Jump to content

35mm, 70mm Film Vs Digital Cinema (Red Weapon 8K S35)


Recommended Posts

Dear DPs,

 

If you see this young generation of filmmakers (Quentin Tarantino, Christopher Nolan, J.J Abrams), they keep shooting their movies in film like Martin Scorsese, Steven Spielberg, …

 

Does it mean that today, we can't get the feeling, the grain, and the warmth of film in digital ?

The response is maybe in the question, right ?

 

But I would like to know if there is a DP who tried to get this feeling of film (35mm (ex: Eastman 100T 5247),70mm (ex: 65 mm Eastman 50T 5250) in digital, especially with the Red Weapon 8K S35 ? And of course with the appropriate lens.

 

Thanks

 

Jean-Marc

Link to comment
Share on other sites

  • Premium Member

Trouble is that what constitutes a "film look" is somewhat personal. There have been digital movies that have applied some grain simulation in post to make the image look more like film, for example, "The Man from U.N.C.L.E." and "Logan" were shot by John Mathieson using classic Panavision anamorphic lenses on the Alexa and I believe some film grain emulation was applied, as well as a LUT that simulated Kodak color negative stock.

 

Some of the faux 16mm documentaries on the parody series "Documentary Now" were shot on a Red camera with the sensor cropped to 2K in order to use 16mm lenses and then a film grain was added in post using Live Grain:

http://www.creativeplanetnetwork.com/news/shoot/documentary-now/611034

 

HBO's "Vinyl" used Sony F55's and the Live Grain process to give it a 16mm look.

Link to comment
Share on other sites

The way colors saturate (depending on degree of under/overexposure) on film, which looks very natural to eye, hasn't been reproduced on video so far, and it's an extremely difficult engineering task.

Then, to have equivalent overexposure latitude and highlight handling on video, you have to severely underexpose it and apply a very smooth rolloff in post - and have your shadows hit the noise floor and look like crap.

So no, you can't fake it, only get somewhat close with G&E virtuosity, filtration and careful production design.

 

Concerning Red's 8K camera. Resolution has nothing to do with image quality as long as it looks sharp on projection. For example, last year I chose Kodak 7203/07 over Alexa XT based on pure image quality, even though resolution of a 2K scan was lower. K's make no difference, it's 1) colorimetery 2) latitude, which are responsible for a good image.

Edited by Michael Rodin
Link to comment
Share on other sites

The way colors saturate (depending on degree of under/overexposure) on film, which looks very natural to eye, hasn't been reproduced on video so far, and it's an extremely difficult engineering task.

Then, to have equivalent overexposure latitude and highlight handling on video, you have to severely underexpose it and apply a very smooth rolloff in post - and have your shadows hit the noise floor and look like crap.

So no, you can't fake it, only get somewhat close with G&E virtuosity, filtration and careful production design.

 

Concerning Red's 8K camera. Resolution has nothing to do with image quality as long as it looks sharp on projection. For example, last year I chose Kodak 7203/07 over Alexa XT based on pure image quality, even though resolution of a 2K scan was lower. K's make no difference, it's 1) colorimetery 2) latitude, which are responsible for a good image.

Dear Mr Rodin, I agree that it's quite hard to get the quality of film, just with the hardware and capacty of the Digital camera. I'm not a fan of fake look on post-prod and a long process, but I took note carefully of your advices and technical aspect, to get the best compromise like anyone else as you know. Thank you very much.

Best Regards,

Jean-Marc

Link to comment
Share on other sites

Trouble is that what constitutes a "film look" is somewhat personal. There have been digital movies that have applied some grain simulation in post to make the image look more like film, for example, "The Man from U.N.C.L.E." and "Logan" were shot by John Mathieson using classic Panavision anamorphic lenses on the Alexa and I believe some film grain emulation was applied, as well as a LUT that simulated Kodak color negative stock.

 

Some of the faux 16mm documentaries on the parody series "Documentary Now" were shot on a Red camera with the sensor cropped to 2K in order to use 16mm lenses and then a film grain was added in post using Live Grain:

http://www.creativeplanetnetwork.com/news/shoot/documentary-now/611034

 

HBO's "Vinyl" used Sony F55's and the Live Grain process to give it a 16mm look.

 

Dear Mr Mullen, thank you very much for your information, the link, and the sharing of your experience. I appreciate. I saw "The Man from U.N.C.L.E". I appreciated the photography, and the feeling of film.

Best Regards,

Jean-Marc

Link to comment
Share on other sites

  • Premium Member

I would like to know if there is a DP who tried to get this feeling of film (35mm (ex: Eastman 100T 5247),70mm (ex: 65 mm Eastman 50T 5250) in digital.

 

Somehow, I have a feeling that the list of DPs who have not tried to emulate the feeling of celluloid film with a digital camera is quite short...

 

I think once you start using film emulation, color profiling, and grain management in post, the discernible differences can become very small. Whether you think that those little film quirks still make a qualitative difference in your overall enjoyment of the image is up for debate. I think people should use whatever format they feel comfortable with.

