Masking Tips from Dave Corbitt

From TIGwiki2
Jump to: navigation, search

Using Masking with Rank Telecines

The colorimetry of video images derived from film will have errors compared to both the original scene and the appearance of that same film if it were viewed optically via projection. Colorimetry errors are caused by several factors including the slight mismatch of the Pass spectrum of the film dyes relative to the analysis curves of the color separation filters in the cell box.

The color separation filters have to have a wide enough spectral window to get a reasonable signal to noise. The wider these windows are the more contamination each of the RGB signals have from each of the other dye layers recordings. The signal contamination caused by overlapping spectral response of each dye layer was anticipated in the original design of the Rank telecines and in the URSA. Log masking was included in the Rank designs to help minimize these errors but the original masks designed in to the original MK3's are no longer current compared to the errors of current film stocks in common use. With each new family of film stocks a new mask has to be designed to recover the original information recorded by the film.

Fortunately Kodak has been distributing TAF films for several years.

Kodak TAF frame

The TAF films are a useful tool to analyze and correct the dye crossover errors inherent in certain film stocks when analyzed by the Rank optics. Different film stocks use different dyes and have different errors which can be minimized with fixed masking trimmed for that particular film stock.

Below are diagrams showing the typical RGB pattern seen by using the 15 line strobe on a WFM and viewing the Color Bar portion of the TAF film. For this example a TAF on a Kodak EXR Negative stock was used. Your results will be similar on either a MK3 or URSA as the optical analysis filters on both families of telecines are essentially the same.

Errors in RGB without masking

Figure 1

How those errors can be minimized with a custom mask

Figure 2

Vectorscope display without masking

Figure 3

Vector display with masking correction

Figure 4

The visual / subjective results are an opening up of the color gamut on the picture monitor and a more accurate image relative to the original scene as compared to masking off. Now, how do we do this? With the MK3, resistors have to be changed on the Analog Masking boards 100418, -419, and -420 to vary the amount of Log R, G, and B added or subtracted from each of the Log R, G, and B signals. Addition of two signals in Log space with an Anti-Log amplifier later on is equivalent to multiplying one signal by another. This is the same concept that was used for correcting shading errors with the Festival.

With URSA it becomes a bit simpler to use the User masks to create an appropriate correction mask with a TAF film in the gate of a similar stock to the client film you wish to transfer. For this info page I will refer to DaVinci Renaissance since that is the controller with which I am most familiar.

How to set up a user mask

1. Lace up a loop with some Negative TAF on it. Select masking OFF for now. Adjust the PECs, URSA Color Channel levels, and DaVinci Color Channel levels for good RGB levels with no clipping in either highlights or lowlights.

2. Select USER Mask 1.

3. On the left side Soft panel, scroll through the menus by toggling the MODE MORE button until the URSA CONTOURS/ MASKING/ SHADING MODE logos are displayed in the top display panel. The first page of this mode is the CONTOURS PAGE. Push PAGE MORE to get LINEAR/ LOGARITHMIC MASKING PAGE displayed. The left 3 columns will display Linear Masking adjustments and the right 3 columns will display Logarithmic Masking adjustments. I don't think Linear works ( at least on our machines). Logarithmic Mask Adjustments are what we want for now. The R, G, and B channels of your machine probably will look like this drawing if masking is OFF.

Figure 1

With the USER Mask selected it may look like anything. Do not readjust the Color Channels at this time to get levels normal if the USER mask has caused them to change. First adjust the nine pots in the 3 right side columns of the Soft Panel (labeled FROM GRN/RED LOG/FROM BLU etc.) to their center points (8000 on the DaVinci VDU). This should be the same as no masking.

4. Look at the Red drawing again. This waveform has mostly Red information and some Green contamination. To minimize the Green contamination adjust the FROM GREEN knob to bring the amplitude of the 6th and 7th bar more in line with the 2nd and 3rd. Adjust the RED LOG knob to restore the overall amplitude to what it was with NO MASK. You may have to adjust FROM BLUE also to minimize Blue contamination in the Red.

5. The thing to visualize here is the ideal characteristic of a Color Bar display in RGB. Red has 2 High level bars (starting from the left ) (White, Yellow), then 2 low level bars (Cyan, Green), then 2 more high level bars ( Magenta, Red), then 2 more low level bars (Blue, Black). The TAF chart has a desaturated color bar pattern recorded with sequential exposures of R, G, and B filtration. The dark level of each bar does not go down to the base line as it would with an electronically generated pattern. But the relative amplitudes of the high and low level chips for each of the RGB colors should be similar if dye and analysis errors are removed. If you look at Red you can visualize the Green contamination of the pure Red signal by seeing that the 2nd 4 color chips have an overall lower level in both high and low values compared to the 1st 4. Since the Green chips are grouped together in 4 up and 4 down it must be Green contamination in Red causing the errors. Use this type of visual analysis to figure which knobs to tweek to minimize the errors in all 3 film layers. (If this explanation is not clear let me know so it can be revised.)

Once you have created an accurate mask save it in User 1 or User 2. Now toggle through the different masks and return to the saved mask. It should return to the values you have just recorded. Do not readjust a mask once it has been set. Writing down the numerical values of each knob is a simple way to insure no one has tweeked your settings to oblivion. The settings are stored in the URSA not the Da vinci. There are 2 User memories for Positive and 2 for Negative. That's it.


Created by Dave Corbitt

Edited by David Tosh - dlt at earthlink.net