Thread Subject: Re: Photo-Seizure Language
This archival content is maintained by WebAIM and NCDAE on behalf of TEITAC and the U.S. Access Board . Additional details on the updates to section 508 and section 255 can be found at the Access Board web site.
From: Gregg Vanderheiden
Date: Thu, Jul 26 2007 7:55 PM
- Return to this mailing list's archives
- View all messages in this thread
- Next message in thread: None
- Previous message in thread: Gregg Vanderheiden: "Re: Photo-Seizure Language"
- Messages sorted by: Author | Thread | Date
Some problems with this - including divergence from international
Will contact you offline with gory details. We also need to address the
non-content hardware aspects per our action item.
Gregg C Vanderheiden Ph.D.
From: = EMAIL ADDRESS REMOVED =
[mailto: = EMAIL ADDRESS REMOVED = ] On Behalf Of Thomas Albin
Sent: Thursday, July 26, 2007 11:27 AM
To: TEITAC General Interface Accessibility Subcommittee; TEITAC
desktop/portable (hardware) subcommittee; = EMAIL ADDRESS REMOVED =
Subject: Re: [teitac-general] Photo-Seizure Language
I've taken a cut at a draft of language to address the Photosensitivity
issue and have also attached some comments to support the draft language.
I tried to organize it by presenting the current draft language, then my
proposed language (in italics), then my comments on the current draft
language and in support of my proposed language.
We still lack a definition of red flash; while it is important, it is
difficult to know what more to say at present other than it is important.
Comments would be appreciated!
* Products must not flash more than 3 times in any one second period,
* Flashing created by software is under the general flash and red
My proposed re-write
Products shall not flash with a frequency greater than or equal to three
times per second or less than 55 times per second (3 Hz ? flash frequency <
55 Hz), unless:
* Event does not meet the general definition of a flash
* Flash avoids saturated reds
General definition of a flash
* An event where there are opposing changes in luminance of magnitude
greater than 20% of screen luminance that occur within a time period of 280
ms (an opposing change in luminance is an increase in luminance equal to or
greater than 20% of screen luminance followed by a decrease of similar
magnitude, or vice versa), and
* The event has a modulation depth of 10 % or greater, and
* The event contiguously occupies more than a total of .006 steradians
(25% of any 10 degree visual field).
Flashing created by hardware
* Shall not meet the general definition of a flash
NOTE: High contrast between text and background is necessary for
legibility, particularly for older individuals. Technical standards such as
ISO DIS 9241-303 recommend contrast ratios of at least 3:1 between text and
background. High contrast text presented on a screen that is being
refreshed periodically is explicitly excepted from the definition of a
1. There is no definition of general flash threshold or of red flash
threshold. Consequently the statement regarding red flash threshold cannot
be evaluated at present. I believe that W3C has discussed a red flash
threshold, but without defining what is meant by saturated red. It's not
clear that that discussion is what is intended here (or that it's even
appropriate). For example, if every pixel on a screen was at R = 255 (fully
saturated red), does a change putting every pixel at R = 244 constitute a
red flash? The epilepsy consensus document simply cautions against
transition to or from "a saturated red."
2. There should be an upper limit on the frequency, e.g. the statement
regarding flashing should read something like:
- Products shall not flash with a frequency greater than or equal to three
or less than 55 times per second (3 Hz ? flash frequency < 55 Hz), unless:
* Flashing created by hardware is either:
* greater than YYY candelas; or
1. There doesn't seem to be anything in the photosensitivity research
literature that supports a limit on the maximum luminance of the type
suggested here, e.g. less than YYY candelas. High luminance displays are
common, especially when intended for use in bright environments, such as
Some references in the photosensitivity literature (e.g. Binnie et al) do
suggest that a difference in the intensity of luminances of 20 cd/m2 or more
between the flicker luminance and the screen luminance may be important as a
trigger for seizure activity; however, it is important to note that the
criteria is for opposing changes in luminance intensity of 20 cd/m2 or more;
that is, there has to be an increase in luminance of 20 cd/m2 followed by a
decrease of 20 cd/m2 or vice versa. The Binnie et al reference that appears
to be the original source for this value was done on a screen with a maximum
luminance of 100 cd/m2. This suggests that the threshold for flicker
intensity is about 20% of the screen luminance rather than an absolute 20
It is also important to note that Binnie et al put a time constraint on this
in addition to the flicker intensity; that is, the opposing changes must
occur within 7 frames on a 50 Hz TV screen (I believe that is about 280 ms)
or about 4 Hz.
