Re: [Frameworks] color separation

From: Abigail Severance <>
Date: Sun, 03 Apr 2011 15:29:48 -0700

thank you for the details - intriguing, and very generous of you to share!
Abigail Severance
Los Angeles

On Apr 3, 2011, at 1:30 PM, Amanda Christie wrote:

> Hi there,
> yes, that is my film. Funny, I almost missed this thread entirely as
> I've been super busy lately... but the title "color separation" caught
> my eye, and I thought... hmmm... I like colour separation... I'll
> read this thread... and without even reading the name on the link, I
> clicked on it, and was surprised to see "oh hey! that's my film."
> so, I know how is was done (conveniently enough)... I had done a crazy
> amount of research and reading on the history of technicolour and dye
> transfer processes, and then I also did an insanely unnecessary amount
> of testing... and I lived and breathed colour separation for about two
> years... so i'm happy to tell you exactly what i did (in more detail
> than you probably want to know... but here it is... it's taking up too
> much space in my head anyway).
> 1. DANCE CHOREOGRAPHY - this was made specifically for the film,
> with the colour separation process in mind. I choreographed three
> separate solos: a red, green, and blue solo. I spent four months
> choreographing and rehearsing with a metronome in the dance studio to
> make sure that my movements would be as frame accurate as possible, so
> that movements would sync when I wanted them to, and so that I could
> also have more ease and freedom when it came time to manipulating
> movement on the optical printer. Given that I knew I was going to be
> using an optical printer to recombine the colours, I choreographed
> accordingly: deciding that this section of the dance would be slowed
> down, this would be in reverse, this would be freeze frame, with would
> be sped up... etc. etc. So I decided long in advance where every
> single cut and movement manipulation would be... i.e. the red dancer
> might be moving forward in 1:1 ratio, while the green dancer went in
> reverse at a 2:1 ratio, while the blue dancer was in freeze frame,
> etc. So all of the editing was essentially done in the dance studio,
> as I kept meticulous second by second notes and logs in a notebook.
> 2. FILM STOCK TESTS - as I neared the end of the rehearsal stage, I
> started testing film stocks - both black and white and colour. I wish
> I could give you specific details, because I found A LOT of useful
> information... however this was years ago, and my notes are packed
> away in some box in storage in another city because I move too much...
> however... here is what I do remember: I shot the live footage on
> Kodak 7222 (double X) BW film. It was not ideal, but it was the best
> option from all of my tests. I optically printed onto Kodak Vision
> 200T (I think).... again, this was not an ideal but it was the best I
> could get... I had spent hours on the phone with one of the Kodak
> engineers on several occasions as I shot and analyzed tests and he was
> very helpful... in the end we found that the best stock was a very
> specific lab print stock (but I can't remember the name)... and I
> wanted to use that one, but it was only available in 2400 foot
> quantities which was fine... but it was also out of stock and would
> have taken too long to come in... so I just went with the Vision 200T
> (which was unfortunate, but sufficient)...
> 3. PUSHING AND PULLING: I spent 45 days straight doing continuous
> testing and analyzing (I'm a bit of a perfectionist). I used an
> analysis projector to project and analyze all of my tests frame by
> frame. I found some very interesting results, especially relating to
> the yellow/blue layer. Basically, this is the trickiest one to get...
> in order for the yellow to show up at all (this is the blue
> choreography filmed with the blue filter, but the dancer looks
> yellow)... I had to increase the contrast. So for the blue / yellow
> layer, I pushed the BW 7222... I can't remember how much... it was
> either 2 or 5 stops... I remember that it reduced the ISO to about ISO
> 5 or 25 or something like that.... the Blue filter cuts out a lot of
> light, and you have to compensate for that as well... I think I pushed
> the blue film 2 stops (but again, my notes are packed away somewhere).
> 4. DOUBLE PERF FILM: because registration was going to be very
> important for having everything line up, I special ordered double perf
> 16mm film and I used an Oxberry optical printer that took double perf
> film.
> 5. SHOOTING THE LIVE ACTION: I used an Arri SR2 and 3 magazines. We
> loaded the double perf 7222 into the magazines and labelled each
> magazine either Red, Green, or Blue. Then we would sandbag the tripod
> to make sure that it didn't move and put on the red magazine and put
> the red filter in front of the lens, and I would perform the red
> choreography for that shot (I had predetermined all of the camera
> angles and shots in the dance studio as well). Then I had someone
> doing continuity who would take a black sharpie and mark on the floor,
> my start and end points (If you look closely in the film you can see
> some of these black sharpie marks). Then we would swap the "red"
> magazine for the "green" magazine, swap the red filter for the green
> filter, and I would go back to the start position for the green
> choreography and we would film that... then we would do the same for
> the blue. Once we were satisfied we would set up for the next shot
> and do the same thing... There were a total of 18 shots, and 3 colour
> layers per shot... so it took us about 2 days to shoot. We needed a
