Sign in with
Sign up | Sign in
Your question

Even I'm impressed...

Last response: in Digital Camera
Share
Anonymous
February 18, 2005 9:01:47 PM

Archived from groups: rec.photo.digital (More info?)

Sheesh, I'm going to have to stop badmouthing the lack of fast lenses for
the E300!

The ZUIKO DIGITAL ED 14-35mm (35mm equiv: 28-70mm) 1:2.0 and ZUIKO DIGITAL
ED 35-100mm (35mm equiv: 70-200mm) 1:2.0? are understood to be the world$B!G(Bs
first? fully interchangeable lenses to feature an aperture of 1:2.0 across
their entire zooming range. These two lenses give photographers the freedom
to apply a 1:2.0 aperture from a wide-angle focal length of 14mm (35mm
equiv: 28mm) to telephoto 100mm (35mm equiv: 200mm).

http://www.dpreview.com/news/0502/05021712oly_3new43len...

We're finally seeing the advantage of the smaller format!

David J. Littleboy
Tokyo, Japan

More about : impressed

Anonymous
February 18, 2005 9:01:48 PM

Archived from groups: rec.photo.digital (More info?)

"David J. Littleboy" <davidjl@gol.com> wrote in message
news:cv4au6$2ii$1@nnrp.gol.com...
>
> Sheesh, I'm going to have to stop badmouthing the lack of fast lenses for
> the E300!
>
> The ZUIKO DIGITAL ED 14-35mm (35mm equiv: 28-70mm) 1:2.0 and ZUIKO DIGITAL
> ED 35-100mm (35mm equiv: 70-200mm) 1:2.0? are understood to be the
world$B!G(Bs
> first? fully interchangeable lenses to feature an aperture of 1:2.0 across
> their entire zooming range. These two lenses give photographers the
freedom
> to apply a 1:2.0 aperture from a wide-angle focal length of 14mm (35mm
> equiv: 28mm) to telephoto 100mm (35mm equiv: 200mm).
>
> http://www.dpreview.com/news/0502/05021712oly_3new43len...
>
> We're finally seeing the advantage of the smaller format!
>
> David J. Littleboy
> Tokyo, Japan
>
>
One advantage is allowing the use of lower ISOs since this format is not as
clean as the other dSLRs. I have to wonder how far this system can go.
Packing more megapixels on this smaller sensor will lead to noise issues.
The APS sized sensors should be good to 12 MP. I doubt the Olympus sensor
will handle this.

One thing I don't get is the high price of the Olympus camera and lenses. If
sensor size dictates the price of the SLR as many claim, The E300 should
cost much less.
bg
bg
Anonymous
February 18, 2005 9:01:48 PM

Archived from groups: rec.photo.digital (More info?)

On Fri, 18 Feb 2005 18:01:47 +0900, "David J. Littleboy"
<davidjl@gol.com> wrote:

>
>Sheesh, I'm going to have to stop badmouthing the lack of fast lenses for
>the E300!
>
>The ZUIKO DIGITAL ED 14-35mm (35mm equiv: 28-70mm) 1:2.0 and ZUIKO DIGITAL
>ED 35-100mm (35mm equiv: 70-200mm) 1:2.0? are understood to be the world$B!G(Bs
>first? fully interchangeable lenses to feature an aperture of 1:2.0 across
>their entire zooming range. These two lenses give photographers the freedom
>to apply a 1:2.0 aperture from a wide-angle focal length of 14mm (35mm
>equiv: 28mm) to telephoto 100mm (35mm equiv: 200mm).
>
>http://www.dpreview.com/news/0502/05021712oly_3new43len...
>
>We're finally seeing the advantage of the smaller format!


Lemme get this straight. We miniaturized
the entire "system" to accomodate the smaller
sensor.

But because its photosites are smaller, it's
noisier at any given ISO. So now we need
wider faster lenses to use lower ISO values..
which bulks up the lenses.

Back to square one, no?

I think I'm going to sit out the next generation
or two and see if they ever figure out how to
cost-reduce the larger sensors.


rafe b.
http://www.terrapinphoto.com
Related resources
Anonymous
February 18, 2005 9:01:48 PM

Archived from groups: rec.photo.digital (More info?)

"David J. Littleboy" <davidjl@gol.com> wrote in message
news:cv4au6$2ii$1@nnrp.gol.com...
>
> Sheesh, I'm going to have to stop badmouthing the lack of fast lenses for
> the E300!
>
> The ZUIKO DIGITAL ED 14-35mm (35mm equiv: 28-70mm) 1:2.0 and ZUIKO DIGITAL
> ED 35-100mm (35mm equiv: 70-200mm) 1:2.0? are understood to be the
> world$B!G(Bs
> first? fully interchangeable lenses to feature an aperture of 1:2.0 across
> their entire zooming range. These two lenses give photographers the
> freedom
> to apply a 1:2.0 aperture from a wide-angle focal length of 14mm (35mm
> equiv: 28mm) to telephoto 100mm (35mm equiv: 200mm).
>
> http://www.dpreview.com/news/0502/05021712oly_3new43len...
>
> We're finally seeing the advantage of the smaller format!

Too bad we're not seeing the size and weight savings of the smaller format.
The new Rebel is smaller and lighter, though.

Seriously, I haven't seen anything from Oly that would convince me to buy
into that system over a Nikon/Canon system. Where's the beef?

>
> David J. Littleboy
> Tokyo, Japan
>
>
Anonymous
February 18, 2005 9:40:23 PM

Archived from groups: rec.photo.digital (More info?)

"David J. Littleboy" <davidjl@gol.com> wrote in
news:cv4au6$2ii$1@nnrp.gol.com:

> Sheesh, I'm going to have to stop badmouthing the lack of fast lenses
> for the E300!
>
> The ZUIKO DIGITAL ED 14-35mm (35mm equiv: 28-70mm) 1:2.0 and ZUIKO
> DIGITAL ED 35-100mm (35mm equiv: 70-200mm) 1:2.0? are understood to be
> the world$B!G(Bs first? fully interchangeable lenses to feature an
> aperture of 1:2.0 across their entire zooming range. These two lenses
> give photographers the freedom to apply a 1:2.0 aperture from a
> wide-angle focal length of 14mm (35mm equiv: 28mm) to telephoto 100mm
> (35mm equiv: 200mm).
>
> http://www.dpreview.com/news/0502/05021712oly_3new43len...
>
> We're finally seeing the advantage of the smaller format!
>

Hmmm --- reading your post some times over and over again
I tried to understand if this is sarcasm or not :) 

Of course - you need a big lens to gather lots of light.
The smaller the sensor - the smaller F-stop number is needed
to get the same system sensitivity.

It is the diameter of the aperture that defines the amount
of light that is gathered from the subject.


/Roland
Anonymous
February 18, 2005 11:17:00 PM

Archived from groups: rec.photo.digital (More info?)

