# The chances Bayer gets 1 pixel color right

Discussion in 'Digital Photography' started by George Preddy, Feb 10, 2004.

1. ### George PreddyGuest

In a 6M recorded output pixel image, the chances are 1 in 11,333 that a
single pixel in a Bayer image is the correct color.

George Preddy, Feb 10, 2004

2. ### perGuest

What if it's 5M?
/per

per, Feb 10, 2004

3. ### DomingoGuest

---

Then I'm one of the luckiest people on the planet - my Olympus turns out
consistently good images, all with faithful colour reproduction.

Statistically, the odds against this happening must be astronomical.

Maybe I should start doing the Lottery?..

Domingo, Feb 10, 2004

5. ### bobGuest

What a silly statement.

Given any recorded output pixel from any sensor, the odds are exactly 0
that it is the correct color.

When you make a print from said pixel, the odds are 0^2.

When you view the image under any given light source, the odds are 0^3.

What difference could it possibly make?

Unless you happen to be a scientist doing research on recorded color
values, the only relavant question is, "given any random photograph
(without anything to compare it to) do you like it?"

A absolutely adore my boxes of Kodachrome.

Bob

bob, Feb 10, 2004
6. ### Simon StanmoreGuest

Simon Stanmore, Feb 10, 2004
7. ### DurumGuest

"George Preddy" wrote

FYI, you've made 3730 postings so far with no noticeable effect as far as I
can see. Why continue?

Durum, Feb 10, 2004
8. ### Dan DunnGuest

Have you devoted your life to running down any equipment which you do
not own?

Why don't you try adding some (respectable) references to back-up your
drivel?

Personally, I think you're pathetic.

Dan Dunn, Feb 10, 2004
9. ### browntimdcGuest

And what are the chances of a single dye cloud in film to have
the correct color?

Tim

--

"The strongest human instinct is to impart information,
and the second strongest is to resist it."

Kenneth Graham

browntimdc, Feb 10, 2004
10. ### eawckyegcyGuest

The Preddy has discovered the binomial distribution, and has, without
surprise, made negatively intelligent use of it.

http://members.aol.com/iandjmsmith/BINEX.HTM

Or you can just approximate it with a simple multiplication:
n=6000000, p=1/2^36 (p=1.455191522836685e-11), then the chance of a
single 'correct' guess in the field of 6000000 attempts is
approximately n*p = 8.731149137020111e-05, or about 1/11453. (Compare
to the Preddy's result.)

Now the Preddy of course has forgotten that a bayer sensor records
1/3rd of his assumed 36 bits/pixel (a fact that it has admitted many
times in the past), so the chance of "randomly" guessing the remainng
bits is is actually p=1/2^24, using his model of the bayer sensor. If
you grind the arithmetic the probability it speaks of is actually
about 1/4, not 1/11500 or so.

But hey, this is Preddy Probability: it _must_ ignore prior knowledge
and the physical reality that presents it. If it did otherwise then
it would not be the negatively intelligent entity it is.

eawckyegcy, Feb 11, 2004
11. ### Rudy GarciaGuest

What pathetic drivel. Get a life.

Rudy Garcia, Feb 11, 2004
12. ### George PreddyGuest

I've been though more Bayers than I can count on two hands. Most were more
expensive than my SD9. Only the Foveon outputs quality images.

George Preddy, Feb 11, 2004
13. ### gsumGuest

So. if you couln't get decent images from Bayer, why did you

Do you know how to use a camera?

Graham

gsum, Feb 11, 2004
14. ### Alan D-WGuest

I doubt you can count that far Georgy Boy.

Alan D-W, Feb 11, 2004
15. ### George PreddyGuest

I was naiive enough to believe the reviews weren't about pushing mass market
junk. It took a while, someday you'll get there too.

George Preddy, Feb 11, 2004
16. ### gsumGuest

Good point. It has become evident that even the Internet
review sites such as DPReview are in the claws of the manufacturers
and cannot be trusted. But exactly the same hype applies to Sigma.

Graham

gsum, Feb 11, 2004
17. ### George PreddyGuest

Bayer output is interpolated, sorry. You were lied to when you bought it.

George Preddy, Feb 11, 2004
18. ### George PreddyGuest

Film always displays the color it senses. Bayer never does. Bayers cannot
display a sensed color, they guess all of the pixel colors in every
interpolated (guessed) image.

George Preddy, Feb 11, 2004
19. ### Bart van der WolfGuest

Wrong, each processed color layer has a single subtractive color dye.

Bart

Bart van der Wolf, Feb 11, 2004
20. ### Simon StanmoreGuest

Hmmm - close top zero chance of that. The only films that have a chance to
are positive (slide) films of course, and each of these renders colour and
contrast differently. Same as for Foveon and CFA sensors really

Simon Stanmore, Feb 11, 2004