 

As for how closely a digital camera with post processing can be made to emulate film, please check out our discussion of a recent test by Steve Yedlin, ASC: http://www.cinematography.com/index.php?showtopic=69957.

Link to comment
Share on other sites

Yedlin's test doesn't show mid-to-low-contrast portraits with highly overexposed regions, like very hot wide kickers, rim lights, etc, as it's where Alexa will scream video. And what can been seen in the test already shows the usual weird oversaturation of highlights corrected by what seems luma-saturation curves and soft clip - which has made them flatter and less textured than on film.

 

How high-key portraits look can be a deal breaker. A recent example - 2015 Russian TV series Quiet Flows the Don shot by Michael Suslov, RGC on Kodak V3 stocks in S16 - is half made of excellent high-key portraiture with broad hot kickers. Would have looked burnt and lit had it been shot on video - looks rich on 7203.

 

Then, how film scans were processed remains a question. What can be said for sure, no special highlight recovery was done. Kodak V3-through-Arriscan has significantly more overexposure latitude than used in the test from what I've seen - and I tend to shoot a dense negative (1/2-1 stop over).

 

Speaking about lenses - Mr Gourmaud touched the subject in the beginning. While they don't influence colorimetery and contrast properties as much as stock or video camera, they do make a difference.

 

Shooting HD on F900 and then F23, I found fast spherical Elite brand (ex-USSR Ekran bureau) T1.5 primes to give the most natural, I'd say (as the word cinematic is overused :)), image. Compared to Digiprimes, they're lower contrast with more veiling glare but very little blooming, which allows you to stop down more for the same exposure and see more and smoother highlight detail. They were also soft wide open but I liked the softness of a Classic Soft 1/2 at T2 (their sharpest stops were 2.8-4) more than the softness of spherical abberations at T1.5. Slight blooming by the way can be a good thing with video - can help conceal clipping. For that reason I never removed "white" diffusion (Schneider White Frost) from an F900, the more "clippy" camera. So the old(er) glass is more about contrast than softness I'd say. Although the colorimetery of older-techology lenses is generally worse; old coatings introduce color shifts which have nothing special about them.

Edited by Michael Rodin
Link to comment
Share on other sites

  • Premium Member

I've rarely found the Alexa to scream video in most situations and I find the overexposure range to be similar to film. Even Kodak claims about a 14-to-15 stop range for film which is about the range of the Alexa, 14.5-stops. Even the way it clips to white is similar to film.

Link to comment
Share on other sites

Somehow, I have a feeling that the list of DPs who have not tried to emulate the feeling of celluloid film with a digital camera is quite short...

 

I think once you start using film emulation, color profiling, and grain management in post, the discernible differences can become very small. Whether you think that those little film quirks still make a qualitative difference in your overall enjoyment of the image is up for debate. I think people should use whatever format they feel comfortable with.

 

As for how closely a digital camera with post processing can be made to emulate film, please check out our discussion of a recent test by Steve Yedlin, ASC: http://www.cinematography.com/index.php?showtopic=69957.

 

Dear Mr Murashige,

 

I agree with you about the short list of DPs who didn't try, but I didn't get a lot of feedback about the topic, so ...

In fact, my initial question is more focused about emulation in hardware (Camera capacities and settings) instead of software in post-prod, even if we can't totally avoid this process as you know. I take in consideration your point of view, the link (great information !) and I thank you very much to light my curiosity and knowledge.

Best Regards,

Jean-Marc

Link to comment
Share on other sites

Yedlin's test doesn't show mid-to-low-contrast portraits with highly overexposed regions, like very hot wide kickers, rim lights, etc, as it's where Alexa will scream video. And what can been seen in the test already shows the usual weird oversaturation of highlights corrected by what seems luma-saturation curves and soft clip - which has made them flatter and less textured than on film.

 

How high-key portraits look can be a deal breaker. A recent example - 2015 Russian TV series Quiet Flows the Don shot by Michael Suslov, RGC on Kodak V3 stocks in S16 - is half made of excellent high-key portraiture with broad hot kickers. Would have looked burnt and lit had it been shot on video - looks rich on 7203.

 

Then, how film scans were processed remains a question. What can be said for sure, no special highlight recovery was done. Kodak V3-through-Arriscan has significantly more overexposure latitude than used in the test from what I've seen - and I tend to shoot a dense negative (1/2-1 stop over).

 

Speaking about lenses - Mr Gourmaud touched the subject in the beginning. While they don't influence colorimetery and contrast properties as much as stock or video camera, they do make a difference.

 

Shooting HD on F900 and then F23, I found fast spherical Elite brand (ex-USSR Ekran bureau) T1.5 primes to give the most natural, I'd say (as the word cinematic is overused :)), image. Compared to Digiprimes, they're lower contrast with more veiling glare but very little blooming, which allows you to stop down more for the same exposure and see more and smoother highlight detail. They were also soft wide open but I liked the softness of a Classic Soft 1/2 at T2 (their sharpest stops were 2.8-4) more than the softness of spherical abberations at T1.5. Slight blooming by the way can be a good thing with video - can help conceal clipping. For that reason I never removed "white" diffusion (Schneider White Frost) from an F900, the more "clippy" camera. So the old(er) glass is more about contrast than softness I'd say. Although the colorimetery of older-techology lenses is generally worse; old coatings introduce color shifts which have nothing special about them.