"TAFKAB" <TheArtist@FormerlyKnownAs.Bowser> wrote in message news:1ErRd.1

> Seriously, I haven't seen anything from Oly that would convince me to buy
> into that system over a Nikon/Canon system. Where's the beef?<


Presumably, all gone - eaten by people who then developed Mad Cow Disease,
and who are now on the point of spending thousands of dollars/pounds on a
lens system for a the Four-Thirds format.
February 19, 2005 3:23:40 AM

Archived from groups: rec.photo.digital (More info?)

BG250 wrote:


>
> One thing I don't get is the high price of the Olympus camera and lenses.
> If sensor size dictates the price of the SLR as many claim, The E300
> should cost much less.
>

Have you checked what you can buy one of these for with their rebate?

--

Stacey
February 19, 2005 3:28:21 AM

Archived from groups: rec.photo.digital (More info?)

TAFKAB wrote:


>
> Seriously, I haven't seen anything from Oly that would convince me to buy
> into that system over a Nikon/Canon system. Where's the beef?
>


Then don't. Nobody is twisting anyone's arm and for many uses these aren't
the best choice.. For me, it works.

OTOH it seems canon owners spend a lot of their time trying to convince
"someone" THERE choice is the best for EVERYONE? I never have figured that
one out. Post the word "Olympus" (or even Nikon) and David/Rafe run into
the fray telling everyone how bad they are while neither have ever even
held one? Just seem odd to me..

--

Stacey
Anonymous
February 19, 2005 3:38:16 AM

Archived from groups: rec.photo.digital (More info?)

"rafe bustin" <rafeb@speakeasy.net> wrote:
> <davidjl@gol.com> wrote:
>
> >Sheesh, I'm going to have to stop badmouthing the lack of fast lenses for
> >the E300!
> >
> >http://www.dpreview.com/news/0502/05021712oly_3new43len...
> >
> >We're finally seeing the advantage of the smaller format!
>
> Lemme get this straight. We miniaturized
> the entire "system" to accomodate the smaller
> sensor.
>
> But because its photosites are smaller, it's
> noisier at any given ISO. So now we need
> wider faster lenses to use lower ISO values..
> which bulks up the lenses.
>
> Back to square one, no?

Yes, Rafe, that's my rant. You're going to owe me royalties if you want to
use it.

> I think I'm going to sit out the next generation
> or two and see if they ever figure out how to
> cost-reduce the larger sensors.

My calculation is as follows. The geometric mean of US$900 (350D body) and
US$10,000 (the latest generation of MF backs) is US$3000. The industry ought
to be able to make a US$3000 full-frame camera. Today. Today's technology.

So it's just a matter of time. Basically, as soon as the 1Dsmk2 market is
saturated and demand drops off, we'll see a "3D" with that sensor in it. In
the US$3000 ballpark.

David J. Littleboy
Tokyo, Japan
Anonymous
February 19, 2005 3:38:17 AM

Archived from groups: rec.photo.digital (More info?)

"David J. Littleboy" <davidjl@gol.com> wrote in
news:cv525j$95m$2@nnrp.gol.com:
>
> "rafe bustin" <rafeb@speakeasy.net> wrote:
>
>> I think I'm going to sit out the next generation or two and see if
>> they ever figure out how to cost-reduce the larger sensors.
>
> My calculation is as follows. The geometric mean of US$900 (350D body)
> and US$10,000 (the latest generation of MF backs) is US$3000. The
> industry ought to be able to make a US$3000 full-frame camera. Today.
> Today's technology.
>
> So it's just a matter of time. Basically, as soon as the 1Dsmk2 market
> is saturated and demand drops off, we'll see a "3D" with that sensor
> in it. In the US$3000 ballpark.

I don't see where the geometric mean comes into the situation...

The semiconductor yield curve is fairly well understood: there's a
defectivity rate per unit area. The best defect density I know of is
0.14, so let's use that.

For the 300D, you get 66 die per wafer on an 8 inch wafer line. 37% of
them work at this defectivity, so you get 24 working die. Each working
die costs about $50 out of the fab (assuming the wafer costs $1000).

For full frame, you get 23 die per wafer on the same 8 inch line. Only
9.6% of them work (ouch), so you get 2 working die per wafer. Each
working die costs $600 out of the fab.

For medium format, odds are you don't get any working die per wafer. For
48x36mm imager size at the above defectivity, then only 1.3% of them
work and there are only 10 per wafer. The costs are not readily
calcuable (presumably they start shipping defective ones and hope no-one
complains when there's a group of totally dead pixels). So any mean
between a camera containing one of these and one containing an APS
imager is kind of meaningless.

And that was the best defectivity I know of: odds are that imagers
aren't made on that fab line and they've a worse defectivity. And that
wafers for the imager process have greater than standard costs (in
particular, the package they go into is going to hurt somewhat). And
that there's some huge R&D cost to recover...

1.3x format cameras are a lot more hopeful. 28x18mm imagers only cost
$100 on the above assumptions.

--Sophie with a semiconductor cost model to hand
Anonymous
February 19, 2005 3:38:17 AM

Archived from groups: rec.photo.digital (More info?)

> My calculation is as follows. The geometric mean of US$900 (350D body) and
> US$10,000 (the latest generation of MF backs) is US$3000. The industry
ought
> to be able to make a US$3000 full-frame camera. Today. Today's technology.

Actually, if they really wanted to, I truly believe that they could offer
a full-frame camera today for $2,000 - perhaps a little less. But you have
to remember, the goal of any company is not to give away everything it can
possibly afford to, but to make as much money as it possibly can.

steve
Anonymous
February 19, 2005 3:38:17 AM

Archived from groups: rec.photo.digital (More info?)

"David J. Littleboy" <davidjl@gol.com> wrote in message
news:cv525j$95m$2@nnrp.gol.com...
>
> "rafe bustin" <rafeb@speakeasy.net> wrote:
>> <davidjl@gol.com> wrote:
>>
>> >Sheesh, I'm going to have to stop badmouthing the lack of fast lenses
>> >for
>> >the E300!
>> >
>> >http://www.dpreview.com/news/0502/05021712oly_3new43len...
>> >
>> >We're finally seeing the advantage of the smaller format!
>>
>> Lemme get this straight. We miniaturized
>> the entire "system" to accomodate the smaller
>> sensor.
>>
>> But because its photosites are smaller, it's
>> noisier at any given ISO. So now we need
>> wider faster lenses to use lower ISO values..
>> which bulks up the lenses.
>>
>> Back to square one, no?
>
> Yes, Rafe, that's my rant. You're going to owe me royalties if you want to
> use it.
>
>> I think I'm going to sit out the next generation
>> or two and see if they ever figure out how to
>> cost-reduce the larger sensors.
>
> My calculation is as follows. The geometric mean of US$900 (350D body) and
> US$10,000 (the latest generation of MF backs) is US$3000. The industry
> ought
> to be able to make a US$3000 full-frame camera. Today. Today's technology.
>
> So it's just a matter of time. Basically, as soon as the 1Dsmk2 market is
> saturated and demand drops off, we'll see a "3D" with that sensor in it.
> In
> the US$3000 ballpark.
>
> David J. Littleboy
> Tokyo, Japan
>
>
>
What I can't figure out is why not use the old 1Ds' 11mp sensor in a "3D".
The R&D has be amortized, put it in a slightly upgraded 20D or non upgraded
1 series body, keep its frame rate at 3-4, raw burst rate under 20, jpeg at
around 30, sell it for $3000 or so...or even the same price as a 1D mkII.

