Re: Does the NikonD700 have image stabilization?

OldBoy <nono@xxxxxxxxx> wrote:
"Wolfgang Weisselberg" <ozcvgtt02@xxxxxxxxxxxxxx> wrote in message
news:1ifgp5-lmi.ln1@xxxxxxxxxxxxxxxxxxxxxxxxxx
OldBoy <nono@xxxxxxxxx> wrote:
"Wolfgang Weisselberg" <ozcvgtt02@xxxxxxxxxxxxxx> wrote in message
OldBoy <nono@xxxxxxxxx> wrote:

Assume human tremor of 3.8 degrees/second.
That's 736 pixels per second (3.8/24 times 4670)

Wrong. You don't have a constant angle tremor --- quite to
the contrary. You have a tremor that vibrates within an 24 pix
radius circle.

I disagree on the 24 pixels because you used the object distance which
has
nothing to do with the angle of view.

"Your source has shake within 0.24° [...] 48 pixels across".
Where do you see an object distance?
How would an object distance --- if I had used it --- influence
camera shake outside of macro situations?

You said:
"At an object distance of 4 meters and f/4, you have only 15.64cm
(3.92m-4.08m) "in focus" (diffraction effects apply) with at most 0.0057 (1
pixel size) image circles (i.e. they can spread onto 4 pixels!)."

And the 0.24° and 48 pixels across have *what* connection
with the object distance?

The tremor is erratic (not a mere sinus), I use the worst scenario.

The tremor is a sum of sinuses. Which implies higher and
lower speed parts.

Yes, I take the higher speed part because I use worst case.

Please don't use words you don't understand. That's not a
'worst' case, but some 'average' case, with some interesting
qualifiers to "average".

Your 736 pix/s is, therefore, completely meaningless.

Not as a worst case.

So 2,649,600 pix/hour is a useful statement, in your eyes?

Just as useful as 4 times 3600 pix/s when using your value.

Good that we agree: 736 pix/s is just a smoke screen, trying
to dazzle with worthless figures. (BTW. where does the
4*3600 pix/s come from? Not from me!)

To counteract tremor at 1 pixel you need 1/736 second shutterspeed.

Only if you wrongly assume a constant velocity.

No, for worst case use maximum velocity.

Darn it! You don't grok "worst case"! Please stop using
words you don't understand.

WORST CASE means literally *worst* *case*. Which isn't "1
pixel per 1/736s". If the speed is an *average* of 1 pix per
1/736s and the speed varies then the worst case is WORSE!

You say it's a sum of sinuses, that means no constant velocity.

So let's do a reasonable assumption on an at least believable
'bad case'. Let's stipulate the speed *usually* varies between
1/8th and 8x the average speed. Then the bad case would need
1/(8*736)s or 1/5888s. And that isn't the worst case.

The "good case", 1/(736/8)s == 1/92s is believable, because these
*are* speeds at which sharp photographs *can* be taken, repeatedly.
Not *every* time, but better than 50% of the time.

Do you now get at least an idea about what "worst case" really
means? Do you also start to understand why your "736pix/s"
is not born out in real life?

Point in fact stays that sharp images exist -- and can be repeated
-- with much longer lenses *and* longer exposures, contrary to
your claim.

My suggestion/theory doesn't exclude the _possiblity_ of getting sharp
images.
It's all about getting sharp images _all_ the time.

Yes, as we saw, 1/6.000s is a good start, but won't guarantee
a sharp image *all* the time. (again, please avoid words
like 'all' and 'every time' etc. unless literally *NEVER
EVER* the opposite could or might happen.)

1/100.000.000s is an even better idea.

However, in the real world, limited amounts of photons exist,
and in practice it's much more interesting to know when you
have a 80, 90, or 98% chance for a sharp shot.

