Is the future FF35?

[M Most]

Sure it exists. I shoot it all the time.When you watch these shots moving, the digital stuff is far, far more clean, stable, detailed and clear, even against the legendary 65mm, shot by a master.

I misunderstood and thought you were referring to the future Red releases, since this is, after all, Red User.

If you feel the way you do, then there's no reason to ask for mine or anyone else's opinion. "Better" is rarely an absolute when referring to images. It is a somewhat subjective opinion, based on one's own perception and one's own personal priorities in terms of the characteristics that catch one's eye and one's storytelling priorities. If you like it better, it's better. My opinion is irrelevant.

 

[Tom Lowe]

I misunderstood and thought you were referring to the future Red releases, since this is, after all, Red User.

If you feel the way you do, then there's no reason to ask for mine or anyone else's opinion.

I don't remember asking for your opinion. Did I miss something? We were talking about the advantages of full-frame 35mm shooting vs S35mm, and got sidetracked into a debate about digital vs film. I don't think there is any question whatsoever whether FF35 produces better image quality than S35.

My guess is that after you account for Graeme's compression, the Epic FF35 6K will probably be pretty close in IQ to FF35 5.6K Canon CR2 RAW. We'll see.

 

[Daniel Browning]

Mike, I think you're right in the general case: post processing latitude is generally very similar between film and raw for the common cases (TV, drama, etc.). However, there are some specific cases where digital creates new options that are impossible with film, and Tom just happened to use one of them as his example.

Tom is correct that there are many things raw can do that film can't, though the reverse is also true. The superiority of digital is the strongest in astrophotography. Not just because of the increased post processing options, but also because of the performance characteristics, such as the superior QE along most (all?) of the EM and much lower noise level.

Chiefly, raw allows "light frame stacking". With film, getting a good astrophotograph meant trying over and over again all night long to get a single one-hour exposure that wasn't ruined by numerous satellites flying through the frame, atmospheric turbulence, tracking mount imperfection, or dozens of other factors. On digital, you simply take dozens of much shorter frames, then let the processing software automatically throw out the ruined frames and stack the rest into one super picture.

All you did was an exposure change.

I'm pretty certain that the image Tom posted is the result of far more complex processing, such as stacking multiple raw files.

But you were implying that somehow a digitally shot RAW file can yield image flexibility that film cannot, and that is just not the case.

There are some ways that it's true. One of the major advantages of digital is the ability to build a custom tone curve for the image after the fact. The film tone curve is built right into the medium, and you can't change it from shot to shot so that it's just what you need. But as you say, it's gneerally flexible enough that you can achieve plenty of stuff, especially if you err on the side of overexposure.

The linear-vs-nonlinear has lots of other implications as well. One is shifts in color. In digital it's possible to push a shot by 8 stops and get the exact same color as if it was exposed correctly. (More noise, though, of course, which may or may not be visible depending on the camera and display size.) Pushing 8 stops on film (chemically or in the DI) would result in major color issues.

Speaking of color, my understanding is that the separation of color filter arrays in Bayer sensors is a significant advantage for flexibility in control over color. If we shot with 3-strip technicolor, film would have the same advantage, but normal film scans have the color layers mixed together, which doesn't allow certain techniques for changing the color that are possible with digital.

My guess is reciprocity failure. Long exposures can make film response inaccurate/weird etc. Perhaps digital sensors are superior for long exposures.

Reciprocity failure is definitely an issue, but even if it did not exist, film would still be inferior due to SNR and stacking. The only advantage that film has in astrophotography is reduced battery usage for star trails, and extra-large film sizes for wide fields (medium format and large format).

Again, though, I want to point out that there are other factors beyond DR.

Stephen mentioned the color palette of film. With raw data and the right software, you can get very close to simulating the color of film, but it's not possible to get every color exactly like film. (If you match film for a certain red/green, then blue/violet will be different, and vice versa.) Of course, that's only an advantage if you're purposefully trying for film-like color, which is pleasing, but highly inaccurate. 3-chip cameras have even worse color accuracy, but CFA raw cameras are capable of higher color accuracy (more like our own human vision) than film

Image capture is image capture, regardless of the medium used to capture it. If you can capture a wide enough range of information, you can then use digital processing techniques to manipulate it. Modern digital cameras capture a very wide range of values. Modern film emulsions capture even more. Once you scan that negative into a digital format, the same processing techniques can be used.

I agree with this for the common case. The times when digital has far more processing techniques are the situations that most redusers would probably consider outliers or too extreme. (Even though they are the very things that Tom and I spend most of our time shooting.)

Software OLPF is bollocks

My new favorite quote. I agree completely.

 

[M Most]

Mike, I think you're right in the general case: post processing latitude is generally very similar between film and raw for the common cases (TV, drama, etc.). However, there are some specific cases where digital creates new options that are impossible with film, and Tom just happened to use one of them as his example.

This seems to have started out as a conversation about time lapse, and then turned into something about astrophotography and advanced multiple image processing. The original question that I responded to was:

is Stephen's S35 timelapse going to be as stable, flicker-free, clear and as full of detail as my FF35 5.6K RAW? Putting DR aside, will his S35 timelapse match the image quality of my FF35 5.6K RAW?

...which, BTW Tom, was when you asked for an opinion. So I gave one.

CFA raw cameras are capable of higher color accuracy (more like our own human vision) than film..

In theory. However, the filters and microlenses used have crosstalk, and a properly designed color matrix is required to compensate for this. Not to mention that there are certain colors in nature that even the best designed Bayer filter and associated color matrix won't properly reproduce. Nothing is perfect.

