Good Example Of What The RED Cams Dont Fix

[S8 Booster]

Autumn in TOHOKU ( Shot on RED ONE )

Shoot.....film

[Nicholas Kovats]

You lost me. What "fix" are you referring too? The RED One fanatics
vastly overstated the original specs focusing on pixel count (nominal
resolution) over everything else including proper recording of color
space. I suspect Red One owners are abandoning these cameras for the
next generation.

This is one of the best analysis of the famously overstated Red One
abilities, i.e. http://rcjohnso.com/REDFACTS.html

So once again. What is the "fix" you are referring to?

[MovieStuff]

Well the only thing I see wrong are blown out highlights, which is the result of improper metering more than anything. I also see that all the time on reversal when auto exposure is used. Like with virtually all digital cameras, you have to underexposure to protect the highlights and then use a gamma adjustment to bring up the detail in the midtones during post. Otherwise, I think it looks great.

Roger

[carllooper]

This is one of the best analysis of the famously overstated Red One
abilities, i.e. http://rcjohnso.com/REDFACTS.html

Very true.

The Red One is highly over rated. The sensor sounds impressive (4K+) but it's a single chip sensor meaning that it's no better (indeed worse) than cameras with 3 chip 2K sensors.

A single chip sensor requires a bayer filter in order to capture colour images whereas three chip sensors don't need a bayer filter. The problem with a bayer filter is that it reduces the spatial resolution by about half (for green) and about a quarter (for red and blue). So the actual full colour resolution is between 1 and 2K.

In other words there's no real point in salivating over a Red. Find a good 3 chip 2K camera instead.

[Nicholas Kovats]

Welcome back, Carl! Looking forward to your entry into UP8 cinematography!

[S8 Booster]

just surfed by this one "live" compare Red vs film.
FilmConvert - Digital vs Film comparison

shoot.......eh.....film

[carllooper]

Ah yes. Nicholas knows I'm very excited by the UP8 film format. Am having my Bolex H16 modded in the New Year.

For most projects I use digital cameras. They are cheap, convenient, and give excellent results.

But there are some projects that just don't work on digital. They require film and the difficulties (of using film) are no reason to abandon such projects, or to compromise them by using digital. The difficulties are just simply an obstacle to overcome. Like walking to work. While it's easier to catch the bus there are many times when a walk is a lot more interesting.

Carl

[carllooper]

Well the only thing I see wrong are blown out highlights ... Otherwise, I think it looks great. Roger

Yeah - without more information I'm not sure what the original post means. My guess is that the intended meaning is that "a Red camera can't fix a god-awful boring wallpaper work".

Carl

[S8 Booster]

bad color neuances, image "flatness" (a typical digital imaging issue as obvious in the Red vs 250D), burnout (waterfall)

shoot.....

[carllooper]

bad color neuances, image "flatness (a typical digital imaging issue as obvious in the Red vs 250D), burnout (waterfall)

shoot.....

Ah ok. Yes. It is very flat. And the burnout. It's interesting that. According to the hype the Red is supposed to have about a 13 stops range (in the latest sensors) which is about the range film is supposed to have, ie. they should look equivalent but they never do. Digital always looks flatter. Perhaps film is under-rated in terms of range, ie. that film actually has a greater range than it's usually given.

Something is obviously amiss in this area. When you think about it, digital must represent exactly what is meant by all the parameters that are otherwise used to describe film. Since digital falls short of film it must mean descriptions of film, in mathematical terms, requires more work.

One of the problems in describing film mathematically (digitally/numerically) is that there are at least two parameters that are needed, but they are at odds with each other. There is range on the one hand ("depth") and resolution on the other. For a given image you can improve the range/depth but at the expense of resolution, and vice versa. This is related to Heisenberg's Uncertainty Principle. The more information you can obtain in one domain means the less information you can obtain in the other.

Hmmm.

