Simon Riera Director of Photography
Article published April 2010,
re-printed here by courtesy of Simon Riera... read more from his site at Cinematographer (blog)

Picscan and the RED Camera

For a number of years I have used the Picscan frame leader system with the Telecine transfer of film rushes. The system is simple yet very powerful. An aspect ratio chart and greyscale is photographed during pre production using a chosen colour temperature and ISO. The recorded chart is then used by the colourist to set framing, white, black and gamma levels and neuetralise greys. The colourist should only have to ride the highlights during the rushes transfer to achieve colour and density accurate images. Picscan is very fast and very accurate. If there is a problem with the rushes it is usually due to human error on the set. Picscan allows a DP to know exactly what is on the negative, with no telecine ?saving of the footage and no suprises as regard framing, colour or exposure.

The problem now is that as DPs shoot less and less on film and more and more on digital cinema cameras such as the RED, they want to apply similar telecine controls to the production of their rushes. Especially with the RED camera. The image created for video assist viewing and the creation of rushes with the RED is at best a proxy. A very good proxy, but a proxy non the less. The Redspace or Rec709 colour spaces that are used for viewing and rushes creation straight out of the camera are not necessarily an accurate indicator of what is captured on the RAW file. So how do you create rushes that reflect what is on the RAW file and indeed reflect your photographic intent… You guessed it, Picscan.

Picscan does not help you manipulate exposure of the RAW on set, niether does it address the fact that on set monitoring is still limited to Redspace or Rec709, (I am ignoring RAW viewing). It does however allow the creation of rushes that show the DPs intent. Picscan can be easily used by your DIT on set or via a more traditional post path, with rushed being generated by the post house.

So how does it work? The simplest way is to leverage as much of the free software supplied by RED, in this case RedCine X. The Picscan frame leader which consists of an aspect ratio chart and greyscale chart is photographed during camera prep, following the Picscan instructions which come with the frame leader system. Each chart is shot at a desired colour temp (usually 3200 and 5600) with meters set at desired ISO. The camera ISO is set at its default 320. Meter ISO can be any where between 160 and 320 depending on your interpretation of the REDONEs (generous) ISO rating. Picscan and the RED camera?To shoot the chart for different ISO settings, for instance if you intend to push or pull the image as you would with film, open or close the aperture accordingly and rephotograph the chart.

I like to do this in ? of a stop intervals up to the equivilent of 500 ISO. In that way if I decide to over rate the camera I can do it using the Picscan chart later rather than adjusting the ISO in camera. Adjusting the ISO in camera seems to quickly change the colourimetry of the image, adding colour casts and noise and producing less than satisfactory results. Remember it is all metadata anyway so adusting ISO in camera is not changing any of the information captured by the sensor in the first place.

Once all your iterations of the chart are captured it is time to ingest them into RedCine X (from now on referred to as RCX). You can use it in other programs such as 3CP, however most people have access to RCX. With RCX open you need to set camera colourspace to Redspace and the gamma space to Redlog. I have tried other camera colourspaces such as camera RGB but found the saturation to be too low.

Using a video I/O card the next thing to do is to attach a wave form monitor to RCX. It is necessary to have a waveform showing RGB parade to set up the Picscan charts correctly. I have even used an external monitor with a waveform built into it. Using the waveform, set blacks of Picscan chart just on or above 0, whites just over 600mv and gamma at 280mv-350mv depending on the density of the midtones you want.

Use the RGB parade to balance out any colour casts. Needless to say the monitor you are using through out this setup should be calibrated and profiled. Store the finished chart as a preset in RCX. I use the ISO and colour temp as a reference for the chart… eg Picscan_250T_3200, Picscan_200D_5600 etc… You now have a Picscan stored for each of your ISO/colour temp combinations.

To apply the chart to your footage, import your R3D file from camera into RCX and apply the appropriate Picscan reference. If necessary adjust the highlight (gain) control to bring back detail in highlights. As only highlights are being adjusted, it is quite possible to use just the histogram and your eye although it is ideal to have a waveform monitor connected.

If the colour temperature used to photograph the scene is different from the Picscan standard, it can also be adjusted here using the colour temp slider in RCX to match the camera colour temperature metadata.

Finally output rushes to desired format. I prefer to keep everything in HD and as data for my rushes, however SD DVDs are still preferred by a lot of productions for rushes distribution.

The next challenge with the RED camera will be to apply these looks to the onset viewing path, and to transmit the Picscan information as metadata for the later conform of the R3D files, however for now Picscan goes along way to creating accurate rushes that not only transmit the DPs intent, but allow consistency of image from the set to the screen.

Simon Riera
April 2010