> lot of light because of the blue layer, and we were in a big warehouse
> studio with lots of windows, but it was a cloudy weekend so we pumped
> in another 7k of light as well from the sides an the back just to get
> the right exposure on the blue film.
> 6. ADDITIVE AND SUBTRACTIVE PRIMARIES: side note here... I was
> wearing black against a white background, so the red layer looks cyan,
> the green looks magenta, and the blue looks yellow. If I had worn
> white against a black background, red would be red, green would be
> green, and blue would be blue... but I found the true red, green, and
> blue to be too garish (the green is very frankensteinish)... you can
> see the true red green and blue in the wider shots when I pass off of
> the white wall and move in front of the darker backgrounds so you can
> get a sense of how garish that green is.
> the BW footage back.. that BW print became my original, so I was very
> gentle with it. I used an analysis projector (Red Lake with digital
> counter and dimmer style speed control) that I borrowed from Christoph
> Runne to log my footage. I used the closing of the slate as my "0"
> and then marked down the frame number for every single significant
> movement. i.e. I was at the height of my jump at frame 263, foot
> touches the ground at 301, bottom of the descent at 348, head turns at
> 409, etc. etc. So I wound up with pages and pages of detailed logs
> of the frame number of every single movement for each layer.
> 8. FINAL PLANNING, RATIOS, ETC: because I was planning to edit the
> entire thing in camera in the optical printer (i don't know why, but I
> always had a fear of neg cuts so all of my films until the one I'm
> finishing now, were edited completely in camera on an optical
> printer)... so I spent a few days with these notes and planned out the
> film. I made an excel sheet (12 point fontsize) with columns to put
> in the following info: shot, description, Colour Filter, Red ND
> filter, Green ND filter, Blue ND filter, Special Instructions
> (Forward, Reverse, Still), Speed Ratio, Source Frame Start, Source
> Frame End, Camera Frame Start, Camera Frame End Notes. So I planned
> out frame by frame when each cut would happen, when the motion would
> change speed, and direction etc. I also colour coded my notes and
> used pink highlighter for anything that was in reverse (so I wouldn't
> miss it), and yellow highlighter for any freeze frames... if it wasn't
> highlighted it was going forward.
> 9. ND FILTRATION: once I had all of that planned out, I did further
> testing on each shot to see how much ND was needed on each colour
> layer. Basically, I had this idea that I needed to be able to achieve
> my own skin tone, a white wall, a black dress, and a 20% grey floor on
> my own in the optical printer... and that if the lab needed to do any
> colour timing that I had failed. so I always told the lab... NO
> COLOUR CORRECTION. And while most shots were mostly the same, some
> had variations, so I'm glad I did this. I then wrote down those ND
> filtrations for each of the shots.
> 10. THE ROOM AND THE SET UP: I posted these notes and charts on the
> walls of the room... I covered two walls from floor to ceiling... it
> was kind of crazy. When a friend walked in to visit me one day, he
> said "oh my God, this is makes me tense just being in here, I have to
> leave".... and it was a bit crazy.
> colour film into the oxberry camera of the optical printer, and I
> would load the black and white original (which I will from now on
> refer to as the record... there was a red record, a green record, and
> a blue record) into the projector, and I used the same filter on the
> optical printer that I used when shooting the original. I lined up
> the record on the projector to the closing of the slate, and set the
> projector counter to "0" and advanced to where I was going to start...
> and away I went... once the red record was filmed onto the colour
> film, then I would rewind the camera the appropriate number of frames,
> then load the green record... do what I needed to do... then rewind
> the camera and load the blue record and do what I needed to do... then
> I was ready for the next shot... and on I went for all 18 shots... all
> done in camera. The actual shooting of the final film on the
> optical printer in this way, I think only took about 7 days of working
> 10-15 hours per day.
> since I was doing colour separation and editing in camera, I really
> wanted to do a few things that would not have been possible
> otherwise... in order to really maximize this process and technique.
> So there are a few of shots where I cut the colours in canon - the red
> cuts to the next shot, then the green, then the blue... this cuts up
> the space and plays with rhythm. I tried to play with the order a bit
> too... Red, Green, Blue... Blue, Green, Red... etc.... and I also
> tried not to do it too much -- I didn't want overkill.
> 13. ERRORS: so I dropped the film off at the lab and when I got it
> back there were 4 mistakes. Perfectionist that I am, I was horrified
> and decided that I would never finish the film or show it to anyone...
> I won't lay them out here for everyone to see... suffice it to say
> that there is one section of the choreography that was supposed to be
> printed in reverse (it looked BEAUTIFUL in reverse on the analysis
> projector)... but one night when I was printing at 2am... I didn't
> notice the pink highlighter on my notes and I printed it forward... to
> me... I cringe whenever I see that section... it just looks so bland
> and terrible! it's just one section of one colour... so I'm sure no
> one ever notices... the other 3 errors were colour correction
> related... one image was too magenta, and I forget what the other
> errors two were.
> 14. FINISHING IT UP: so a year later, my friend Chris Brabant was
> finishing a film, and he's a whiz at optical printing and asked me how
> the RGB dance film went, and I said. "It's a failure I'm not
> finishing it." and he replied... well, I'm premiering my new film at
> the Pacific Cinematheque, and I want yours to premiere with it too...
> It's all booked and the publicity's done... you've got 'till Nov. 28
> to finish the sound!" I panicked... but it was a good thing. I am so
> grateful to Chris! Had he not done that I would not have finished
> it. sooo... here are the last steps:
> 15. CREDITS: I shot the title and credits separately (edited in
> camera) and I used a neg cutter for the first time just to cut the
> title on the front and the tail credits on the end.
> 16. COLOUR TIMING: in the end I did use a colour timer... at the
> Technicolour Lab in Vancouver... they were so so super helpful! I
> went in after hours between midnight and 6am and I basically job
> shadowed the guy as he worked on other projects and when he was in
> between other jobs we worked on my film together. It was super
> educational for me too as I watched him colour timing other films
> (Alien vs. Predator, Fantastic Four, and a Neutrogena commercial) were
> the three other jobs being done that night.
> 17. SOUND: we filmed this film in sync sound. From the very
> beginning my plan was to create the music sourced from the movements
> of my body... skin on skin, breathe, feet on the floor, hand in hair,
> etc... At the 11th hour at the last minute... I wound up hiring a
> professional sound designer to help me get it done on time, and we
> worked together on that. I then added the musical bits by watching
> the red version and humming with it (improv), watching the green
> version and humming with it (improv), and doing the same with the
> blue... so the humming tracks were all done independently improv
> style... it took about 4 takes to get something I liked... a nice play
> of both harmony and dissonance. And all of those electronic sounds
> you hear are actually processed from the raw sync sounds of my body in
> the space at the time of filming.
> 18. WHAT ELSE? - there are 18 shots in the film, so I feel like I
> should have 18 points about process... but I feel at a loss... OH! I
> KNOW! yes... it's called "3part Harmony: Composition in RGB #1"
> because my original plan was to do others... composition in RGB #2,
> #3, #4, etc... but this was so consuming and exhausting that I swore I
> would never do one again... and also, I was relatively unhappy with
> the end product... ironic since it seems to do quite well in terms of
> screenings and feedback... it's actually one of my least favourite
> films... however... recently... I've started playing with the idea of
> doing #2... in a much more relaxed and improvisational way... less
> anal retentive... now that I've got the research and testing out of
> the way... I want to sink into it more comfortably (maybe that's what
> I don't like about this film... maybe it feels too much like a mere
> technical study or a sketch from something more interesting and
> finished... and yet, that's the part that I like about it too).
> So there you have it.
> All you ever wanted to know about the technical process behind 3part
> Harmony and more.
> I still have the original 16mm film footage of all of my tests and an
> analysis projector, so if any of you ever want to invite me to give a
> talk to a class or something about colour theory, I have more than
> enough visual teaching material (power points too that I showed my
> crew in advance with charts and graphs on the panoptic curve and ways
> that the human eye responds to colour)... so yeah... I could probably
> go for about 2-4 hours on colour theory and optics if anyone ever
> wanted me to. I was quite sick of it for a while... but the knowledge
> is there, taking up space in my head... valuable mental real
> estate.... sigh.
> Thanks for your interest in my film, and I hope this info is useful to
> some of the people on this list (and that the others just skipped the
> message without reading it all and getting annoyed).
> Amanda Dawn Christie
> --------------------------------
> 506-871-2062
> On 2-Apr-11, at 1:18 PM, Todd Eacrett wrote:
>> Have you tried contacting the filmmaker? Amanda is on Frameworks, so
>> I'm sure she will see this and respond.
>> As I understand, it was indeed 3x B&W with filters, optically
>> printed one at a time on colour stock, similar to the technicolor
>> process.
>> ----- Original Message -----
>> From: Daniela Zahlner <>
>> Date: Saturday, April 2, 2011 7:25 am
>> Subject: [Frameworks] color separation
>>> hello,
>>> does anyone know how to exactly make a color separation?
>>> i want to have a similar effect as in this film:
>>> rgb-1.html
>>> as far as i understood you shoot b/w film three times, each time
>>> with another filter (or two?).
>>> i found these numbers:
>>> Wratten color separation filters
>>> Filter no. 47 blue
>>> 58 green
>>> 25A red
>>> and then just print to color print stock?
>>> and what if i shoot reversal?
>>> does anyone have experience? any lead would be helpful!
>>> daniela
>>> --
>>> GMX DSL Doppel-Flat ab 19,99 Euro/mtl.! Jetzt mit
>>> gratis Handy-Flat!
>>> _______________________________________________
>>> FrameWorks mailing list
>> _______________________________________________
>> FrameWorks mailing list
> _______________________________________________
> FrameWorks mailing list

FrameWorks mailing list
Received on Sun Apr 03 2011 - 15:30:17 CDT