[Tig] Peter Swinson on Super 16

[Rob Lingelbach]

There was a local problem at colorist.org with Peter's message so he  
asked me to post it.   --Rob
----

So the world of imaging has lost interest in satisfying the subconscious
part of the human visual system.

Texture and Stochastic resonance are what makes an image look real.
Take away the minutia of the texture and the tiny amounts of "noise"  
and we
see a very clean image that consciously looks good, but can make us feel
uneasy. Why?

Our visual system is still tuned to our prehistoric ancestor's view  
of the
world, 100 years of viewing film or video has not changed our brain's
subconscious assessment of our surroundings. Very very few, if any,  
natural
world objects are devoid of texture, indeed very few man made objects  
are
totally devoid of such texture.

Examples. Grass, no not the type you smoke! The type you walk on.  
Look at
it, fine detail, maybe only about 1% brightness modulation, but you know
its grass, you don't think about it, your subconscious "feels it". A  
carpet
with no real pattern, the weave from any normal distance will be very  
fine
the difference in brightness between strands may only be 0.5% but the
visual system instantly and subconsciously recognise it as a carpet,
likewise with many plain wooden structures such as tables chairse  
etc, the
grain s fine with very low modulation but we know without thinking  
that it
is unpainted wood.

In none of these instances do you think, grass, carpet or wood because I
can see the clues. It all happens subconsciously. Now add the random
sampling of these textures by the frame to frame granularity of film,  
where
the granularity is below that of the object.the grains random sampling
still allows the texture to be gathered. This granularity, due to
stochastic resonance in our visual cortex enhances the extreme subtle
density differences seen by our visual system.  Remove the grain and we
loose these subtle shades, suppress the very low level but high  
resolution
detail and we get clean images BUT they no longer looks "real". Why not?
Because subconsciously we have lost those clues as to what the materials
are. Yes we can often guess from the image context, but that guess takes
subconscious effort. And that's why such "cleaned up images" no  
longer look
relaxing. Our recognition system is working overtime to tell us what  
we are
seeing. It doesn't exactly hurt us, but there is a certain unease when
items look too "clinical". This is where film as an acquisition  
medium has
it over nice clean video.

And where does this show most, human faces, low texture looks weird,  
just
look at the commercials with the grain reduction for women with those
removed wrinkles, if they look real to you then heaven help us.  I have
sent Rob an example of a noise reduced facial shot with grain added  
back to
show how stochastic resonance assists our visual system see subtle  
shades.
Hope he posts it. Look in particular at the scar on her face, almost
imperceptable in the "clean" version, but shows just that bit better  
in the
same image with the noise added!.

If you want to keep the subconscious vision happy shoot film, S16mm  
or 35mm
and be careful what you remove. If keeping digital compression bandwidth
low is your goal, then clean up as much as you want, but do not claim  
that
the images look more real.

Nuff said !

Peter