--
Skip Middleton
http://www.shadowcatcherimagery.com
Anonymous
February 19, 2005 3:38:18 AM

Archived from groups: rec.photo.digital (More info?)

Sophie Wilson <sophie.wilson@bigfoot.com> wrote in
news:Xns9601A6D01D01Asophiewilson@130.133.1.4:

> The semiconductor yield curve is fairly well understood: there's a
> defectivity rate per unit area. The best defect density I know of is
> 0.14, so let's use that.
> ...

Does your calculations assume that a sensor shall have no
defects to work?

There are defects on sensors that are shipped. And those
sensors are working. They might have some dead pixels.


/Roland
Anonymous
February 19, 2005 3:38:18 AM

Archived from groups: rec.photo.digital (More info?)

Sophie Wilson <sophie.wilson@bigfoot.com> wrote:
[...]
> For the 300D, you get 66 die per wafer on an 8 inch wafer line. 37% of
> them work at this defectivity, so you get 24 working die. Each working
> die costs about $50 out of the fab (assuming the wafer costs $1000).
[...]

I thought pretty much everybody had switched to 12" wafers these days.
IIRC, there are big cost savings to using 12" wafers.

--Paul
** Note "removemunged" in email address and remove to reply. **
Anonymous
February 19, 2005 3:38:18 AM

Archived from groups: rec.photo.digital (More info?)

On 18 Feb 2005 16:23:54 GMT, Sophie Wilson <sophie.wilson@bigfoot.com>
wrote:

>"David J. Littleboy" <davidjl@gol.com> wrote in
>news:cv525j$95m$2@nnrp.gol.com:
>>
>> "rafe bustin" <rafeb@speakeasy.net> wrote:
>>
>>> I think I'm going to sit out the next generation or two and see if
>>> they ever figure out how to cost-reduce the larger sensors.
>>
>> My calculation is as follows. The geometric mean of US$900 (350D body)
>> and US$10,000 (the latest generation of MF backs) is US$3000. The
>> industry ought to be able to make a US$3000 full-frame camera. Today.
>> Today's technology.
>>
>> So it's just a matter of time. Basically, as soon as the 1Dsmk2 market
>> is saturated and demand drops off, we'll see a "3D" with that sensor
>> in it. In the US$3000 ballpark.
>
>I don't see where the geometric mean comes into the situation...
>
>The semiconductor yield curve is fairly well understood: there's a
>defectivity rate per unit area. The best defect density I know of is
>0.14, so let's use that.


Some counter arguments to this:

1. Photosites are absolutely enormous
compared to conventional silicon features.
Photosites (good ones) are on the order
of 5 to 7 microns, whereas "standard"
features now are on the order of 0.1 micron.
That's about 5000 times more area.

Therefore, defects which might destroy
a RAM or VLSI cell might have negligible
impact on a photosite.

2. Firmware compensation for dead pixels.
As long as the defect doesn't extend over
several pixels, it can probably be dealt
with in firmware. AFAIK, most silicon
area sensors are *not* defect free.

The data sheet for the Phillips 24x36 mm
6 MP sensor (now about 4 or 5 years old)
showed the chip offered in several defect
grades, ranging from a few thousand bad
pixels down to a few hundred bad pixels.

3. Defects in interconnections can be
fatal to the die, but in a 6Mpixel sensor,
there aren't that many interconnections
(by VLSI standards.) In any case, these
could be made "fat" by VLSI standards
without appreciably impacting the photosite
size.

I think what it's really going to take is
for some company to tool a fab specifically
for sensor production.


rafe b.
http://www.terrapinphoto.com
Anonymous
February 19, 2005 3:38:19 AM

Archived from groups: rec.photo.digital (More info?)

Paul Wylie <paul@teamwylie.removemunged.org> wrote in news:cv5h4k$i5f$2
@reader2.panix.com:

> Sophie Wilson <sophie.wilson@bigfoot.com> wrote:
> [...]
>> For the 300D, you get 66 die per wafer on an 8 inch wafer line. 37% of
>> them work at this defectivity, so you get 24 working die. Each working
>> die costs about $50 out of the fab (assuming the wafer costs $1000).
> [...]
>
> I thought pretty much everybody had switched to 12" wafers these days.
> IIRC, there are big cost savings to using 12" wafers.

Yes. But the problem is that the defectivity of the processes I know about
on 300mm wafers is higher than the 0.14 on an old 8" line. So its not as
good at making imagers, which are critically dependent on good defectivity
- there's no way to (say) put in redundancy like is done with memory
arrays. There's another problem: if you're using a dedicated line, you may
not be able to afford the costs of 300mm equipment "just" for imagers. And
currently the raw cost of a 12" wafer is larger. Certainly its something
that might change in the future (e.g. when someone is decommissioning a
300mm line!) and it does have benefits for the larger imagers (less
wastage). [there's also some technical stuff with large reticles that isn't
good for imagers on 300mm - the prime users of 300mm are making chips that
just aren't very large compared to an imager (up to about 20mm on a side)]

I had a go at introducing a new thing to the cost model - a 300mm process
with a defectivity of 0.14 - but I simply don't know how much the wafer
would cost (its clearly more than just area related since its harder to
refine the silicon "rod"): I copied another process's 300mm wafer cost - it
was high, so I used 2008's cost. It didn't pay off - you got 61 36x24mm
imagers on the wafer, 9.6% of them worked (same defectivity) i.e. only 5
and the final imager cost $830. Currently, defectivity of 300mm is higher
than that, so I guess no one is doing this for imagers. It really is
incredibly sensitive to this - the difference between 0.14 and 0.15 shows
up (5.7% of 36x24mm imagers work at 0.15 defectivity - i.e. only 3 on the
300mm wafer). If someone has a process skewed towards making imagers and
can get the defectivity down, then they can clean up!

--Sophie
Anonymous
February 19, 2005 3:38:19 AM

Archived from groups: rec.photo.digital (More info?)

Roland Karlsson wrote:
> Sophie Wilson <sophie.wilson@bigfoot.com> wrote in
> news:Xns9601A6D01D01Asophiewilson@130.133.1.4:
>
>
>>The semiconductor yield curve is fairly well understood: there's a
>>defectivity rate per unit area. The best defect density I know of is
>>0.14, so let's use that.
>>...
>
>
> Does your calculations assume that a sensor shall have no
> defects to work?
>
> There are defects on sensors that are shipped. And those
> sensors are working. They might have some dead pixels.
>
>
> /Roland

A few dead pixels can be handled by software. Too many and you end up
with an image that looks like you used too much jpeg compression....


