Fuji`s new Eterna stock and RED comparsion

[MJ KERBER]

Someone asked about defining the word "natural."

I won't try to define it, but I will say that every single test that I've seen that compared high end electronic cinematography cameras to film (same scene, lighting, actress, etc., etc.), the film originated skin tones just look better. One could try to express the difference in words--more subtle gradation of tones, better rendering of the various hues that make up skin, bla bla--but it's just something everyone has to see for themselves. The above is most true when you're not aggressively pushing the look; when you're going for a 'natural' representation of what your eyes see.

Having said that, when Epic is released (or before) I think it's time to revisit such comparisons. If Epic delivers on Dynamic Range (which it looks like it will) AND it nails skin-tones like film can/does, then....

And viewing it on the internet isn't sufficient. You need a large screen.

Best,

 

[Antoine Fabi]

This film stock makes me wish to shoot film, really.

So beautifull.

Film is not limited to 10 bit, 12 bit or even 20 bit, it's infinite, analogue, and it shows, even once it's been scanned digitally.

Every time i look at film, it looks better than digital.

That said, film is not convenient and is very costly, so, sadly, it is disappearing.

I am curious to see how the HDRx™ will improve things on the EPIC.

Not only it will improve the DR, but it could also close the gap a little more because it should look smoother, richer.

Antoine

 

[NormLi]

Love these stocks and can't wait to shoot on it. Wow! Exciting times.

 

[Michel Hafner]

This film stock makes me wish to shoot film, really.
So beautifull.
Film is not limited to 10 bit, 12 bit or even 20 bit, it's infinite, analogue, and it shows, even once it's been scanned digitally.

Analogue does not have infinite resolution. Far from it. Noise level sets hard limits to the dynamic range of any analogue source. Film is no different with its built in noise levels.

 

[LeeArcher1]

Looks fantastic and everything, but i cant help thinking; that looks alot like RED...

The lack of noise and richness of colour only enhances the DIGITAL look.

Obviously FUJI wouldnt do it, but it would be great if we could see a comparison side by side in the future. Using the best "DIGITAL" cinematographers on EPIC, (R1-MX), ALEXA, etc and see the same shots on 35mm or 16mm.

Similar to what zacuto did on their shootout.

Hopefully in the new year we can see some tests.

 

[Stefan Antonescu]

IMHO it doesn't look better than R1 footage... Can't see any difference.
I don't hate film but it will be buried alive when EPIC ships.

The only thing I'm missing with digital is the "film motion blur".
And that's why I'm so excited about EPIC. That special "Magic Motion" looks so sexy and filmic.

 

[Lorenzo Straight]

Yeah, great test footage. The dynamic range is amazing. Thanks for sharing.

 

[Antoine Fabi]

Analogue does not have infinite resolution. Far from it. Noise level sets hard limits to the dynamic range of any analogue source. Film is no different with its built in noise levels.

Michel,

Well...sure it has limits...everything has ...but i mean it does gently and smoothly infinitely falls into vagueness. That's the nature of things. Same with analog sound recording. That why most big recording studios still manipulate most of the sound in the analog world before "freezing" it digitally.

For instance, shoot a 4K digital image and then resize it at 8K, it will look blocky.

Now scan a film at 8K, it will look smooth for sure.

Now, that said, i shoot digital
Repeatable results, flexible post work, convenient, more affordable, and more and more organic looking. The more info you record ( bit depth, sample rate etc...) the more you close the gap with analog, that's for sure.

I like RED philosophy of recording the highest possible resolution and other sensor info because that closes the gap between digital and analog.
To my eyes, RED has the best looking image in the digital acquisition.

Again, i have high hopes that RED new HDR will bring an unprecedent image smoothness in digital acquisition.
To me, it is as much important as the DR improvement.

That said...i love film

Antoine

 

[Michel Hafner]

Michel,

Well...sure it has limits...everything has ...but i mean it does gently and smoothly infinitely falls into vagueness. That's the nature of things. Same with analog sound recording. That why most big recording studios still manipulate most of the sound in the analog world before "freezing" it digitally.

?? What are you talking about?

For instance, shoot a 4K digital image and then resize it at 8K, it will look blocky.

No, it will not, unless you use a really bad interpolation algorithm like nearest neighbour. It will look pretty much the same. Just with 4 times the number of pixels.

Now scan a film at 8K, it will look smooth for sure.

Actually it will look quite rough with all the grain detail in your face. Detail that does not make it to release prints and on the screen.
There is no magic about film. It's a high resolution binary system and itself digital in nature, just instead of pixels you have grain that is developed or not, coupled with dies. Lots and lots of it. Lots of little bits to be on or off and create images with a dithering process. In the end you get limited real world detail and very real noise levels. But it behaves smoothly for sure. Do not try to zoom in much, though. It will quickly reveal its nature and limitations.

