[opendtv] Re: News: High Dynamic Range imaging
- From: Jeroen Stessen <jeroen.stessen@xxxxxxxxxxx>
- To: opendtv@xxxxxxxxxxxxx
- Date: Tue, 8 Aug 2006 10:47:47 +0200
Hello,
Bert Manfredi:
> I can see the point. You don't need the full dynamic range for any given
> scene, so it's best to shift the quantization as a function of average
> image brightness (or some other similar strategy). Equivalent to dynamic
> compression/expansion for audio.
But note that if the method is used only for defining the quantisation
table then there will be no visible change in brightness. Only the
quantisation errors would dynamically vary over the scenes.
> In audio, sometimes this creates audible artifacts, as the playback
> system constantly readjusts to the sound levels. I think they call it
> "pumping." It's basically a control system with a feedback loop, which
> can become less than linear and stable if the feedback is shifted in
> phase wrt the input. I wonder if there wouldn't be similar risks with a
> video equivalent?
Not if the quantisation table is transmitted with the video data and
then used for a perfect inversion. Only the fixed gamma functions are
replaced with a dynamically variable quantisation table.
Of course, this could also be combined with a more intelligent (MPEG)
compression algorithm, which is also all about quantisation. I don't
know if you could design a compression that can work well in the linear-
light domain, if you feed it -say- a histogram of the pixel values.
(As was suggested by Tom Barry.)
> Perhaps, the idea of *not* doing what you and Tom suggest is that, given
> a good display of course, this allows any individual viewer's eyes to
> react naturally, without making assumptions on how fast or slow that
> individual's vision adjustment is?
That would make the display unnecessarily expensive. If you know that
the high dynamic range is going to get wasted before it reaches your
brain then you can waste it earlier and not spend good money on it.
> Me, I wish my LCD with not-very-adequate 600:1 contrast ratio (dynamic
> range) would amplify the dark scenes. Seems to be a problem with LCDs,
> at least of the vintage of mine. I'd be satisfied with a static dynamic
> compression, actually.
600:1 is not a bad display. It gets better with dynamic backlight dimming,
of course. This is all described in our recent SID (2006) paper:
Session 26: Power Savings in Backlights (Display Systems)
26.4: Algorithm for Contrast Reserve, Backlight Dimming, and Backlight
Boosting on LCD Page 1249
J. H. Stessen (Philips Applied Technologies)
J. G. R. van Mourik (Philips Consumer Electronics)
The .pdf file is 1.23 MB. Ask me for a copy, but not everyone please !
Regards,
-- Jeroen
+-------------------------------+------------------------------------------+
| From: Jeroen H. Stessen | E-mail: Jeroen.Stessen@xxxxxxxxxxx |
| Building: SFJ-5.22 Eindhoven | Deptmt.: Philips Applied Technologies |
| Phone: ++31.40.2732739 | Visiting & mail address: Glaslaan 2 |
| Mobile: ++31.6.44680021 | NL 5616 LW Eindhoven, the Netherlands |
| Skype: callto:jeroen.stessen | Website: http://www.apptech.philips.com/
|
+-------------------------------+------------------------------------------+
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:
- Using the UNSUBSCRIBE command in your user configuration settings at
FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word
unsubscribe in the subject line.
Other related posts:
