Grain Aliasing

[carllooper]

This is something I don't think I've ever addressed in this forum.

Grain aliasing is that which produces what might be called a "pseudo-grain" in a digital transfer. Anyone who has compared the projection of film, next to a digital transfer of the same, will probably understand what is meant here. One will see the transfer is grainier than the projected film.

If we call this additional grain "pseudo-grain" it's only because it's not what we might otherwise call the "native grain" of the film. But what is causing such pseudo-grain?

Well, it's not due to the digitising camera, nor is it due to the film. It's the result of interference between the two ways of representing an image (analog and digital).

Its something that has been known about for more than a decade now. One of the earliest online insights into such (dated Sept 2000) is here:

http://www.photoscientia.co.uk/Grain.htm

The simplest solution (based on traditional sampling theory) is to scan the film at a definition higher (at least 2X) than that which will be one's target output definition, apply a low pass filter that removes frequencies above the target definition, and then re-sample the result down to the target definition (the release version).

Now importantly this is not the removal of that native grain which belongs to the film. It's the removal (or avoidance) of that additional grain which would otherwise be created in a lower definition transfer.

An interesting idea that follows from this is that the native grain of film (so called) could also be regarded as "pseudo-grain". That native grain is also an alias of some more fundamental "grain". But that will have to remain science fiction for the moment.

-----

But what is grain anyway?

Its not the individual clusters of silver in a film (or dye clouds in colour film). Although often the term "grain" is used to mean exactly such a thing. Rather it's that phenomenological effect - that shimmer in an image. If we otherwise keep the term "grain" to mean clusters of silver or dye clouds, then we could use terms such as "graininess" or "granularity" or "noise" for what we're otherwise talking about.

Our question could then be rephrased as "what is noise"?

At a fundamental level, noise is a physical expression of the mathematical concept of zero.

But from a mathematical point of view there is no such thing as noise. "God does not play with dice" as Einstein once said. Within a mathematical framework "noise" would be a signal, and such a signal would simply represent whatever reality was the cause of the so called "noise". If we knew the cause it wouldn't be noise. It would be an image of the cause. A signal, signifying the cause.

But from an alternative perspective a signal doesn't represent anything at all. On the contrary it will be "things" that represent a signal. Things such as pixels, silver, paint, roman tiles. The signal itself is the reality.

Now we make a distinction between a signal and noise. Simply put: noise is not a signal. Noise expresses the absence of a signal. Expresses a zero signal.

And as a provocation to the mathematically inclined (of which I am one myself) we propose that there is no such thing as zero, or more generously, not physically so. There is instead the idea of zero. The concept of such. Which we might say is no less real but on a different plane of existence perhaps. Or if we are to be more provocative we could say zero is a figment of the imagination. But we can nevertheless physically express this idea through noise. But it can only be an expression since any exact equation between noise and zero would require an infinite sum, which is unobtainable (and justification for calling zero a figment of the imagination). In it's place we have instead: noise.

If we are to recast this in terms of a traditional cause and effect way of speaking then we might say noise (as an effect) is the result of an absent (zero) signal (the cause). It is due to a lack of information.

...

C