--
Ron Hunter rphunter@charter.net
Anonymous
February 19, 2005 10:56:45 AM

Archived from groups: rec.photo.digital (More info?)

"Skip M" <shadowcatcher@cox.net> wrote:
> >
> What I can't figure out is why not use the old 1Ds' 11mp sensor in a "3D".

It's a dog. It's the noisiest sensor Canon has ever produced.

> The R&D has be amortized, put it in a slightly upgraded 20D or non
upgraded
> 1 series body, keep its frame rate at 3-4, raw burst rate under 20, jpeg
at
> around 30, sell it for $3000 or so...or even the same price as a 1D mkII.

Whatever. As long as they don't include a built-in grip so I can lift the
damn thing.

David J. Littleboy
Tokyo, Japan
Anonymous
February 19, 2005 11:00:05 AM

Archived from groups: rec.photo.digital (More info?)

"Roland Karlsson" <roland_dot_karlsson@bonetmail.com> wrote in message
news:Xns9601C81FACB35klotjohan@130.133.1.4...
> "David J. Littleboy" <davidjl@gol.com> wrote in
> news:cv4au6$2ii$1@nnrp.gol.com:
>
> > Sheesh, I'm going to have to stop badmouthing the lack of fast lenses
> > for the E300!
> >
> > The ZUIKO DIGITAL ED 14-35mm (35mm equiv: 28-70mm) 1:2.0 and ZUIKO
> > DIGITAL ED 35-100mm (35mm equiv: 70-200mm) 1:2.0? are understood to be
> > the world$B!G(Bs first? fully interchangeable lenses to feature an
> > aperture of 1:2.0 across their entire zooming range. These two lenses
> > give photographers the freedom to apply a 1:2.0 aperture from a
> > wide-angle focal length of 14mm (35mm equiv: 28mm) to telephoto 100mm
> > (35mm equiv: 200mm).
> >
> > http://www.dpreview.com/news/0502/05021712oly_3new43len...
> >
> > We're finally seeing the advantage of the smaller format!
> >
>
> Hmmm --- reading your post some times over and over again
> I tried to understand if this is sarcasm or not :) 

It's not. The smaller sensor should allow faster, cheaper, and more extreme
range zoom lenses. Oly finally coughed up.

> Of course - you need a big lens to gather lots of light.
> The smaller the sensor - the smaller F-stop number is needed
> to get the same system sensitivity.

Yes. That's what I've been saying about the E-1 and now E300 for a couple of
years.

> It is the diameter of the aperture that defines the amount
> of light that is gathered from the subject.

Uh, no. It's the diameter divided by the focal length. It's the size of the
pixels that determines the sensitivity.

David J. Littleboy
Tokyo, Japan
Anonymous
February 19, 2005 11:00:06 AM

Archived from groups: rec.photo.digital (More info?)

In article <cv5sdk$h5n$3@nnrp.gol.com>,
"David J. Littleboy" <davidjl@gol.com> wrote:

> > Of course - you need a big lens to gather lots of light.
> > The smaller the sensor - the smaller F-stop number is needed
> > to get the same system sensitivity.
>
> Yes. That's what I've been saying about the E-1 and now E300 for a couple of
> years.

Actually it is a big sensor that NEEDS more light than a small one,
because the same exposure will be spread out over a larger area, which
reduces light-intensity quadratically. The aperture therefore NEEDS to
be bigger to still get the same amount of light on the chip per sq. mm.

If you move a lightbulb/slide-projector/camera-lens further away so it
lights a four times larger area, you lose 2 stops. This is all very
basic.

Lourens
Anonymous
February 19, 2005 11:00:07 AM

Archived from groups: rec.photo.digital (More info?)

In article <smak-D035F4.01375219022005@news.wanadoo.nl>, smak@wanadoo.nl
says...
> Actually it is a big sensor that NEEDS more light than a small one,
> because the same exposure will be spread out over a larger area, which
> reduces light-intensity quadratically. The aperture therefore NEEDS to
> be bigger to still get the same amount of light on the chip per sq. mm.

Uh, aperture is expressed as a ratio for a reason. F/2.8 is the same
amount of light no matter what the format. So why even bother with this
line of thought?
--
http://www.pbase.com/bcbaird/
Anonymous
February 19, 2005 11:00:08 AM

Archived from groups: rec.photo.digital (More info?)

On Sat, 19 Feb 2005 00:42:29 GMT, Brian C. Baird <nospam@please.no>
wrote:

>In article <smak-D035F4.01375219022005@news.wanadoo.nl>, smak@wanadoo.nl
>says...
>> Actually it is a big sensor that NEEDS more light than a small one,
>> because the same exposure will be spread out over a larger area, which
>> reduces light-intensity quadratically. The aperture therefore NEEDS to
>> be bigger to still get the same amount of light on the chip per sq. mm.
>
>Uh, aperture is expressed as a ratio for a reason. F/2.8 is the same
>amount of light no matter what the format. So why even bother with this
>line of thought?


I agree entirely. Lourens is just blowing smoke here.


rafe b.
http://www.terrapinphoto.com
Anonymous
February 19, 2005 11:41:16 AM

Archived from groups: rec.photo.digital (More info?)

On Sat, 19 Feb 2005 00:28:21 -0500, Stacey <fotocord@yahoo.com> wrote:


> OTOH it seems canon owners spend a lot of their time trying to convince
>"someone" THERE choice is the best for EVERYONE? I never have figured that
>one out. Post the word "Olympus" (or even Nikon) and David/Rafe run into
>the fray telling everyone how bad they are while neither have ever even
>held one? Just seem odd to me..


Nah, it's more like, where the word "Olympus" appears
there's likely to be a foolish post with your name on
it, implying that you're the only smart digicam buyer
on this forum. You're our new George Preddy.

My experience of something doesn't make it true or
untrue, better or worse. Reviews and opinions matter,
to an extent. Physics matters more. You used to
understand this, kievgurl, but since you've gone
digital, you've done an about face.

Why don't you admit to the world that your 35 mm
gear was also Oly, and that you might, just might
be a bit biased in your choice here?


rafe b.
http://www.terrapinphoto.com
Anonymous
February 19, 2005 1:10:12 PM

Archived from groups: rec.photo.digital (More info?)

"Brian C. Baird" <nospam@please.no> wrote in message
news:MPG.1c80557ac0f22b8c98a608@news.verizon.net...
> In article <smak-D035F4.01375219022005@news.wanadoo.nl>, smak@wanadoo.nl
> says...
> > Actually it is a big sensor that NEEDS more light than a small one,
> > because the same exposure will be spread out over a larger area, which
> > reduces light-intensity quadratically. The aperture therefore NEEDS to
> > be bigger to still get the same amount of light on the chip per sq. mm.
>
> Uh, aperture is expressed as a ratio for a reason. F/2.8 is the same
> amount of light no matter what the format. So why even bother with this
> line of thought?