 

[D Fuller]

There is no magic about film. It's a high resolution binary system and itself digital in nature, just instead of pixels you have grain that is developed or not, coupled with dies. Lots and lots of it. Lots of little bits to be on or off and create images with a dithering process. In the end you get limited real world detail and very real noise levels. But it behaves smoothly for sure. Do not try to zoom in much, though. It will quickly reveal its nature and limitations.

Aaaah... no. It is not.

Film is not binary in nature at all, and not digital. You see, the size of the dots you refer to varies, and that is a key difference between film and a binary system. Yes, there are lots of little bits, but as they are developed, they grow in size as well as number. This creates an image in quite a different way than a matrix of on/off pixels that are always the same size. That is precisely why it behaves so smoothly.

You are correct that detail is limited, and that grain is limiting in much the same way as digital noise, but here, the randomness of the pattern of film grain makes it much less objectionable than most sensor noise, which has a distinct, geometric pattern. One of my favorite features of the MX sensor (and of the Nikon D3s) is that it has a much more random-seeming noise than the M sensor so that even when there is a bit of noise, I don't have the feeling of looking through a screen at the image.

And scanning 35mm film at 8K is a bit of overkill, but it will not look any rougher at an equal projection size than a 4K scan. The 8K scan scan will more accurately render the film's grain, but it won't make more of it. (It also won't resolve any more resolution from a 35mm film frame. The film frame has a resolution of a little less than 4K.)

 

[Stefan Auer]

Aaaah... no. It is not.

Film is not binary in nature at all, and not digital. You see, the size of the dots you refer to varies, and that is a key difference between film and a binary system. Yes, there are lots of little bits, but as they are developed, they grow in size as well as number. This creates an image in quite a different way than a matrix of on/off pixels that are always the same size. That is precisely why it behaves so smoothly.

You are correct that detail is limited, and that grain is limiting in much the same way as digital noise, but here, the randomness of the pattern of film grain makes it much less objectionable than most sensor noise, which has a distinct, geometric pattern. One of my favorite features of the MX sensor (and of the Nikon D3s) is that it has a much more random-seeming noise than the M sensor so that even when there is a bit of noise, I don't have the feeling of looking through a screen at the image.

And scanning 35mm film at 8K is a bit of overkill, but it will not look any rougher at an equal projection size than a 4K scan. The 8K scan scan will more accurately render the film's grain, but it won't make more of it. (It also won't resolve any more resolution from a 35mm film frame. The film frame has a resolution of a little less than 4K.)

at a certain level you can say that it is binary. because either there are enough photons to split the SilverBromide or not. There is no other case (as far as I know), so its binary at that level.

 

[D Fuller]

at a certain level you can say that it is binary. because either there are enough photons to split the SilverBromide or not. There is no other case (as far as I know), so its binary at that level.

You could, but that would require stretching the definition of binary to a really meaningless point. Let's see... "At it's absolute lowest limit of sensitivity and below, film is binary, but it ceases to be binary at any light level above that." How's that? Pretty absurd, I think.

You could use film as a binary system by recording binary data on it á la punch cards. (Oh, does that make paper binary?)

This is a silly discussion.

 

[D Fuller]

An interesting aspect of this test is our reaction to the dynamic range it presents. There is no detail at all in most of the fire shots, yet no one has commented on that. If I were shooting this scene on my Red, I would worry about that, thinking it was a failure of DR in the digital world. It's interesting to see that film would not hold detail the fire either, and that it's OK. Papamichael has made the choice to expose for the skin tones and let the firelight go as far over as it likes. I'd probably have underexposed the skin trying to hold more of the fire detail. Good lesson.

 

[Michel Hafner]

Aaaah... no. It is not.
Film is not binary in nature at all, and not digital. You see, the size of the dots you refer to varies, and that is a key difference between film and a binary system.

It's a matter of definition. Size can be modelled by clustering binary elements. And the 3D aspect using binary voxels in 3D space. The point is that film's smoothness is ultimately the result of a dithering process with binary elements at the lowest level. That gives film strengths but also a built in weaknesses. The resolution and noise limits become quickly apparent when dealing with blow up and insufficient light/exposure/sensitivity. The dithering shapes the MTF curve in specific ways. Film is great, but not magic or perfect. Traditional digital is "few" non binary elements (each capable of representing an average of "many" binary elements) and film is "many" binary elements together forming "fewer" non binary elements. Which prevails depends on what you are trying to represent and how many elements you can work with.

And scanning 35mm film at 8K is a bit of overkill, but it will not look any rougher at an equal projection size than a 4K scan. The 8K scan scan will more accurately render the film's grain, but it won't make more of it. (It also won't resolve any more resolution from a 35mm film frame. The film frame has a resolution of a little less than 4K.)

Correct. More K will show the grain and ultimately the binary nature of film more accurately, but is pointless for real world detail once some format specific limit is crossed. My point was that more Ks do not make things smoother, only more correctly rendered which means sharper as far as grain detail is concerned. Rough is probably a misleading term for what I meant.