Rob Harrison wrote:
> C.H., I also found those comparison images very interesting. Makes you want
> to ask--"and the big deal is....what?"
The big deal is that the F30 achieves it's results with a sensor that is
7.6x5.7 mm - 43.3 sqmm, while the 4/3 sensor is 18x13.5 mm - 243 sqmm.
The E-1 sensor is over twice as big in each linear measurement and 5
times as big in area.
So the speculation is about what kind of results might come from scaling
the Fuji sensor technology us to 4/3. Direct scaling at the same 6.3 mp
would result in sensor sites over 5 times as large. Scaling to about 12
mp would result in sensor sites about 2.5 times as large as in the F30.
Since sensor site size determines the size of the signal from each
pixel, larger sites result in lower noise, all other things being equal.
Since the F10/11 already far surpassed the noise performance of other
cameras with similar size sensors, and the F30 widened the difference,
it's easy to speculate that a 4/3 sensor using this technology might
result in a 4/3 camera with noise comparable to say the 30 D or even 5D
- a very big deal indeed to many buyers.
It is clear, as you can see in CH's examples and in also tests I've done
and posted some time ago, as well as dpreview tests, that the Fuji P&S
cameras with very low noise at high isos achieve that feat partially
through noise resolution processing in the camera. That's nothing new,
all the manufacturers of this class of camera do that. But Fuji clearly
does it very well, especially compared to some like Panasonic, and, it's
sensor system may just be a great deal better too.
The thing open to debate is how much of Fuji's performance advantage is
actual low noise signal off the sensor system and how much is just the
cleverest software engineers reducing noise more effectively with less
effect on detail than anybody else.
If it's all software ledgerdemain, then a scaled up 4/3 system would be
quite good, with the larger sensor sites allowing the NR to be turned
down. Even that might mean a 10+mp sensor with less noise than the E-1.
If it's mostly superior sensor system before NR, then the scaled up
system could be revolutionary, suddenly catapulting 4/3 into competition
with FF sensors in this area of performance.
It seems CH's peering at his samples leads him to believe it's mostly in
the NR. I think others here at least partially agree. Peering at my
samples, especially at the F10 vs. the F30, lead me to believe that a
significant part is in some kind of improvements in the sensor/processor
independent of NR. The F30 improvements at iso 400 and above seem to
involve only a very little less noise, but significantly increased
detail retention, something like one stop worth, as well as better
shadow retention.
> Except for the fact that the F30 is
> 25% of the cost of the E-1 and lens and has twice the pixels
Huh? E-1 is 5.1 mp and the F30 is 6.3, a difference that's not really
significant in potential resolution.
> , when viewed side by side like this the noise doesn't seem like it's a real
> issue.
My reading of CH's samples is:
- At all isos, E-1 captures subtle color graduations in wood grain and
upholstery that the F30 does not.
- 100 & 200 - F30 sharper/more detailed both near and far, noise not
significant in either.
- 400 - Crossover. E-1 has more noise, but detail rendered is close to
identical.
- 800 - E-1 has more noise, but sharpness/detail of F30 now less than E-1.
- 1600 - E-1 pretty noisy, but detail/resolution still pretty good. F30
noisier and, worse than that, has visible artifacts at tonal
transitions. E-1 noise can be reduced considerably without much loss of
detail and with no artifacts
This wouldn't mean much if these were comparable cameras. In addition to
being much cheaper, as you point out, and using a much smaller sensor,
the F30 is 10% of the volume and 18% of the weight of the E-1 with
comparable lens, the 14-45. That makes it very interesting!
Moose
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|