Technicolor. When you say the name it harkens back to the golden age studio era. It brings to mind the larger-than-life stars of the silver-screen, epic production values, and a hyper-vibrant punchy, onslaught of color that overwhelms the eye. While the complicated technological and chemical process of 3-strip Technicolor film is no longer available to filmmakers, we can approximate the signature Technicolor look digitally in DaVinci Resolve. The benefits to using Resolve instead are: you can process it on your own computer rather than at an industrial lab; you have much more control over the final image; you do not have to worry about the hassles of shooting at ASA 5; and you save yourself about a metric ton of film-stock in the process. Let’s get started…
When you’re replicating a specific film look, it’s always necessary to do the research first. understanding the analog process involved goes a long way to simplifying the digital approximation. There’s a lot of great information on the internet on the specifics of Technicolor and how it worked, but below is a quick and simplified explanation, so you can see how the analog process is replicated digitally.
3 Strip Technicolor, as it’s name suggests, used three different strips of film running through the same camera at the same time. This used three times the film stock as regular black and white film, and was very complicated and expensive. The film stock itself was monochromatic (black and white), and the camera used a beam splitter and colored filters to send all of the RED light to one strip of film, GREEN light to another strip, and BLUE light to the final strip. So what you ended up with was three strips of black and white film with slightly different exposures of the same image, based on the color of the original scene.
These three black and white strips were then run through a tinting process, that colored them with a dye of their complimentary color. The RED strip was tinted CYAN, the GREEN strip was tinted MAGENTA, and the BLUE strip was tinted YELLOW. One at a time these strips were physically pressed onto a single strip of clear print stock, and the dyes imbedded in the gelatin of the final positive film. What’s interesting is that making the final print was an entirely mechanical process, not photographic. There were no photosensitive chemicals used at all, just physically pressing film together with very tight tolerances. Pressing the CYAN, MAGNENTA, and YELLOW dyes together, made the full range ,vibrant, ultra-saturated, glorious Technicolor image we know today.
Step 1- Setup and Splitting the Shot into RGB “Strips”
Now, on to replicating this process in DaVinci Resolve. First lets start with our initial image. I’m going to be using some footage released for the BlackMagic Cinema Camera by cinematographer John Brawley. It’s freely downloadable, so we can all work from the same shot, and the RAW Cinema DNG files have a lot of latitude.
This tutorial pre-supposes that you already have basic knowledge of using Resolve. First setup your project settings in DaVinci and add the file to the media pool. It’s important to make sure that the CAMERA RAW settings for CINEMA DNG are set to BMD FILM for both GAMMA and COLOR SPACE. Otherwise the computer will try to automatically correct the frame for REC709, and it sucks at it. The frame itself should look something like this to start, washed out and low-contrast.
Let’s enter the COLOR room and get started. For the first node you should adjust the lift/gamma/gain to get the full dynamic range of luminance out of the shot. Don’t adjust color or saturation here, just adjust the brightness and contrast as if you were doing a simple primary grade.
Next you will want to add three more corrector nodes, and name them RED Layer, GREEN Layer, and BLUE Layer. Then connect the orange dots (RGB Output) from your initial node to the orange dots on your new nodes as in the node tree below.
Now we need to break up the color image into its component RGB parts, and we’ll do that using the RGB MIXER, which looks like a triangle of three dots. First select your BLUE Layer and open up the RGB mixer. Make sure that in the drop-down menu you select MONOCHROME and PRESERVE LUMINANCE (This should already be checked). This will remove some sliders, and make sure you’re working with the true RGB levels of the image only.
Then simply slide the RED and GREEN sliders all the way to the bottom. Leaving you with the BLUE Channel isolated (below). The image itself will look monochrome, but is composed of only the BLUE information in the picture, just like how it would be recorded in a 3-strip technicolor film camera.
And this is the resulting image
Repeat the same technique for the RED and GREEN Channels respectively, pulling down the sliders of the other primary colors (pull BLUE and GREEN out of RED, and pull RED and BLUE out of GREEN), and make sure that MONOCHROME is selected in the drop-down menu for each. Your three nodes should look like this:
Step 2- Creating the CMY “Strips”
Now that we’ve taken our original footage and broken it up into our “3 strip” RGB components, we will now need to tint them with their respective complimentary colors, just like they did with the film. In order to do this, we need to make three more nodes called CYAN Layer, MAGENTA Layer, and YELLOW Layer; then connect the orange dots from their opposite colors. RED-CYAN, GREEN-MAGENTA, and BLUE-YELLOW. See the node structure below.
