Tuesday, January 20, 2009

Ofcom and PSE - there is no secret sauce Mr Harding!

The issue of Photosensitive Epilepsy comes up often in television - back in the late nineties there was an episode of Pokemon that featured flashing images that provoked kids in Japan to have seizures. Since then Ofcom have been very keen to avoid this on British television and since 2003 have produced the following guidelines;

OFCOM Guidance(extract).pdf

This is an extract from the document but it does include the important details which hinge around the following;

3. A potentially harmful flash occurs when there is a pair of opposing changes in luminance (i.e., an increase in luminance followed by a decrease, or a decrease followed by an increase) of 20 candelas per square metre (cd.m-2) or more (see
notes 1 and 2). This applies only when the screen luminance of the darker image is below 160 cd.m-2. Irrespective of luminance, a transition to or from a saturated red is also potentially harmful.

3.1.1. Isolated single, double, or triple flashes are acceptable, but a sequence of flashes is not permitted when both the following occur:
i. the combined area of flashes occurring concurrently occupies more than one quarter of the displayed (see note 3) screen area; and
ii. there are more than three flashes within any one-second period. For clarification, successive flashes for which the leading edges are separated by 9 frames or more are acceptable, irrespective of their brightness or screen area.

These parameters are well defined and so anyone who can understand them can build a PSE detector that will with certainty detect when a violation occurs. This is how the situation should be as it avoids any one manufacturer of test equipment having a monopoly. Unfortunately this is just the situation that has nearly developed with the Harding FPA detector. Their machine is a PC with SDi capture card that you digitise the video sequence to be checked into and it runs an analysis. The other popular unit is the GordonHD which is more like a traditional piece of equipment in that it sits in the signal chain and gives an alert when it sees a violating sequence go past.
We have a couple of customers who like the idea of realtime performance that the Gordon gives and don't like having to capture (the Harding doesn't support standard codecs so no Quicktime reference export from Avid!), analyse (in slower-than-realtime) and then get a report - only to repeat it all after you've corrected the offending clips (because the broadcaster likes to see a full 10:00:00:00 - 10:54:00:00 report!). The Gordon on the other hand is cheap (£3k against £13k) and just sits there taking a feed of HD/SDi video and Timecode and firing a GPI when a violation is detected (and even entering the TC into a file) which means you can have it hanging off the Avid (or whatever) and the editor can rock'n'roll over a piece of footage adjusting his edit point over the flash frames (it's mostly paparazzi footage with all those camera flashes that cause it) until he gets a sequence that doesn't cause a problem.

Anyhow - you can tell which machine I think is best. Harding is a great self-publiciser who gives you the idea that he alone knows the secret-sauce of PSE. The guys at Tektronix tell me it's on the way as an upgrade for their WVR-series 'scopes but they are worried that Harding has all the patents stitched up.

Anyway - a quick once around pals revealed the following;
Ascent Media check all of five's output (including the two daughter channels) on a GordonHD, ITV's QC department at Upper Ground use a Gordon as their first-pass analyser and Channel Four specify it as well. However, talking to everyone in facilities reveals that they (almost) universally believe the Harding to be the only machine capable of doing the job.
I had several earlier-model Gordon's at Resolution and they were superbly fitted to the job (and cheap enough to have one in every suite). We never had a tape sent back that had gone through it and that included many more hours of terrestrial television than most facilities ever turn out (including quick turn-around stuff with lots of potential trouble - think the Friday night eviction show for Big Brother).

John Emmett of BPR is a gentleman of the old school (they make the Gordon) and you can find an interesting paper on the subject of PSE he co-authored here.


Saul said...

Well said Phil!

"However, talking to everyone in facilities reveals that they (almost) universally believe the Harding to be the only machine capable of doing the job."

This'll be because facilities have to deliver to the BBC, and the BBC specifies that it uses a Harding for Tech Review ... so those facilities are right in a sense because the Harding box is the only machine capable of telling you with 100% certainty whether you'll pass through another Harding box!

The problem is that whilst luminance flashes are well defined and therefore easy to detect without a harding, the "sustained failures", "spatial patterning" and "red saturation" errors are not as clearly defined and therefore do constitute a kind of "special ingredient" ... at least until someone else is willing to copy Harding's formulas exactly (if they are even publicly known).

I've lost count of the number of times we have been caught out by seemingly harmless stuff ... people waving, shots of traffic in daytime, a cake with candles, stirring ingredients in a pan, a locked off shot of a harbour, Roman columns, even the old "Grange Hill" opening titles trigger it off (and how many millions of kids must have seen them without any ill effect?)

You're entirely right that this is an unnacceptable monopoly, particularly for something which is all about protecting the public ... it is particularly surprising that the BBC (usually so opposed to proprietary systems) is upholding it by specifying a Harding in their standards document.

Good news if Tektronix are going to incorporate one into their scopes though ... that will stop us having to wheel the Harding from building to building!

As a final note, the Harding we have can do real-time analysis ... but the price really is a killer.

Phil Crawley said...

So the question is; Who decided to grant Harding the right to define the delivery standard over Ofcom?

If that is the case then where can we find the standard?
Nowhere - Harding guards it as a trade secret!

Saul said...

Hi Phil,

I just did a Google for "Ofcom PSE guidelines" (to print them off for a colleague) and this page came top!

Good work, I only hope that others stumble across this page in a similar fashion!

Anonymous said...

Hey Phil,
Apparently Harding do have a file based system now so you can analyse things like quicktime ref clips on it.

hks1966 said...

Digging up an old blog here but here's my tuppence. The Harding PSE analyser is seen as an inconvenience in the world of post production. We only bought one because We had to! NOT ONCE have I heard anyone say that this test is a good idea and PSE sufferers will reduce their exposure to PS hazardous material. It is seen as an expensive inconvenience which holds up deliveries. Red buses, red tiled roofs, fences, etc. A quick fix for most fails is to reduce the brightness in the scene by a few percent. Other fails require more work by re-editing.

Is it just me or does anyone else find the BBC coming down harder on external suppliers and give themselves more slack by letting issues go by they would normally condemn?

Thad Beier said...

From what I have read, though, the incidence of PSE from TV has decreased by 75% since it has been required that content be checked -- so there is some value to the inconvenience.

Phil Crawley said...

So this is an entirely un-scientific response but I asked a friend who works in A&E if he had EVER seen a hospital visit due to film or TV-induced PSE and he hadn't, nor did he know of any colleagues who had.
I'd be interested if anyone had any real-world statistics relating to TV-induced PSE?