He has a point: the absolute size of the aperture for a "normal" f/2.0 lens
on a 4/3 camera is 1/2 the absolute size of the aperture for a "normal"
f/2.0 lens on a full-frame camera. So there could be as much as a factor of
_eight_ difference in weight.

Where the 4/3 partizans are misleading in their claims is that the format
multiplier from 4/3 to APS-C isn't 2.0, but 1.2. So the difference is pretty
small. A 25mm f/1.4 lens on the Oly would be only a tad smaller/lighter than
the Stigma 30/1.4.

By the way, 4/3 _is_ a good idea. Larger formats always provide better
performance, and smaller formats always are cheaper, lighter, and more
convenient. And since there's an enormous gaping hole between 15x23 and
6.6x8.8, it makes sense that a camera in that space would be cheaper,
lighter, and more convenient than APS-C and better quality than the consumer
dcams. (Unfortunately, Oly hasn't coughed up on the cheaper/lighter bit, and
only now is coughing up on the more convenient bit.)

By the way, the reason the 4/3 cameras have such lousy noise is not so much
the format as the fact that Canon makes better sensors than Kodak
(FillFactory).

My objections to 4/3 are (1) the claim that it's just as good or better as
APS-C. That's simply wrong. And (2) that spending that kind of money on
lenses for a system that has no upgrade path is a waste of money.

David J. Littleboy
Tokyo, Japan
Anonymous
February 19, 2005 4:22:43 PM

Archived from groups: rec.photo.digital (More info?)

Sophie Wilson wrote:
> The semiconductor yield curve is fairly well understood: there's a
> defectivity rate per unit area. The best defect density I know of is
> 0.14, so let's use that.

Correct me if I'm wrong: that's defects per cm**2 for the
processes' designed lambda (feature size). Most of the industry
wants to maximize the device-count a process can support, so
they want to keep line-size small. Imagers are different; we
increase area by increasing feature size. Larger features lead
to a lower rate of defects per unit area.

The effect of increasing lambda on defectivity seems not to be
well-studied. I've read that defect density per unit area
decreases as approximately the inverse cube of the line width,
though that may be optimistic. At constant device count, the
area of a chip increases as the square of the line-width, so the
percentage of chips that are good actually increases with size.


--
--Bryan
Anonymous
February 20, 2005 2:39:13 AM

Archived from groups: rec.photo.digital (More info?)

"David J. Littleboy" <davidjl@gol.com> wrote in
news:cv5sdk$h5n$3@nnrp.gol.com:

>> It is the diameter of the aperture that defines the amount
>> of light that is gathered from the subject.
>
> Uh, no. It's the diameter divided by the focal length. It's the size
> of the pixels that determines the sensitivity.

No - actually it is the diameter of the aperture that
is important.

The object you want to photograph emits/reflects light.
Some of the light it sends out hits the aperture. The
bigger the aperture, the higher percentage of the light
is catched.

Thats it really. A larger aperture (in mm or inch) catches
more photons. This means that a system with a smaller sensor
needs a lower F-number to be equally light sensitive.



/Roland
Anonymous
February 20, 2005 1:29:41 PM

Archived from groups: rec.photo.digital (More info?)

In article <1pfe11pfi90olroismd3pjt4l0op7php7k@4ax.com>,
rafe bustin <rafeb@speakeasy.net> wrote:
>On Sat, 19 Feb 2005 00:28:21 -0500, Stacey <fotocord@yahoo.com> wrote:
>
>
>> OTOH it seems canon owners spend a lot of their time trying to convince
>>"someone" THERE choice is the best for EVERYONE? I never have figured that
>>one out. Post the word "Olympus" (or even Nikon) and David/Rafe run into
>>the fray telling everyone how bad they are while neither have ever even
>>held one? Just seem odd to me..
>
>
>Nah, it's more like, where the word "Olympus" appears
>there's likely to be a foolish post with your name on
>it, implying that you're the only smart digicam buyer
>on this forum. You're our new George Preddy.

That's possibly a bit strong. Stacey has only just "gone digital",
apparently after a road to Damascus type experience, and is understandably
very enthusiastic about her new purchase. We've all been there, haven't we?

GP, on the other hand, is a net.kook in the best tradition of them.
Anonymous
February 20, 2005 5:07:15 PM

Archived from groups: rec.photo.digital (More info?)

"Skip M" <shadowcatcher@cox.net> wrote in message
news:1BtRd.32133$xt.2830@fed1read07...
> "David J. Littleboy" <davidjl@gol.com> wrote in message
> news:cv525j$95m$2@nnrp.gol.com...
>>
>> "rafe bustin" <rafeb@speakeasy.net> wrote:
>>> <davidjl@gol.com> wrote:
>>>
>>> >Sheesh, I'm going to have to stop badmouthing the lack of fast lenses
>>> >for
>>> >the E300!
>>> >
>>> >http://www.dpreview.com/news/0502/05021712oly_3new43len...
>>> >
>>> >We're finally seeing the advantage of the smaller format!
>>>
>>> Lemme get this straight. We miniaturized
>>> the entire "system" to accomodate the smaller
>>> sensor.
>>>
>>> But because its photosites are smaller, it's
>>> noisier at any given ISO. So now we need
>>> wider faster lenses to use lower ISO values..
>>> which bulks up the lenses.
>>>
>>> Back to square one, no?
>>
>> Yes, Rafe, that's my rant. You're going to owe me royalties if you want
>> to
>> use it.
>>
>>> I think I'm going to sit out the next generation
>>> or two and see if they ever figure out how to
>>> cost-reduce the larger sensors.
>>
>> My calculation is as follows. The geometric mean of US$900 (350D body)
>> and
>> US$10,000 (the latest generation of MF backs) is US$3000. The industry
>> ought
>> to be able to make a US$3000 full-frame camera. Today. Today's
>> technology.
>>
>> So it's just a matter of time. Basically, as soon as the 1Dsmk2 market is
>> saturated and demand drops off, we'll see a "3D" with that sensor in it.
>> In
>> the US$3000 ballpark.
>>
>> David J. Littleboy
>> Tokyo, Japan
>>
>>
>>
> What I can't figure out is why not use the old 1Ds' 11mp sensor in a "3D".
> The R&D has be amortized, put it in a slightly upgraded 20D or non
> upgraded 1 series body, keep its frame rate at 3-4, raw burst rate under
> 20, jpeg at around 30, sell it for $3000 or so...or even the same price as
> a 1D mkII.
>
> --
> Skip Middleton
> http://www.shadowcatcherimagery.com
>

I was hoping that would happen also. It may be that the full frame sensors
are so expensive that it is just not possible.
Anonymous
February 20, 2005 9:41:14 PM

Archived from groups: rec.photo.digital (More info?)