* greater than 20 candelas/sq meter and contiguously occupies more
than a total of .006 steradians (25% of any 10 degree visual field).
1. The definition for flashing created by hardware is flawed. As written, it
can be interpreted to not allow any current CRT monitors with refresh
frequencies of 3 Hz or greater to have average luminances greater than 20
cd/m2. Unfortunately, that is nearly every monitor on the market. At least
some flat panel displays would also likely be adversely affected by this
flicker limit of ? 2 Hz with luminance greater than 20 cd/m2, e.g. some of
those intended for viewing movies.
There are other references that suggest that the difference in flicker
luminance contrast should be limited, e.g. the lower luminance should be no
less than 80 percent of the greater luminance. Binnie et al refer to this
in a discussion of modulation depth of the flicker. Modulation is defined
as the difference of the higher and lower luminances divided by the sum of
the two ((Lh - Ll)/(Lh + Ll)) x 100%. They cite a study (Harding and Fylan)
that showed that flicker with a modulation depth of approximately 10% or
less did not appear to constitute a risk for photosensitive seizure. Binnie
et al use this to define a "potentially harmful flash". Binnie et al define
a harmful flash as " 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 cd/m2 or more (see Notes 1 and 2). This applies only when
the screen luminance of the darker image is below 160 cd/m2". Note the
previous comment that the 20cd/m2 value for opposing changes may be in fact
20% of screen luminance rather than an absolute value.
In an earlier comment, Binnie et al clearly indicate that the 160 cd/m2
restriction on the darker luminance is further restricted to a situation
where the peak luminance of the screen is 200 cd/m2, referring to ITU-R
BT.500 and stating that "This specifies a peak luminance value of 200 cd/m2
for a reference home viewing environment. There is an upper threshold of
160 cd/m2 at or above which the darker image of a cycle of flashing images
is acceptable, as this limits the contrast to levels below those at which
PSE is a risk".
If we calculate the modulation depth for a flicker on a screen whose
greatest luminance is 200 cd/m2 and whose darkest luminance is 160 cd/m2 (?
20% absolute difference in luminance), we find the modulation depth to be
approximately 11%, which is consistent with the Harding and Fylan study
Finally, we need to address the issue of contrast ratio. Contrast ratio is
defined as the ratio of the highest screen luminance to the lowest (Lh/Ll).
Technical standards such as ISO DIS 9241-303 specify minimum contrast ratios
for electronic displays of approximately 3:1 or greater, noting that higher
contrast is of great utility when reading, particularly for older
individuals. A contrast ratio of 3:1 is equivalent to a modulation depth of
0.5, much higher than the photosensitivity threshold suggested by Binnie et
For example, if not changed, the current PSE limits could be interpreted as
forbidding the display of type with a 3:1 contrast ratio on a screen that
was refreshing at a frequency of 100 Hz.
On Jul 1, 2007, at 11:26 PM, Gregg Vanderheiden wrote:
We are supposed to come up with language for general coverage of seizure
We have agreed to follow the Web language (except tweaked to fit 508 -
e.g. imperative voice and use of "must")
For hardware we talked of restricting this to only severe problems so that
LEDs etc are clearly NOT covered.
The approach below eliminates all small flashing objects unless they are
VERY bright (like a flash tube) - or very large ( so it would be comparable
to the software provision).
If this looks OK we can go with this as a placeholder for our next draft and
work to figure out the proper YYY value and language.
Product software must not flash more than 3 times in any one second period
unless the flash is under the general flash and red flash threshold.
Product hardware must not flash more than 3 times in any one second period
if the flash either:
* is greater than YYY candelas; or
* is greater than 20 candelas/sq meter and contiguously occupies more
than a total of .006 steradians (25% of any 10 degree visual field).
Gregg C Vanderheiden Ph.D.
Professor - Depts of Ind. Engr. & BioMed Engr.
Director - Trace R & D Center
University of Wisconsin-Madison
< <http://trace.wisc.edu/> http://trace.wisc.edu/> FAX 608/262-8848
DSS Player at http://tinyurl.com/dho6b
If Attachement is a mail.dat try <http://www.kopf.com.br/winmail/>
- Next message in Thread: None
- Previous message in Thread: Gregg Vanderheiden: "Re: Photo-Seizure Language"