I.
First, just a quick comment that while talking about outresolving
lenses, one should also keep in mind that due to the discretization of
the pixelated media, the effective resolution of a lens can be still
reduced, even if the sensor's pixels/mm number is larger than lines/mm
number for the lens. (The guy in that recent Nikon vs Canon video that
was linked a few days ago in a different thread discusses that.) So,
increase in the pixel density of a factor of 1.28 in principle can
produce some significant effect for some lenses if you are going from
6
MP to 7.7 MP.
In this case, the same argument may apply for the sharpest lenses.
(Maybe even prime * lenses of Pentax?)
II.
With respect to the increase in the sensor's MPs in question, -
if we are talking about the technology advancement, - what is relevant
is increase in the density of the pixels. Indeed, the density
increased
by
a factor of 1.13 (=Sqrt(1.28)).
What is true, is that the higher the starting density is, the harder
it
is to enhance it further. This applies to the recording media density
(as in HDDs), as well as the sensors.
So, from the SENSOR technology point of view, this is a valuable
advancement.
III.
As for practical advantage of having such a sensor in the camera, -
besides the possible one mentioned in I. above, - there might
be some others. Just one example is the possible improvement in
the optical stabilization (again due to discrete nature
of the pixels).
Also, ultimately, the availability of cameras with the
high-resolution
sensors can boost the efforts of producing (and even mass-producing)
higher-resolution lenses, which make the more easily available and
hence
(eventually) cheaper.
But I think the practical advantage (in terms of quality of the
photos)
of going from 5 to 7.8 MP is higher than that of going from 36 MP to
46
MP.
(see I. above).
[Nerd ON]
IV.
Sometimes percentage of the increase might not be a good measure,
while the absolute increase is.
I just wanted to give a few clear examples where the percentage is not
a
good measure:
1. A temperature increase of, say, 2 degrees F (1 degree C).
Does it matter if it from 2 F to 4 F or from 34 F to 36 F?
The corresponding percentages would be 100% and ~6%.
And, of course, in Celcius, the equivalent would be
... well... 100% in the second place (from 1C to 2C), but I am not
even
sure what to say about the first case, when it is from approximately
-16.5C to -15.5C.
So, it obviously depends on the scale!
Well, I agree, this example is not directly applicable to sensors,
whose
dimensions are using the absolute scale. And to make the comparison, I
should've used the absolute temperature scale (Kelvin). In Kelvin,
the changes would be from about 256.5 K to 257.5 K, and 274 K -> 275
K,
respectively.
2. Relevance of the change also depends on the effect that we are
considering.
For many physics processes that rely on the behavior of electrons, the
relevant measure would be the relative change (on the absolute scale,
in
Kelvin). For many biological processes, the direct percentage change
probably would not be a good measure. The absolute change wouldn't
matter
either...
E.g. a change of a human body temperature from 37C (98.6F) to 38C
(100.4) is unpleasant, but the same absolute change from 40C (104F) to
41C (105.8) (which would be smaller in percentage) can be lethal ou.
So, this shows the relevance of the change to the process that gets
affected by this change.
3. Stan, with respect to your "home-economy" example. You are right,
except for those cases, where there is an "offset" of fixed costs.
E.g. if the bare minimum cost of housing is, say $800 a month.
Then for somebody earning $1000 a month, an extra $100 is more
valuable
then $300 for somebody who is earning $3000/mo. The reason is
that the effective increase in the income that available for things
other then the housing would be 50% (from 1000-800=200), and a
comparable relative increase for the second person would be smaller
(from
3000-800=2200 to 2500 is less than 15%).
Effect of the lenses becoming the bottle-neck of the performance is
similar to the "fixed-cost" offset described above.
All these examples are just to support the point that one cannot just
blindly use relative increase (percentage or factor) in all cases.
[Nerd OFF]
Thu Sep 11 09:45:58 EDT 2014
Darren Addy wrote:
I'm not denying that there is an appropriate place to use percentages.
It is especially useful in apples to apples comparisons.
I'm just saying that comparing APS-C to full frame AND to a completely
different era is apples to oranges, in my book.
To go back to Mark's numbers, he's saying a 27% increase is
insignificant except from a marketing standpoint. The new K-S1 is a
25% increase in megapixels over the past several years' 16MP models.
Even ignoring the other technology improvements along the way, I think
that 25% is a pretty significant increase. I can make is sound smaller
by terming it a 1.25 "factor" if I want to minimize it.
Feel free to disagree, but that's my opinion. I think that we may just
be spoiled by seeing the flagship go up 50% from 16MP to 24MP. That's
partly due to the disruption caused by no (really) new DSLR models
during the Hoya to Ricoh transition.
On Thu, Sep 11, 2014 at 8:36 AM, Stanley Halpin
<stan at stans-photography.info> wrote:
On Sep 11, 2014, at 8:23 AM, Darren Addy <pixelsmithy at gmail.com>
wrote:
That's a funny way of looking at it. First of all, almost all change
is incremental, but that doesn't mean it is insignifcant. If
reducing
things to percentage increase was a valid way of comparing things,
then someone who went from bench pressing 460 lbs from 360 lbs
shouldn't be any prouder of the accomplishment than someone who went
from 60 lbs to 76 lbs. It's just a funny way to make comparisons,
unless you are trying to purposely minimize accomplishment.
I.ll leave aside Mark.s point, I don.t know enough to agree or
disagree.
But Darren, your notion of percentages as a bad thing is just wrong.
Lets say I earn $100 an hour. Then I get a $100 raise, am now earning
$200 an hour.
You are earning $1000 an hour, and then you also get a $100 raise. So
you are at $1100 an hour.
We both get an added $100 an hour, but my increase was 100%, yours
was
only 10%. Don.t you think that percentages better reflect the
perceived
value in this case? Ask the buyer of a new $20,000 car how important a
$2000 discount would be. Ask the buyer of a new $100,000 car how
important a $2000 discount would be.
There is a long history of trying to use numbers in various forms to
represent perceived value of one sort or another. Most systems fall
apart because our underlying value systems are not linear and cannot
be
fairly represented with a simple linear scale. Percentages do a pretty
good job capturing some of that underlying non-linearity and I think
Mark.s usage helps to provide a valid alternative perspective on this
"breaking news.. Log scales can be another useful tool.
stan