My feeling is that grain is still not well understood in the digital domain. When describing film in numerical terms the grain is usually removed from the equations. This probably affects the numerical representation more than is understood. I've certainly found that in my own work. You have to take the grain out because there is no such thing as noise in mathematics. You use statistics (instead of math) to represent the grain (as noise). But to take it out you need to specify a threshold (based on the statistics)- and the selected threshold can affect the resulting numerical representation considerably. The point is that digital systems are built according to the mathematical representation. So if the mathematical representation (of film) is incorrect in the first place, the digital system is being built (and assessed) according to an erroneous representation of film.

To put it another way: when we say film has a "13 stop range" (for example) we're representing film in terms of a number (13). But that range is defined by we're we are drawing the line between information and "noise". However we don't really know we're that line really is because we we don't really know how good our brains are at reading a noisy signal. Or how good we can make AI capable of doing the same.

A digital camera, on the other hand, can be made to that numerical spec (13) but in doing so it cuts out, in advance, any information beyond that spec. Film doesn't cut it out. My feeling is that what we appreciate as "depth" is this extra information in film. Numerically it's "noise" but that's a limitation of the numerical representation rather than the actual film.

[MovieStuff]

bad color neuances, image "flatness" (a typical digital imaging issue as obvious in the Red vs 250D), burnout (waterfall)

I dunno. If anything, this clip shows how good the Red Camera can look for probably 90% of the shots even when used by someone with obviously limited photographic skills. Could it look better? Of course. But I am fairly positive that a roll of K40 in the hands of that same person would not even begin to look as good. The difference is the experience of the shooter; not the medium.

Roger

[Nicholas Kovats]

Carl. It will be interesting to see your algorithms applied to UP8 as it is a 113% and 39% increase in imaging area relative to R8 and S8 respectively.

Roger. I am in agreement to a point. There are amazing experimental artists who are very comfortable with the imaging limitations of their chosen film/video system and still produce mesmerizing work. I call it intimate film-making.

[carllooper]

If anything, this clip shows how good the Red Camera can look for probably 90% of the shots even when used by someone with obviously limited photographic skills. Could it look better? Of course. But I am fairly positive that a roll of K40 in the hands of that same person would not even begin to look as good. The difference is the experience of the shooter; not the medium.

What about in the hands of someone who can get the best out of both film and digital? Would they get a better result in film or digital? I'm sure they could get better depth in film. From a purely observational point of view I haven't seen any digital work that has the depth of the best film work. Regarding spatial resolution, in most practical circumstances (for which even the best are constrained) modern digital cameras compete quite easily with film.

But it's the depth which is difficult to appreciate - even in the hands of the best digital filmmakers. Digital sucks. It sucks out all of the depth. Single chip cameras such as the Red even more so. Presumably future cameras will solve that problem. They just need to study how film works a bit more.

[carllooper]

Carl. It will be interesting to see your algorithms applied to UP8 as it is a 113% and 39% increase in imaging area relative to R8 and S8 respectively.

Yes - I'm quite excited by UP8. Am looking forward to shooting UP8 and then pumping it up through the digital processes I've been developing over the last year.

To obtain a 2.8:1 aspect on Super8 would require either cropping S8 or using an anamorphic.

If using an anamorphic (hard to get, clumsy, colour refraction problems, etc) the comparison on bandwidth would just be a comparison of the emulsion area used in each:

UP8 = 10.52 x 3.75 = 39.5
S8 (anamorphic) = 5.69 x 4.22 = 24

UP8 = 164% over S8

Cropping is the only other way. Doing it that way the comparison becomes:

UP8 = 10.52 x 3.75 = 39.5
S8 (cropped) = 5.69 x 2.0 = 11.56

UP8 = 346% over S8

Carl

[S8 Booster]

Well, i do have a Sigma DSLR 14 which has a 3 layer RGB chip aka film but it is nothing near the colour balance of that of my iPhone on colour balance (real compare)
Sigma sample image:

Which again is a far cry from film.

Shoot....8.....