Friese-Greene Colour Process

Colour movies using a single filter and no specialist projection equipment...

This was an attempt to simulate the two-colour photographic process used by Friese-Greene for cinema film, shown in Dan Cruickshank's 2006 BBC TV Series The Lost World of Friese-Greene. I was intrigued by this series, because of its promise of colour cinematography from an era usually associated with black and white images. His system was quite ingenious, amazingly using just one coloured filter in recording - the camera captures alternate images through a red filter on black and white film. These images are tinted red on the print of the film. The other images are unfiltered and then tinted blue-green (cyan), to complement the red. When the projected film is viewed, the brain sees a colour image, presumably through the 'persistence of vision' effect. This process requires no specialist projection equipment and is capable of producing realistic reds and whites, which must have been quite novel for the time. The drawbacks include flickering images, unrealistic blues and greens, coloured fringes for moving objects and the need to film in bright light.

First I took a colour picture, of my son,  chosen for his red rugby shirt, blue trousers and green background. (Although he was delighted to be wearing the Wales 'Grand Slam' shirt, he wasn't too thrilled about having his picture taken).

Using photo editing software, I converted it to a black and white picture, filtered in red, to simulate a picture taken in black and white through a red filter. This is the first photo taken with my virtual F-G camera!

Then I tinted it red to produce the first image in the finished film.

Next, I converted the original picture to black and white, but this time with no filtering. This is the second virtual F-G photo.

Then  I tinted it blue-green (cyan), to produce the second image in the finished film.

I then merged the red and cyan images together to produce a two-colour picture.

The  reds are faded, because red light in the original also contributes to cyan in the final image. Also, skin tones look slightly green.  Increasing red saturation reduces these problems, but does reduce the colour of the grass, as shown below. There is no discernible blue at all in either image.

If a scene contained a large expanse of blue, for example a cloudless sky, then it should be possible to reproduce this more accurately, by replacing the cyan colouring on the developed film with a colour having a higher blue content. Similarly, for large expanses of green (fields or trees), a colour with a higher green content could be used. I wouldn't be surprised if this method was used by Friese-Green, because the blues and greens are quite convincing in such footage of his that I've seen. I tried this here, tinting the unfiltered black and white image picture green, not cyan. The grass looks more realistic, but the green cast to the skin is more pronounced.

Alternative ways of combining the red and cyan images:

1. Simulate the projection method used by Friese-Green. To do this, I used Windows Movie Maker to create a clip. The minimum duration was 1/8 second. I played it back with Windows Media Player using fast forward and repeat. There was too much flicker for prolonged viewing, but there was an indication of colour. 

2. View cyan and red images side-by-side as you can for stereo images. This doesn't work at all! It just results in a black and white image! So, this method could not be used to produce stereoscopic colour slides. This shouldn't have surprised me, because when you see 3D images using red and green glasses, the image always appears black and white.

So, this is an interesting process. Using just one filter on the camera can give quite convincing results, especially with some forethought in setting up the 'shoot' and in preparing the film afterwards. It was of course, a technological dead-end. Once the hurdle of creating a usable and affordable 3-colour process had been cleared, then this system and others like it would have become redundant.