Cryogenia mixed with Sunny by Rocky Gresset and David Reinhardt
Merci a joethepro over at personal view’s forums
GH2 with Valkyrie 444 TYPE-ZERO4 BETA5 patch
mostly VFR 80% • Edited and graded in Resolve
Ok I stop pelling the potatoe and release a magnificent shake of pink poopooza
The starting point was very simple, once again I had ascertain that even at the lower ISO of my camera (160) there was already a lot of noise being cooked in. As in most if not all modern single-chip digital cameras, from the three R(ed)G(reen)B(lue) channels the stream output’s composed, blue is the noisiest, followed by red and finally green is the cleaner by a tyronosarius’ head. POC:
Without going into the properties of light itself (e.g. hue’s wavelenghs) this happens because the current technology mimics the physiology of the human eye, which already is an incredible subject for study – see cone cells. Thus the Bayer grid (named after its inventor, Bryce Bayer of Eastman Kodak) that filters the fotons, has 2 green “cells” for each Red and Blue. # See PS at the end for the analog process
There’s many many many implications/consecuences/”problems” deriving from the adoption and implementation of this technology; despite I have no degree in physics or digital image ingeneering, I could probably go over some of them in a reasonably informed way, but quite honestly I don’t think there’s any benefit (4 you or me) in doing so. Nevertheless and as a footnote, never stops to amaze me that digital images are just an algorythm’s interpretation (quite credible too) of an already constrained portion of what’s possibly there … and yet the human eye functions pretty much the same… so what’s realy out there that we cannot see? I shall ask my neighbor who is brazilian (???)
Ok, so coming back to my very concrete “problem”, in the proccess of trying to mitigate the noise, I end up testing a couple of several things:
A – Inspect, reflect, hollyflec and conspurcate each RGB channel separatelly to understand better the bitterness of my good seagul friend’s soul and how to ease it without robbing a boat
B – Use only the information in the green channel (again, the cleaner) to construct the luminance of the image
C – Blur the noise on the chroma side of life just a tiny fly-fart, chuf
D – If possible, adopt the technique into my already chaotic workflow, particularly experimenting with (cross-procces kind of) negatives
MAKE A PAUSE TO ENJOY LIFE WITH MONONEON
After learning how to do the routing, omelette without eggs and of course much researching, I was left with a final issue; I had been employing free Tuttle’s ChannelShuffle OFX to achieve the green-luma thing and though this plugin works in Davinci Resolve it wasn’t build with the platform in mind which produces a slow and not as stable as desirable performance. So trying to mimic the effect while speeding up the process natively in Resolve I turned to PV forums. There the user joethepro suggested DR’s channel mixer, which if not e x a c t l y the same, producced a very similar result; my cow’s a photocopy of your ponny man!!
Up mixer, down shuffle, stage before color grinding; once again thanks Joe
Maybe all of this sounds nerdy and complicated, but once the proccess was integrated I ended up having more control over the textures (imagine the threads that make a blanket), ditching some of the luma noise and bluring a bit the chroma one without compromissing the speed and quality (gypsy selling stolen women under-ware in sundays’ market CHIP CHEAP, 5 for 1, only best quality!!!!)… stone soup nevertheless; still difficult to accept how much coherent and pleasant my 10 year old pany dvpal 3ccds is colour wise =)
To finish this little chapter, now – and while there are some nodes that do nothing, too lazy to delete them and once in a while they alert from police – I can apply the franken-tree-grade to a group and then just balance a shot in the very first node and changing the whole look of it by exchanging the LUT node at the end. Worse enemies than noise also peep out from the footage but that battle, BandingLoo, I already fought it some time ago with mixed results.
Joethepro also suggested trying the technopotatoe vintage strip colour thingie (basically “reverting” RGB channels to its complementaries – see spectral senttizers aka dyes below), so we stript and stript and stript, with this very well explained fashion by DIT Spot
The footage is from BlackMagic’s commercial shot by the good John Brawley, you might find it here, 11GB though!!! –Internal_House_Shot.zip
For just 1 DNG frame, 10 MB DL
Soft-stript-frame-grab-xample of quick and dirty grade
For comparison sake and because it’s such an organically balanced complex achievement of humans, here’s a basic layered structure of photographic panchromatic film before and after being developed and explanation of the process, courtesy of Rochester
Photographic film captures the image formed by light reflecting from the surface being photographed. The sensitive elements in the film are crystals of, most often, silver halide which can change their structure when excited by light (photons). In general less sensitive films (slower films) have finer grains that are closely packed and more sensitive films (faster films) have courser grains. A film may have a distribution of grain sizes to obtain certain desirable properties. The reason for the sensitivity relationship to grain size is related directly to how the grains are converted from a stable non-developable state to another stable state (latent state) from which they can be developed chemically. This happens in something like the following way. When a photon of light strikes a grain it dissipates its energy in the crystal (grain). This energy may or may not be enough to flip the crystal into a latent state. Generally it takes a few photons to flip the grain (depending on its size and sensitivity). In the meantime, thermal energy is jiggling the grain and tending to drop it back into its normal state. If enough photons strike the grain in a given time, the grain flips to a latent state and sticks there. We then have a grain that can be turned opaque chemically. Thus the photons build up a latent image that is later developed. The darkness of the image is more or less proportional to the light striking the film. It takes about the same number of photons to flip a large grain as a small one. Since the larger grain intercepts more light more of the larger grains will be flipped and thus less light is required to create a latent image. This later phenomenon makes coarse grained films faster (more sensitive). All the silver halide salt particles reside in some emulsion layer depositing on the base layer of films. Color film has three layers of emulsion for three kind of silver halide which are sensitive for the light of 3 different light wavelengths.
And a diagram showing how the photons make it through, literally flipping our allies, I mean hallides, exciting ONLY the layer which contain crystals’ grains sensitive to their wavelength; those layers have been dyed with organic molecules known as spectral sensitizers that help with the “colour blindness” of the crystals. If the silver-halide crystals get crazy they’ll go looking for holes to make unstable silver (uuuhhh); given the miracle that enough photoelectrons and positive holes gather in the crystal lattice’s party club (at least two to four silver atoms per grain combine) a stable latent-image site will be borned.
You’ld never want to “see” that, cause you’ll end up killing the baby; it is only when the film’s processed – using reducing chemicals that make exposed silver-halide grains develop to pure silver (later bleached) and washing away the crystals that hadn’t change at the fixing stage – that it gets inverted to its complementary colours and the overall orange-brownish looks (from further dyes to fix stuff). That’s so because of the oxidized developer reacting with chemicals (couplers) in each of the image-forming layers (like in life, many insensitive layers in there too). This reaction causes the couplers to form a colour which varies depending on how the silver-halide grains were spectrally sensitized (dyed for traumatised people)
- Red-sensitive layers form a cyan-colored dye.
- Green-sensitive layers form a magenta-colored dye.
- Blue-sensitive layers form a yellow-colored dye.
After washing and drying, there’s no silver anymore, ohhhh… just a substractive based colour long delicate magic band we now “have to print” and print real good and then we restart the process with exposure and contrast tests’ strips and meet our miracle silver workers again… pure TAO, cho biuchiful… sniff
For an in depth explanation see – How Photographic Film Works
A last nostalgic tearful, heart bleeding, soul ripping angst look at the faces of our good hard working little crystal meth friends, I miss y’all crazy dudes =(
Finally a quick aniso vs iso tropics’ grid 😛 or how apparent chaos can be more diligent than rational order or still how intuition caught a giant tuna with a lullaby song.
As now you might want to replay the video and it’s so up there, here: