feature, and I don't think anybody seeing the final was aware of the cut-ins. They looked fine, both in the theater and in home video. I would be reluctant to use them side-by-side, but I think a whole sequence isn't going to stick out, even if you shot it on Red One, Red Epic, Alexa, and Film. It can all look good if the DP knows what they're doing.

The film was scanned on two different scanners with special attention on focusing the the grain on the actual film used. Minimal 'aperture correction' (term coined by Kodak, but is sharpening ). I do know a thing or two about motion picture scanners, since 1986 actually.

Funny, the first motion picture scanner I know of was the hand-built unit built by ILM in San Rafael in 1988 for Back to the Future II (with Kodak right behind them in 1990 with Cineon). Tell me about what you were doing in 1986.

I'd also be curious to know the specific brand names and models of the scanners you used for this test. Speaking only as a colorist, the shots don't match to me, but bear in mind I'm looking at this board on an sRGB monitor slightly tweaked for Rec709 colorspace and 2.2 gamma. No doubt on a fully-calibrated display, the differences would be more meaningful -- but my experience is that they would be worse, not better. It is possible to match digital and 35mm so they're almost identical in terms of contrast and color, depending on color. Have you seen the ASC StEM test?

http://www.dcimovies.com/StEM/

 

[Les Dittert]

Anamorphic lens' tend to be softer than spherical. I'm dubious whether you can even get 4K out of one.
Spherical lens' are the best for a film vs digital shootout because you don't want to even have the glass resolution even be a question.

I agree that stills are an issue. But if one still is much better looking than another, it will follow that it still wins the test.
If the grainy slightly softer film does look better to a regular person ( not a dyed in the wool grain lover geezer on some blog group ) then we can always degrade the digital with grain and blur to mimic film.
Anyhow, this is really just a fun test I did, as most normal people won't be shooting film anymore anyhow.
I didn't think it would happen this fast, but it's great!

The soft rolloff in the film file and the many tests I've done in this same regard against epic, alexa, etc along with the natural color rendition of the film lends itself to my personal taste between the two for narrative work. That said, we always tend to judge these images in still frames as if we are still photographers, and I think that's not quite a fair test. In most cases, stills will always impress due to the clarity of the picture as your eye has time to scan the frame and see all the details but in motion is what these images are intended to be viewed and often film feels more natural to my eye when you press play.
I'd love to see 35mm anamorphic versus RED Epic shooting 5K anamorphic. I have an Arri 435es 4 perf that i'd be willing to donate for that test if you want to rent some lenses Anamorphic would yield the most resolution and best 35mm film has to offer which would be a worthwhile test IMO.

 

[Les Dittert]

Funny, the first motion picture scanner I know of was the hand-built unit built by ILM in San Rafael in 1988 for Back to the Future II (with Kodak right behind them in 1990 with Cineon). Tell me about what you were doing in 1986.

I started working at ILM in 1986, to take over the Pixar laser film recorder/scanner when Pixar co. was sold by George. That's where I was .
We also scanned before the Cineon scanner was done. If memory serves, it was an Ikonix line array camera that Scott S. bolted to an optical table behind the laser machine. It took a god awful time to scan each frame. We were using 256meg big removal diskpacks, a big pain in the ass. Dennis M. would not let is drop the resolution below 3456 across, and we used 12bits/color.
The 3456 resolution was what the Pixar Image computer liked.
And I was the digital opticals supervisor on Back to The Future 2, doing the first ever fully digital composite.
Before that ILM was filming out mattes on hicon and using the optical printer to comp. the shots.

Funny right ?

 

[Casper Warnich]

I started working at ILM in 1986, to take over the Pixar laser film recorder/scanner when Pixar co. was sold by George. That's where I was .
We also scanned before the Cineon scanner was done. If memory serves, it was an Ikonix line array camera that Scott S. bolted to an optical table behind the laser machine. It took a god awful time to scan each frame. We were using 256meg big removal diskpacks, a big pain in the ass. Dennis M. would not let is drop the resolution below 3456 across, and we used 12bits/color.
The 3456 resolution was what the Pixar Image computer liked.
And I was the digital opticals supervisor on Back to The Future 2, doing the first ever fully digital composite.
Before that ILM was filming out mattes on hicon and using the optical printer to comp. the shots.

Funny right ?

Snap! :001_cool:

 

[Phil Holland]

Funny right ?

http://www.imdb.com/name/nm0228545/awards

"For their development work with area array CCD (Charge Coupled Device) film input scanning systems."

One of these other very intelligent individuals was my boss, manager, and I'd say mentor for a good chunk of my life.

A bit challenging jumping into the industry and learning how to work with that bleeding edge tech, but I loved all the shiny things

 

[Mike P.]

Hey Les thanks for your comparison frames. Like others have mentioned, RED, in my mind, always looks better when it's framegrabs/stills. You compare a RED frame to a film or Alexa frame and 4 out of 5 times the RED *still image* will look objectively better than the other two... However, put that shot in motion and it tends to look more digital than the other two (partly due to the high resolution which, ironically, helps make the stills so much better.) Motion mount *might* help, but something tells me its more than just rolling shutter vs mechanical shutter (after all, Alexa doesn't have to have a mechanical/global shutter to look like it does in motion.)

I find the second set of stills (girl standing in front of grass/trees) to be very interesting. From a latitude/dynamic range perspective it it seems they've captured (and rolled off) very similarly, but looking at her eyes in the two different shots shows some pretty big differences in colour tonality -- on film her eyes look way more green, whereas on RED they look icy blue. Is that because the frame is just different enough? Or maybe because the lenses are just different enough?

Another thing I'm noticing is that, despite the increased latitude and highlight handling of film in the first shot, the RED still manages to look essentially the same from a colour/latitude point of view, but again, the motion might make one stand out over the other. Definitely matchable, though.

Any chance we can see these in motion?