 

Thank you very much Mr Rodin, for all these information.

Best regards,

Jean-Marc

Link to comment
Share on other sites

I've rarely found the Alexa to scream video in most situations and I find the overexposure range to be similar to film. Even Kodak claims about a 14-to-15 stop range for film which is about the range of the Alexa, 14.5-stops. Even the way it clips to white is similar to film.

 

I took note of this Alexa aspect. Thank you Mr Mullen,

Best regards,

Jean-Marc

Link to comment
Share on other sites

I've rarely found the Alexa to scream video in most situations and I find the overexposure range to be similar to film. Even Kodak claims about a 14-to-15 stop range for film which is about the range of the Alexa, 14.5-stops. Even the way it clips to white is similar to film.

It's not as bad as F900 or Red One, of course. Where highlights can look noticeably unnatural is on skin given enough overexposure. And if skin meters at +2...3 it's usually where a kicker ("modelling" light I'd say in Russian, seems not really equivalent terms) hits - often at angle that makes skin more specular to camera, making it all even worse... And the giveaway is, IMO, mostly not clipping (which, again, happens not so early on Alexa) but highlight desaturation which looks artificial - because it is artificial: in video saturation rises with exposure, it's clever processing which tries to compensate.

It's a subjective thing anyway, so I don't blame video much for slightly more blown practicals and windows, but when it comes to portraits, viewer will notice it, consciously or not.

 

As to DR stops - well, we both know these figures can't really be compared between manufacturers as there's no standart for DR measurement.

With film, there's density range and gamma which can be precisely measured. How many stops it gets you depends on how much of its curve's nonlinear part you can scan and use.

With digital, you have properly measured signal to noise and you decide how much noise you can tolerate to get your DR.

So the only real comparison is a test which's a very individual thing - too much variables, esp. with film (exposure, scanning, LUT/matrix, etc).

Edited by Michael Rodin
Link to comment
Share on other sites

  • Premium Member

Doesn't film emulation all come down to software, at a certain point though? Silicon sensors all have fixed imaging characteristics that differ fundamentally from logarithmic film capture simply due to physics.

 

If you push any digital sensor into clipping, the RGB channels clip hard in linear light when the photosite saturates. What makes the digitally captured image look like film capture is how the camera's color science encodes the gamma curve, debayers the color data, and controls the highlight roll-off. And also any subsequent color grading and VFX to further emulate the characteristics of film capture.

 

In order for that to happen, the full dynamic range of the scene has to be captured without clipping or excess noise in the shadows. I think Yedlin's demonstration was to show exactly this overall approach to image capture. He's on record as viewing both film and digital capture formats as just a means to record the maximum detail from a scene, from which he creates a look in post. Which is just as valid a working method as Wally Pfister's film purist approach, or Conrad Hall's happy accidents, etc.

 

Personally, I really couldn't tell which shot was Alexa and which was 35mm for most of the shots in Yedlin's side-by-side demo. And I looked pretty closely. So as far as I'm concerned, they can be made to look identical (at least for the web) with enough work in post. I still prefer shooting on film, but I like most of the little happy accidents that come with it. So it just ends up being less work for me. To each their own though.

Link to comment
Share on other sites

Doesn't film emulation all come down to software, at a certain point though? Silicon sensors all have fixed imaging characteristics that differ fundamentally from logarithmic film capture simply due to physics.

 

If you push any digital sensor into clipping, the RGB channels clip hard in linear light when the photosite saturates. What makes the digitally captured image look like film capture is how the camera's color science encodes the gamma curve, debayers the color data, and controls the highlight roll-off. And also any subsequent color grading and VFX to further emulate the characteristics of film capture.

 

In order for that to happen, the full dynamic range of the scene has to be captured without clipping or excess noise in the shadows. I think Yedlin's demonstration was to show exactly this overall approach to image capture. He's on record as viewing both film and digital capture formats as just a means to record the maximum detail from a scene, from which he creates a look in post. Which is just as valid a working method as Wally Pfister's film purist approach, or Conrad Hall's happy accidents, etc.

 

Personally, I really couldn't tell which shot was Alexa and which was 35mm for most of the shots in Yedlin's side-by-side demo. And I looked pretty closely. So as far as I'm concerned, they can be made to look identical (at least for the web) with enough work in post. I still prefer shooting on film, but I like most of the little happy accidents that come with it. So it just ends up being less work for me. To each their own though.

 

Thank you Mr Murashige for this further information. I had a look on Yedlin's demonstration. Very impressive I agree.

I understand better now, why some DPs still love shooting on celluloid film.

I have all elements now, to get the best compromise on my projects, and above all the budget of course !

Best,

Jean-Marc

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...