Re b&...
Could you explain a bit about what made CRT designs have a wrinkly
bed-sheet response?
Based on the essential physics, I wouldn't guess that behavior. I'd guess
the opposite.
OTOH I would for a printer, but not a color CRT monitor.
I would also expect modern matrix displays (LCD, etc) to be more
susceptible by design to wrinkly-bed-sheet response because they are run by
programs (abstract logic circuits) rather than relatively linear analog
amplifiers and timing circuits, and programmers are well-known to get up to
all sorts of tricks.
I don't intend to grind any axe here, I'm looking for further explanation
per your comments re CRT and the movie screen analogy, which I like
/wire
On Fri, Jun 18, 2021 at 15:34 Ben Goren <ben@xxxxxxxxxxxxxxxx> wrote:
On Jun 18, 2021, at 1:09 PM, graxx@xxxxxxxxxxxx wrote:
I could almost get away with only measuring 200 patches?
Indeed, that’s quite likely.
Forget color for a moment.
Let’s say you’ve got a video projector, and you want to correct its
geometry. If everything is perfect, you can project a square grid and every
line will be straight and evenly spaced and all the angles will be right
angles.
Next, assume the optics of the projector are perfect and the screen is
perfectly flat, but the two aren’t perfectly aligned. The projection shows
“keystoning.” Then you just need a near-trivial correction, one for the
horizontal skew and the other for the vertical. This would equivalent to a
matrix color profile.
Now, imagine a stiff screen, but torqued. The math to “de-torque” this
isn’t terribly complex, and is much like a shaper profile.
Finally, imagine you’ve hung a bedsheet up by clothespins. Depending on
how good a job you did, this might not be too bad…but there’s no simple map
that will correct everything. You need a bunch of precise, detailed
measurements of how far each point on the grid is off from where it should
be. This is your LUT profile.
How many points do you need? Well, how lumpy is your bedsheet projection
screen? If it’s nicely ironed and you took good care to get the tension
straight in every direction with lots of clothespins, you don’t need very
many. If the sheet hasn’t been ironed and it’s just dangling by the top two
corners, it might be impossible to build a corrective map.
Early CRTs were wrinkly bedsheets. But modern displays tend to be very
well behaved.
Considering both your measurements had the same error (within rounding), I
take that as a strong suggestion that your display is very well behaved.
Indeed, you might consider building shaper and matrix profiles from that
800-patch set and see how they compare.
Cheers,
b&