Once again, we’re going to be using the RGB MIXER for this, but this time make sure that MONOCHROME is OFF, because we’re going to be using a subtractive process to create the tints.
Select the YELLOW layer and the RGB MIXER (again, making sure that MONOCHROME is now off), now what we want to do is to remove all the BLUE information in this node. We do this by simply pulling down the blue slider for each RGB channel in this node, but leaving the RED and GREEN sliders alone.
The image for this node now looks like this, a nice yellow tint over a black and white image.
We then simply repeat the process with the RGB MIXER in the CYAN and MAGENTA nodes by pulling out the complementary colors respectively. Pull out all the REDS in CYAN, and all the GREENS in MAGENTA. See the image below:
One thing we have to reckon with before we move on, is the overall strength of the image. Right now, if we wanted three different tinted versions of the shot, we could stop right here. But our overall goal is to combine these three nodes into one overall image. We now have to adjust the transparency of each of the CYAN, MAGENTA, and YELLOW nodes to show only 33.3% of their intensity, so when mixed together they add up to 100% instead of 300%. Also by adjusting the relative strengths of these nodes, you can adjust the color characteristics of the final image as well.
We do that in the KEY section of Resolve (its shaped like a little key), by modifying the OUTPUT GAIN section to read .333 instead of 1.000. You need to make this adjustment on each of the CYAN, MAGENTA, and YELLOW nodes. See the image below:
Step 3- Combining the Final Technicolor Image
Now we need to combine the 3 strips, and since we can’t physically smash bits together, we’re going to use a parallel node. So create a new parallel node by right-clicking in the node window and selecting ADD NODE -> PARALLEL. Then we need to add two more inputs to the parallel node to make three total. You do this by right clicking on the parallel node and selecting ADD ONE INPUT, twice.
Connect the orange dots of the CYAN, MAGENTA, and YELLOW nodes to the three inputs of the parallel node. Then create a new serial node (option-s) to that and label it WORKING NODE, connect the output of the WORKING NODE to the Output Bar at the right side of the node editor, as in the picture below:
This is the final node structure of the 3 Strip Technicolor look, but it’s going to need some additional tweaking. The first thing you probably notice is that your image is somewhat backwards. The parallel node has equally mixed the CYAN, MAGENTA, and YELLOW nodes together. However due to the fact that we’re using positive images as our source, rather than negative images like used in film, we have this weird in-between stage where the Luma (black and white) part of the image is positive, but the Color part of the image is negative, like below:
This is why we placed the WORKING NODE in line, so we can adjust this, and the final saturation of the shot. Select the WORKING NODE and the COLOR WHEELS interface, and find the control HUE at the bottom. Simply type in 100 instead of 50 (which inverts the colors). Use the Vectorscope and the SATURATION tool in the WORKING NODE to make sure your colors are still legal, and make any final adjustments to contrast, add power windows, etc.
And Voila! There’s your glorious Technicolor image!
Compared to the original image:
Compared to just cranking the saturation all the way up:
At this point your shot is about 80% done, you still need to add a node or two and tweak it depending on your footage to make it look more like film. For example below I took the initial look, and crushed the blacks a little and shifted the shadows into cyan a bit. I then increased the contrast and left the highlights nice and broad. I added just a hint of halation to the whites, and a slight softening to the whole image. The result is below:
Here’s a downloadable link to the 3 Strip Technicolor .DRX file for DaVinci Resolve.
Well I appreciate that my little corner of the web has been discovered, the Technicolor 3-Strip look I created here has been both complimented, and dismissed outright in the same threads. You can find two of the main threads below, and I imagine as the discussion continues that both better, and more complicated ways of trying to replicate this interesting analog process will emerge, and give you some great insight into working with Resolve. Definitely worth a look.
Thread on Creative Cow: http://forums.creativecow.net/thread/277/23054
Thread on Lift Gamma Gain: http://liftgammagain.com/forum/index.php?threads/technicolor-3way-the-correct-way.951/
UPDATE 2: The upcoming Resolve version 10 will have the ability to separate the RGB channels for each node. This will make a three way technicolor look fully operational, rather that this way which is a bit of a hack. I’ll update this page once 10 comes out with a new and improved method.