A in-camera sensor movement would be 4.2mm (3.8/24 times 26.7)

Sure, and if you drive the 24 hours of Le Mans, you move ~5000
km, and thus are (by your logic) halfway across the Atlantic,
while you don't leave a circle of <7km radius.
[...]
Your 4.2mm is completely spurious. You only need a 0.28mm
movement, which you may need to cross and recross several times.

No, see above.

Your 24h le Mans race ends in the Atlantic. Fancy that.

Again, that's a silly remark.

Again, if you don't grasp the difference between "distance
travelled" and "maximum distance from the starting point
reached", you'll insist on huge travel distances --- which
mean absolutely nothing in that comparison.

If you say 100%, you must proof against even extreme and silly
circumstances. Unless you misunderstand what 100% means, in
which case you shouldn't use words you don't understand.

However, you might want to look at dpreview.com's IS
evaluation. e.g.

http://www.dpreview.com/lensreviews/canon_70-200_2p8_is_usm_c16/page5.asp

I know that test.

and see that at 1/f most images are sharp, and a few slightly
blurred.

I wouldn't call 30-40% a few :-)

Call them what you want, it's proof that your "1/f is wrong"
claim isn't true.

No, the test proofs the 1/f rule is wrong 30%-40% of the time.

Indeed, it does not. It shows that you have a good chance to
get a sharp shot, while not asking for 1/6000s or shorter.
Which is quite important in the real world.

Yes, that's a visual approach, with "sharp" being sharp
at pixel level. Now, how can they produce 60-70% "sharp" images
without IS and at ~1/f, if, as you 'prove' above, they cannot?

If they used the shutter speed I suggest they had 100% sharp images
handheld.

You don't understand the concept of 100%.

Yes, I do: 100 shots taken, 100 shots sharp.

Obviously you don't grasp percentages. It has nothing to do
with 100 shots.

Dpreviews handheld tests misses 30%-40%, that's normal for you?.

As I am interested in limits, yes, that's the expected and
completely normal result. I am *very* interested in using 1/800s
for 100mm, especially when I am at wide open, max ISO of the
camera, already underexposing by a stop or two and 1/80s.

But you probably only shoot in full sunlight.

And even if they had had only sharp images, that wouldn't prove
your theory, because 1/2 or 1/8 of your sugessted exposure time
would yield the same result.

Not "would yield" but "could yield" :-)

Yields for most photographers.

My suggested shutterspeed would give 100%.

No, it wouldn't. Just *one* not sharp shot out of millions
and millions ... and there goes your 100%.

Why don't you advocate 1/1.000.000.000.000.000.000s? That
would be *at least* as good as your 1/800s for 100mm. When
you find the right answer, you'll understand (one day) why
your 100% is not of interest to the photographer.

-Wolfgang
.

Relevant Pages

  • Re: Possible to extract high resolution b/w from a raw file?
    ... pictures with about 12 million pixels of image. ... JPEG images produced by the camera in fact are ... million sensor locations, ... information comes from the green sensor locations. ...
    (rec.photo.digital)
  • Re: TIFF vs JEPG
    ... individual pixels, you will see that some of the pixels are ... I downloaded the two images that Raphael referenced in his reply ... see JPEG artifact squares, and still wasn't sure which image was ... TIFF and good JPEG. ...
    (comp.periphs.scanners)
  • Re: DPI and PPI
    ... Can one change the camera setting of how the images are compressed? ... meaningless 72 ppi number (and the associated 28x21 inch printed size ... because both images are the same size (pixels). ...
    (rec.photo.digital)
  • Re: Ralph Rabbidge and relativity of simultaneity
    ... |>|>| galaxies are not likely to be standing perfectly still. ... |>The images are not well-defined, ... |>has pixels to display them all. ... Mean distance between stars, 10 LY. ...
    (sci.physics.relativity)
  • Re: Scottish Munro Website
    ... images are much larger than they're displayed on the page. ... of your images seems to be roughly 1024 pixels wide on the page, ... but you'll be alienating a lot of people by going wider. ... you should allow for scroll bars. ...
    (uk.rec.walking)