Grain Size vs Pixel Size

[carllooper]

The left image simulates a film camera image and the one on the right simulates a digital camera image.

The sampling window in both cases, is the same average size (10 pixels) but the position of the sampling window for the film simulation is random (and overlapping), while the position of the sampling window for the digital simulation has a fixed pitch (and doesn't overlap).

Note that the goal here is not to create a simulation of film as such, but to look at how randomly positioned and sizing of a sampling window (ie. in general) provides for an arguably richer and more organic mediation of the original signal than does a fixed pitch sampling window of the same average size. It also shows how the size of film grains (or dye clouds) with respect to pixel size, would not provide a meaningful metric in terms of the relationship between the two.



The technique used in the left image is more clearly visible in the following images where I'm using larger sampling windows that reduce in average size towards the one used in the above comparison. Each sample captures the average of the original signal within a randomly positioned and sized circular/elliptical region, approximating the way in which grains (or dye clouds) would register light and overlap each other in film.

[Nicholas Kovats]

Once again an excellent overview, Carl.

Aaton (France) knows film structure and fim camera manufacturing. Hence their recent debut of their all digital camera called the Penelope Delta whereby the Canadian manufactured sensor (Delta) actually moves to simulate the inherent randomness of film particles and therefore "mimic" film transport induced high resolution 24fps motion! I am repeating myself once again but we still have NOT exhausted the capabilities/potential of the film medium and film camera/transport systems. Film is THE definitive engineering reference point in my book.

http://nofilmschool.com/2012/10/aatons- ... ta-camera/

"Something even more space-age about the Delta’s sensor, though, is an option which allows the sensor to actually move in place. This capability oscillates the imager by a half-pixel offset each frame, randomizing the noise structure of the image. Fixed noise structure is a difficult problem to bypass in the digital world, but it seems as though Aaton has done so to a point never before achieved. Moreover, Aaton claims that the resolution created by this sensor movement actually increases the effective spatial resolution of the imagery over time. Though any given single frame will resolve whatever a debayered 3.5K image equates to, in motion the consecutive frames combine to virtually resolve an estimated 7K before debayering. That’s right — the Penelope Delta’s sensor can resolve 7K RAW through a method of time-travel (not really but you can understand what I mean)."

http://blog.abelcine.com/2012/09/27/aat ... pe-camera/

"Another revolutionary design is in the sensor, which is the first in the industry to be mounted to a moving assembly. By offsetting the physical position of the sensor by half a pixel with each frame, the spatial resolution is increased over time.

This is akin to film image capture, where the random structure of grain and silver halide crystals in each frame creates greater image resolution in the changes that happen on the image surface level between frames. In other words, while an individual film frame may appear relatively low in resolution and high in grain, the random structure of the actual imaging material on a strip of filmstock means that information is captured in different spots and grain deposited in different positions from frame to frame. When these frames are shown successively one after another, the cumulative effect is an overall increase of resolution and a decrease in visible grain structure. This is increased temporal resolution via increased spatial resolution.

In the case of the Delta Penelope, when the sensor movement option is engaged, the resolution increases from its native 3.5K to a virtual 7K. The movement of the sensor also eliminates fixed-pattern noise, a major issue with standard electronic sensors; the offset is also tracked and noted in each frame’s metadata for post interpretation and plotting. This unprecedented moving sensor may well be the ultimate in Aaton “lateral thinking.”

[PyrodsTechnology]

Again, why do not shoot directly with REAL film ? (gaining better resolution and quality by the way)
Regards.
Roberto

[Nicholas Kovats]

Roberto,

I admire your passion. Both Carl and I are both film advocates and users. However we both work in the digital realm and have similar interests in originating and facilitating film based digital hybrid approaches.

[carllooper]

Again, why do not shoot directly with REAL film ? (gaining better resolution and quality by the way)
Regards.
Roberto

Hi Roberto,

the work presented was done entirely in the digital domain (from a digital original found on the web) but the ideas being explored there are not in any way advocating shooting digital, or advocating the digital simulation of film (although I'm not necessarily dis-advocating that either). Rather it's an attempt to demonstrate, if only from a digital perspective, why you might indeed shoot film instead of digital.

It's a way of convincing myself, from another perspective, why I'd shoot film. To put it another way, even though when I see film projected I can already see why I'd shoot film, it's also comforting to partly understand why that might be the case.

As Nicholas has suggested, the workflow being pursued here remains shooting film (for the reasons given as much as any other reason) but there is also it's subsequent scan to digital as part of the process (if one does that). But rather than treating the scan as an afterthought, or no more than a copy of the film, (eg. for the purposes of distribution) it is to treat the digital part of the pipeline as an equally important part of the overall process being advocated. That wasn't discussed here but the discussion feeds into that. What happens during the transfer? What might you lose? What might you gain? The question could equally threaten to become: why not shoot digital in the first place?

The real answer is that such questions are irrelevant. Work in film, work in digital, or in my case, a hybrid of both. Work in clay. Work in paint. Work in chemistry. Work in astronomy. Technical considerations don't really tell you what domain to work within. It's only when you've sort of already decided on a particular process that the technical considerations start to play a role. And then only to clarify what you've probably, intuitively, already grasped.

Carl