An+examination+of+source+material

The below images show magnified sections taken from two different sources of the same individual frame of video material. Fig.2.1(a) shows a section taken from a VT copy of original PAL source material. Fig. 2.1(b) shows a broadly similar section taken from a high-definition scan of a 16mm film recording negative.


 * [[image:image001.jpg]]©BBC || **Fig. 2.1(a): Section from original video** ||
 * [[image:image002.jpg]]©BBC || **Fig. 2.1(b): Section from film recording** ||
 * [[image:totp_sd_9_31_pal1_adj_patch.jpg]]©BBC || **Fig. 2.1(c): A fresh attempt at PAL encoding the colour frame depicted in 2.1(a)** ||

In fig. 2.1(a), there are contributions from around 100 lines of video. Half of the lines originate from the odd field and the other half originate from the even field. The lines from each field are interleaved in the image due to the effect of interlace. The ‘comb’ effect that can be seen along the top edge of the index finger portrayed in the image is due to the fact that the fields are separated in time by 1/50th second, and the object had a component of motion (vertical in this case).

In fig. 2.1(b), the same combing effect can be seen – showing that the film frame is also composed of contributions from two discrete fields. This also shows that the film recording system possessed a resolution sufficient for this level of detail to be captured. Also in fig. 2.1(b), a pattern of finely spaced dots can be seen running horizontally along the length of the lines. These are seen due to the fact that the film has captured **//luminance + chroma//** and not just the **//luminance//** part of the PAL signal.

Fig. 2.1(c) shows a fresh attempt (courtesy Richard Russell) at PAL encoding the colour frame depicted in fig. 2.1(a). This is one of four possible variations that arise from the PAL 8-field sequence, and I have shown the version that seems to best match the film recording frame. It can be seen that there doesn't appear to be much difference between this image and that scanned from the film (fig. 2.1(b)).

The chroma signal in PAL is a quadrature modulated signal derived from U and V (the colour difference signals). The sub-carrier used for colour modulation is a very precise 4.43361875MHz. Therefore, the horizontal spacing between dots represents exactly 2.2555 x 10-7 seconds (within an area of consistent hue).

The level of ‘presence’ of the dots is a measure of the saturation of the original colour. Broadly speaking, an area where no dots can be determined indicates an area where colour was weak or absent, and, an area where the presence of dots is very strong indicates an area of highly saturated colour.

Film recorders used a technique called ‘spot-wobble’ to obliterate the line-structure of the filmed monochrome CRT, but still retain vertical resolution. In effect, spot-wobble broadened the scanned path of the line on the CRT. Judging from fig. 2.1(b) spot-wobble does not appear to have had much of a detrimental affect upon the presence of the chroma dots on film.

A former member of the BBC staff, who was involved in the BBC film unit, informs me that a very slow film was used in the construction of film recordings. Something like ASA 3 [|film speed] was used. Such film stock had a very fine grain - meaning that it was capable of capturing very fine detail, and that the structure of the grain should not interfere unduly with the depiction of the embedded chroma dots. A specification for the film stock used can be seen here: A couple of film-recording reports, from the FR negative cans, can be seen here: Interestingly, the first relates to an FR **without** chroma-dots and the second relates to an FR **with** chroma-dots. The signature on the form is the same, but two different film recording channels were used. No.23 for Ep1 and No.24 for Ep2. Perhaps use (or otherwise) of a chroma notch filter was dependent upon the channel used rather than the operator?

Explanation of the reports (courtesy of Brian Pritchard): //"////The Sheets show the following information:// //Process to Instruction K6A would specify the Gamma to which the negative should be processed.// //Negative:// //Eastman Television Recording Film 7374 Emulsion Number 541 Camera Number 23 Density of optical track: 1.30 1.37 (Inside and outside of track) Density of clear area of track; Fog 0.25 The Level 1 to 8 are density readings of a grey scale so that level can be checked and gamma can be calculated.// //Positive:// //Eastman Fine Grain Release Positive 7633 (Grey base) Emulsion Number 857 Picture Printer Light 15 Exposure setting on the printer ( Probably a Lawley or Model D Printer) Printer Number 1 Track Density: 1.28 (aim 1.4) Track Printer Light: 20 Exposure setting for exposing the optical track.// //Track Fog Level: 0.28// //Track Printer Number 1// //The Level 1 to 8 are density readings of a grey scale so that level can be checked and gamma can be calculated.////"//

Jonathan Wood, who has operated telecine equipment within the BBC for many years, says that the vast majority of film recordings that pass through his hands that were made from a PAL source exhibit embedded chroma. This would suggest that there would be considerable benefit to be had in the development of a technique to extract and employ the colour information usefully.

-- James Insell 06/03/2007