Roland Karlsson <roland_dot_karlsson@bonetmail.com> writes:

>No - actually it is the diameter of the aperture that
>is important.

Important for what? A larger aperture collects more photons. But it's
photons *per unit area at the sensor* that determines exposure time
needed. That depends on focal length, so f/number is generally a more
useful measure of light throughput than aperture diameter.

For example, suppose you have two cameras, one with a sensor 2X the
dimensions (4X the area) of the other. Both cameras have lenses with
the same field of view, so the lens on "big" is twice the focal length
of the lens on "small". When both cameras are set to f/4, the actual
entrance pupil on "big" is twice the diameter of that on "small", so
"big"'s lens captures 4 times as many photons. But those photons are
spread over 4 times the sensor area, so light flux at the sensor is the
same for both cameras.

It's up to the camera designer to decide what that extra sensor area
will be used for. The big sensor could have the same number of pixels
as the small one, with each pixel having 4X the area and thus either
less noise at the same ISO, or higher ISO at the same noise. Or it
could have 4X as many pixels, with the same noise/sensitivity as the
small sensor. Or somewhere in between.

>Thats it really. A larger aperture (in mm or inch) catches
>more photons. This means that a system with a smaller sensor
>needs a lower F-number to be equally light sensitive.

It all depends on what you assume is kept constant. If the pixel size
is constant, then the camera with the smaller sensor has *the same*
sensitivity and noise performance as the larger sensor, and performs as
well at the same f/number, which means a smaller absolute aperture
diameter.

On the other hand, if you assume that the number of pixels is kept
constant, and thus the larger sensor has 4X the area per pixel, and you
also assume that the larger sensor is operated to get the same S/N as
the smaller one (not the better S/N it is capable of), THEN the large
sensor can be considered to be 4X as sensitive as the small one, and
then both cameras need the same aperture diameter from their lens.

But this is a pretty long list of assumptions. In reality, larger
sensors tend to have more pixels (as well as somewhat larger pixels),
not the same number as smaller sensors. And people with large-sensor
cameras often operate them at ISO 100 to get better S/N performance,
instead of using ISO 800 or 1600. Under these conditions, f/number is
more useful than aperture diameter.

Dave

PS: there is one exception to the above: astronomy. Stars are optical
point sources, and they are imaged as points on the sensor (or as close
to a point as the lens can manage). If you use a longer focal length
lens, the star images *do not* spread out to cover more sensor area;
each star is still a point on the sensor. The magnification change
merely spreads the start further apart. Under these conditions, star
image brightness *does* depend on lens absolute aperture, not f/number.
But this doesn't apply to general photography.
Anonymous
February 20, 2005 10:05:33 PM

Archived from groups: rec.photo.digital (More info?)

davem@cs.ubc.ca (Dave Martindale) wrote in news:cvalka$q8c$1
@mughi.cs.ubc.ca:

> Important for what? A larger aperture collects more photons. But it's
> photons *per unit area at the sensor* that determines exposure time
> needed.

No - it is not.

It is the number of photons per time unit and pixel that is important.
If you have the same f-stop (e.g. F/4.0) and have a twice as large
diameter of the aperture, you also have twice as large detectors for
each pixel. Thus - you collect four times as many photons.

It is the diameter of the aperture that determines the light
sensitivity of the system.



/Roland
Anonymous
February 21, 2005 12:29:05 AM

Archived from groups: rec.photo.digital (More info?)

On Sat, 19 Feb 2005 20:11:11 -0500, Stacey <fotocord@yahoo.com> wrote:


>Notice you don't see me posting to canon threads about how I didn't care
>for the image quality even though I have used one of them, unlike you and
>David. Yet you and David spend plenty of time posting your opinion about a
>camera you have never even held in your hand? Just seems very odd to me.


This isn't quite the way I remember it.
You've referred to those "other" dSLRs
producing images that look "waxy"
or "plastic" for example. By contrast,
you claim the Oly's images "don't look
digital."

Most (all?) of your statements in favor
of the E300 (or against other brands/models)
have been in subjective terms. So there's
not much basis for an argument.

No, I've not used an E300, and have no
opinions on it... my statements are
based simply on area per photosite,
which is considerably smaller in the
Oly 300 than the Canon 10D. I think
Roger Clark explains quite clearly
why this matters, if it wasn't obvious
from experience or from your personal
comparisons. (And as you've worked
with film formats from 35mm up to LF,
none of this should surprise you.)

Note well that my "criticism" extends
to the Canon 20D also, ie., it's getting
33% more pixels than its immediate
predecessor without any increase in overall
sensor area.


rafe b.
http://www.terrapinphoto.com
February 21, 2005 4:10:45 AM

Archived from groups: rec.photo.digital (More info?)

rafe bustin wrote:

> On Sat, 19 Feb 2005 20:11:11 -0500, Stacey <fotocord@yahoo.com> wrote:
>
>
>>Notice you don't see me posting to canon threads about how I didn't care
>>for the image quality even though I have used one of them, unlike you and
>>David. Yet you and David spend plenty of time posting your opinion about a
>>camera you have never even held in your hand? Just seems very odd to me.
>
>
> This isn't quite the way I remember it.
> You've referred to those "other" dSLRs
> producing images that look "waxy"
> or "plastic" for example.

In a thread about =olympus cameras= that you and dave were on one of your
bashing rants... I posted that I HAD checked out the other brands
personally (actually used them, imagine that!) and explained why I didn't
buy a Canon.


>
> Most (all?) of your statements in favor
> of the E300 (or against other brands/models)
> have been in subjective terms. So there's
> not much basis for an argument.

So? There is nothing subjective about what someone considers a "quality"
image? Is there a mathmatical formula for "good color" or "too much"
saturation? Or is high ISO noise the only criteria?

>
> No, I've not used an E300, and have no
> opinions on it...

??? You sure seem to have one to me! :-)

> my statements are
> based simply on area per photosite,
>

You ignore what while the per pixel noise may be higher, there are MORE of
them so on the same size image they aren't as visible. The end result is
more resolution of fine details if the optics are good enough.

--

Stacey
Anonymous
February 21, 2005 12:53:57 PM

Archived from groups: rec.photo.digital (More info?)

On Mon, 21 Feb 2005 01:10:45 -0500, Stacey <fotocord@yahoo.com> wrote:


>You ignore what while the per pixel noise may be higher, there are MORE of
>them so on the same size image they aren't as visible. The end result is
>more resolution of fine details if the optics are good enough.


This reminds me of that film where
Woody Allen and his yuppie friends are
kvetching about the poor quality of food
in some swank restaurant... and also
complaining that the portions are small.

The Oly tries to get more pixels out of
less silicon than the 10D. There's no
free lunch, not unless you can change the
laws of physics.


rafe b.
http://www.terrapinphoto.com
Anonymous
February 21, 2005 10:11:50 PM

Archived from groups: rec.photo.digital (More info?)