I agree with this for the common case. The times when digital has far more processing techniques are the situations that most redusers would probably consider outliers or too extreme. (Even though they are the very things that Tom and I spend most of our time shooting.)

I think it's sometimes useful to consider the point of view that a poster is coming from in evaluating the statements they make. My background is primarily the entertainment industry - film and television. Most of what I say here and elsewhere is predicated on that experience, just as much of what you say is predicated on image analysis, post processing techniques, and, in general, a number of things that are generally a bit too time consuming to be in common use in the rather accelerated - both in production and in post - world of entertainment imaging. That doesn't make either view any less valid, because each one's view is based on their particular point of reference.

 

[Tom Lowe]

This thread is now wildly off track.

I thought the discussion was about FF35 vs S35 in general. Maybe we could cool it with these tangential posts.

 

[David Rasberry]

Software OLPF is bollocks, unfortunately. You can, to an extent, deal with chroma moire in software, but luma moire is "burned in" as the point of sampling and if it were easy to seamlessly remove afterwards, we'd all be doing it. Fact is, there's no way to remove it. There is some software that can do a ok guess - Hassy's raw software tries, but when it fails, and it does, it looks ugly. And there's no way it would stand up to motion.

In the end, the "correct" engineering and aesthetical approach is to optically ensure that the input signal to the sensor is sensibly band limited, and that there's enough photosites to get a reasonable degree of oversampling.

Graeme

Thanks Graeme

Makes perfect sense.

 

[Stephen Williams]

This thread is now wildly off track.

I thought the discussion was about FF35 vs S35 in general. Maybe we could cool it with these tangential posts.

Hi Tom,

In short the future may not be FF35 as you suggested.

Best,

Stephen

 

[Tom Lowe]

Hi Tom,

In short the future may not be FF35 as you suggested.

Best,

Stephen

Hi Stephen,

In short, it might be.

Let's stay tuned and find out. :wink:

Best,

Tom

 

[Oliver Skibbe]

This link may have been posted before, but I think it is quite interesting:

Canon product manager Tim Smith talking about 5D2 video and RED

 

[Daniel Browning]

I would be happy to do the experiment myself and post the results (including raw files)...

OK, here it is:

Comparisons demonstrating equivalence between different sensor sizes.


Click the link for a longer description, raw files, etc.

Here is 70mm f/4 ISO 640 on a 22.5x15mm sensor:

Here is 111mm f/6.3 ISO 1600 on a 36x24mm sensor:

Notice how similar the noise levels are. (The test is not meant to compare sharpness.)

The third comparison shows the same thing at low ISO:

70mm f/4 ISO 160 on 22.5x15mm sensor:

111mm f/6.3 ISO 400 on 36x24mm sensor:

As you can see, I didn't get the brightness exactly the same in post processing (the f/6.4 ISO 400 shot is slightly brighter), but even so the noise is highly similar.

The test images were not made with full scientific rigor, but hopefully they are still good enough to help demonstrate the validity of the point that f-number scales with sensor size as it pertains to noise.

 

[Ian MacLean]

There are a couple of issues with FF35 that I have in the context of this discussion.

One is that by the time you go to the ratios likely to be used most commonly, you lose much of the advantage...

This bothers me too. I suppose that with a spherical lens that needs to cover the longest (width) part of the ratio, the aspect of the sensor can have data that you'll never use without costing you much more than extra bandwidth in your workflow. I don't like IMAX aspect ratio, and can't imagine ever shooting anything in a squarer ratio than the golden rectangle. Most of the time, there's going to be wasted bandwidth with FF35, and that's a negative, and no one except Christopher has talked about it here yet. Or am I wrong?

*Edit: That's not quite right. Looking at Matthew Duclos' Circle of Confusion web page (which I found earlier in this thread, I believe), http://web.me.com/ducloslenses/DataRef/ImgCir.html makes it clear that if you're indeed not going to project or do a final render in a 3:2 ratio, you're paying for glass you're never going to use as well. This is just a rephrasing of the lens arguments already posted, but format aspect ratio should be factor in answering this topic's question. Maybe more so than it's already been. I've seen Mountain Light, and Learning to Fly, but they sure weren't in a 3:2 ratio... I mean no offense, but I am curious: what is your favorite ratio for theatrical presentation Tom?

...I will explain why I participated in this thread: I hate myths. The idea that smaller formats have an advantage in deep DOF, diffraction, noise, etc. is nonsense. I take efforts to share the facts so that misinformation might be reduced.

This thread has now been read more than 12,000 times, and has been an AMAZING resource, and a pleasure to read. Your insights have contributed more than anyone's. Your attitude, analogies, and expertise have been a pleasure to read, and I wanted to thank you for reducing 'misinformation.' There's nothing more gratifying, and I've learned a lot.

...I really think I have the right content, but my delivery must be very poor…

On the contrary. I find your comments concise, accurate, and your attitude exemplary. Don't apologize for anything. This is a technical subject with techie readers. Anyone with a disagreement is free to make a better argument, or offer a better, more concise proof. Good luck to them.

Mike's right in that a film image is essentially a raw image simply stored on film. It's just waiting to be transferred to to a computer. Film is still one of the most cost effective long term storage media in existence. There's a ton of information packed into a very little space.

And if I'm not mistaken, as Motion picture film scanners increase in sophistication and sensitivity, each new generation (of scanner design) is able to pull information from the film that wasn't previously attainable. That's a plus that shouldn't be overlooked, even if it is off-topic (sorry Tom).

 

[Gunleik Groven]

With all the great S35 gear out there fully fitting my needs, I really hope it becomes obsoleted by FF so that I can pick it up at good second-hand rates...