Roland Karlsson <roland_dot_karlsson@bonetmail.com> writes:

>> Important for what? A larger aperture collects more photons. But it's
>> photons *per unit area at the sensor* that determines exposure time
>> needed.

>It is the number of photons per time unit and pixel that is important.
>If you have the same f-stop (e.g. F/4.0) and have a twice as large
>diameter of the aperture, you also have twice as large detectors for
>each pixel. Thus - you collect four times as many photons.

As I pointed out, there are a bunch of assumptions that affect whether
this is true or not. You're now being more explicit about one of them:
you assume that when you double the aperture diameter you also double
the pixel pitch, and thus keep the same number of pixels in the image.
That's true when comparing an 8 MP P&S camera to an 8 MP DSLR, but
many other combinations are possible. (I have a 4 MP P&S camera partly
because I don't think 7 or 8 MP makes sense with that sensor size.)
So, this assumption is sometimes true, but not always.

You're also assuming that, even though the double-size sensor has 4X
the light collection area, you only want the same number of electrons
for a "correct" exposure. That's true if you're happy with the same
signal-to-noise. If everyone was happy with that, DSLRs would have a
minimum ISO of about 800. But, in fact, having 4 times the area per
pixel means you have 4 times the full-well capacity at each pixel, and
you can usefully collect 4 times as many electrons to give you 2 times
better signal to noise. In other words, although you have 4 times the
collection area, you *want* 4 times as many photons for the best quality
exposure. In this case, the sensor sensitivity is not higher. That's
why DSLR ISO ratings start about 100, the same as P&S cameras, even
though the DSLR sensors have about 10 times as much area per detector.

>It is the diameter of the aperture that determines the light
>sensitivity of the system.

Assuming the same number of pixels in the final image, and assuming that
you don't want to make use of any of the extra full-well capacity of the
larger sensor. But those are both questionable assumptions.

Dave
February 21, 2005 10:52:25 PM

Archived from groups: rec.photo.digital (More info?)

rafe bustin wrote:

> On Mon, 21 Feb 2005 01:10:45 -0500, Stacey <fotocord@yahoo.com> wrote:
>
>
>>You ignore what while the per pixel noise may be higher, there are MORE of
>>them so on the same size image they aren't as visible. The end result is
>>more resolution of fine details if the optics are good enough.
>
>
> The Oly tries to get more pixels out of
> less silicon than the 10D. There's no
> free lunch, not unless you can change the
> laws of physics.
>

Who's claiming there isn't more noise at high ISO's?

What you and others seem to ignore is this has a lot more to do with the
design of the sensor than the small difference in sensor size. The FFT
kodak sensors are noisier but IMHO noise isn't ALL there is to image
quality. I guess if the only advantage your camera has is high ISO noise,
you tend to focus on that?

--

Stacey
Anonymous
February 22, 2005 12:20:06 AM

Archived from groups: rec.photo.digital (More info?)

davem@cs.ubc.ca (Dave Martindale) wrote in news:cvdbpm$m1f$1
@mughi.cs.ubc.ca:

> As I pointed out, there are a bunch of assumptions that affect whether
> this is true or not.

Hello! Don't make things too complicated :) 

Of course I assume taking the same picture, i.e. the same
number of pixels and the same FOV. That is enough to make it true.


/Roland
Anonymous
February 22, 2005 1:02:36 AM

Archived from groups: rec.photo.digital (More info?)

On Mon, 21 Feb 2005 19:52:25 -0500, Stacey <fotocord@yahoo.com> wrote:

>rafe bustin wrote:
>
>> On Mon, 21 Feb 2005 01:10:45 -0500, Stacey <fotocord@yahoo.com> wrote:
>>
>>
>>>You ignore what while the per pixel noise may be higher, there are MORE of
>>>them so on the same size image they aren't as visible. The end result is
>>>more resolution of fine details if the optics are good enough.
>>
>>
>> The Oly tries to get more pixels out of
>> less silicon than the 10D. There's no
>> free lunch, not unless you can change the
>> laws of physics.
>>
>
>Who's claiming there isn't more noise at high ISO's?

Who said anything about ISO? The Oly will have
more noise than a 10D or 20D at any ISO.

>What you and others seem to ignore is this has a lot more to do with the
>design of the sensor than the small difference in sensor size. The FFT
>kodak sensors are noisier but IMHO noise isn't ALL there is to image
>quality. I guess if the only advantage your camera has is high ISO noise,
>you tend to focus on that?


Small difference in sensor size?
Do the math, Stacey.

Canon 10D/20D sensor = 15 x 22.5 mm = 337.5 mm^2
E300 Sensor = 13 x 17.3 mm = 224.9 mm^2

That's 50% more area for the 10D...

337.5 / 6M = 5.6E-5 mm^2/pixel (10D)
224.9 / 8M = 2.8E-5 mm^2/pixel (E300)

Thats 100% more area per pixel for the
10D, as compared to the E300. That
translates to cleaner images, at any ISO.


rafe b.
http://www.terrapinphoto.com
February 22, 2005 3:22:13 AM

Archived from groups: rec.photo.digital (More info?)

rafe bustin wrote:

> On Mon, 21 Feb 2005 19:52:25 -0500, Stacey <fotocord@yahoo.com> wrote:
>
>
> Small difference in sensor size?
> Do the math, Stacey.
>
> Canon 10D/20D sensor = 15 x 22.5 mm = 337.5 mm^2
> E300 Sensor = 13 x 17.3 mm = 224.9 mm^2

SInce I always print 8X10, the difference is the 2mm in hight. I don't like
3:2 proportions, why would a I want a camera that shoots that ratio?
Obviously you do.


Again there is more to image quality than low noise to me, I guess for you
that's all that matters? How much of this "noise" are you going to ever
seen in a 8X10 print? I'm sure not seeing it.

--

Stacey
Anonymous
February 22, 2005 5:19:10 AM

Archived from groups: rec.photo.digital (More info?)

Roland Karlsson <roland_dot_karlsson@bonetmail.com> writes:

>Hello! Don't make things too complicated :) 

>Of course I assume taking the same picture, i.e. the same
>number of pixels and the same FOV. That is enough to make it true.

No it's not. You're assuming that if the photosites are 4X the area,
you will always shoot at 4X the ISO rating even though the photosites
have 4X the full well capacity and can deliver 2X better signal to noise
if you use it.

For an example closer to the real world: many DSLR cameras have sensors
with a pixel pitch about 3X that of the good P&S cameras. Thus, the
pixels have about 9X the area. Thus, if the P&S camera gives reasonably
clean images at ISO 100, the DSLR should give similar images at ISO 800
(the standard value closest to 900). And it probably does.

But you go further and assume that nobody will shoot below ISO 800, and
that the sensor *is* 9 times more sensitive by your definition. That's
not true. The normal definition of sensitivity of an electronic sensor
means exposing to some fraction of full-well capacity, rather than a
fixed number of electrons regardless of sensor area. So as area
increases, sensitivity does *not* increase - but signal to noise does
improve. And that's why DSLRs have minimum ISO ratings of around 100,
just like P&S cameras, even though the sensor cells are much larger.

If ISO is constant with photosite area change, you need a constant
f/number, NOT a constant aperture diameter, to keep the exposure the
same.

Dave
Anonymous
February 22, 2005 11:56:31 AM

Archived from groups: rec.photo.digital (More info?)

On Tue, 22 Feb 2005 00:22:13 -0500, Stacey <fotocord@yahoo.com> wrote:

>rafe bustin wrote:
>
>> On Mon, 21 Feb 2005 19:52:25 -0500, Stacey <fotocord@yahoo.com> wrote:
>>
>>
>> Small difference in sensor size?
>> Do the math, Stacey.
>>
>> Canon 10D/20D sensor = 15 x 22.5 mm = 337.5 mm^2
>> E300 Sensor = 13 x 17.3 mm = 224.9 mm^2
>
>SInce I always print 8X10, the difference is the 2mm in hight. I don't like
>3:2 proportions, why would a I want a camera that shoots that ratio?
>Obviously you do.
>
>
>Again there is more to image quality than low noise to me, I guess for you
>that's all that matters? How much of this "noise" are you going to ever
>seen in a 8X10 print? I'm sure not seeing it.



In May 2003 I made a concerted effort to
compare my Canon G2 against my Nikons,
loaded with Reala. This should have been
a no-brainer, after all there is a 5:1
difference in image size (measured in
megapixels) between the G2 captures and
the film scans.

Much to my surprise, the main difference
between the images was in shadow noise,
and to a lesser extent, resolution.
Given perfect light, either one could
make a great 8x10" print, or an acceptable
13x19" print.

This was most obvious in my sunrise/
sunset pix of the Grand Canyon, where
most of canyon was in deep shadow.

Mind you, even the film images had lots of
difficulty with shadow noise, but given
a choice, I find film noise somewhat less
annoying than blue-channel noise in
digicam captures.

As I recall, when I posted full res film
scans on the rpe.m-f way-back-when, your
initial reaction was "Eeeewwww... look at
the noise." As you know, with our friend
Dave L., banishment of noise is practically
a religion.

Were it not for noise, C41 film would rule,
and color slide film would wither. IMO,
noise is the *only* significant disadvantage
of C41 film.

Lack of noise is the only way (in my mind)
to explain why 11 Mpixels from a 1Ds can
match 55 Mpixels from a MF (645) film scan.



rafe b.
http://www.terrapinphoto.com
Anonymous
February 23, 2005 1:08:30 AM

Archived from groups: rec.photo.digital (More info?)

davem@cs.ubc.ca (Dave Martindale) wrote in news:cve4qu$207$1@mughi.cs.ubc.ca:

> For an example closer to the real world: many DSLR cameras have sensors
> with a pixel pitch about 3X that of the good P&S cameras. Thus, the
> pixels have about 9X the area. Thus, if the P&S camera gives reasonably
> clean images at ISO 100, the DSLR should give similar images at ISO 800
> (the standard value closest to 900). And it probably does.
>
> But you go further and assume that nobody will shoot below ISO 800, and
> that the sensor *is* 9 times more sensitive by your definition. That's
> not true. The normal definition of sensitivity of an electronic sensor
> means exposing to some fraction of full-well capacity, rather than a
> fixed number of electrons regardless of sensor area. So as area
> increases, sensitivity does *not* increase - but signal to noise does
> improve. And that's why DSLRs have minimum ISO ratings of around 100,
> just like P&S cameras, even though the sensor cells are much larger.

What you write is true. If you have a larger sensor you have a potential
of better dynamic range by not using higher ISO sensitivity. So, if you
use ISO 100 on both systems and F/4.0 on both systems, you will need a
bigger (in mm) aperture for the bigger system.

But - you will then not get the same picture - you will get a better
picture. If you want the same picture, you shall use ISO 100 on the
smaller system and ISO 800 on the larger. And then you shall use
F/2.0 on the smaller system and F/5.6 on the larger system, which both
have the same aperture size.

So - I maintain - it is the size of the aperture (in mm) that
determines the sensitivity if you want the same picture. If you
want the best picture - thats a totally different issue.


/Roland
Anonymous
February 23, 2005 8:22:37 PM

Archived from groups: rec.photo.digital (More info?)

Roland Karlsson <roland_dot_karlsson@bonetmail.com> writes:

>What you write is true. If you have a larger sensor you have a potential
>of better dynamic range by not using higher ISO sensitivity. So, if you
>use ISO 100 on both systems and F/4.0 on both systems, you will need a
>bigger (in mm) aperture for the bigger system.

>But - you will then not get the same picture - you will get a better
>picture. If you want the same picture, you shall use ISO 100 on the
>smaller system and ISO 800 on the larger. And then you shall use
>F/2.0 on the smaller system and F/5.6 on the larger system, which both
>have the same aperture size.

Actually, it's never "the same picture". If you use ISO 800 on the
DSLR, you can choose the same shutter speeds and different apertures as
you suggest, but the DSLR lens is probably sharper at f/5.6 than the
P&S lens at f/2.0. Or you could choose to use the same aperture but 8X
higher shutter speed on the DSLR, giving much better motion-stopping
and shallower DOF than the P&S can provide. Or you can shoot both at
ISO 100, giving better S/N and shallower DOF in the DSLR images.

No matter what you choose, you don't get "the same picture", and the
DSLR image can always beat the P&S camera in some parameter while
keeping the others the same.

>So - I maintain - it is the size of the aperture (in mm) that
>determines the sensitivity if you want the same picture. If you
>want the best picture - thats a totally different issue.

You're using a very narrow definition of "the same picture". You're
assuming equal S/N performance from the two cameras, while allowing other
things to vary. You're assuming everyone will shoot a DSLR at ISO 800
in order to get "the same picture", without stating that. But, in fact,
most people with a DSLR will use lower ISOs to get "better" pictures
much of the time, and that may have been their whole motivation for
buying the camera in the first place.

So, given a particular set of assumptions about keeping S/N constant,
your statement about aperture diameter equalling sensitivity is correct.
But you need to state your assumptions alongside the claim, particularly
when the conditions are not the way most people will actually use their
DSLRs.

Dave
Anonymous
February 23, 2005 11:00:58 PM

Archived from groups: rec.photo.digital (More info?)

davem@cs.ubc.ca (Dave Martindale) wrote in news:cvie4t$3i6$1@mughi.cs.ubc.ca:

> You're using a very narrow definition of "the same picture". You're
> assuming equal S/N performance from the two cameras, while allowing other
> things to vary. You're assuming everyone will shoot a DSLR at ISO 800
> in order to get "the same picture", without stating that.

You are assuming what I am assuming a bit too much :) 
Frankly, I cannot really regognice my own assumptions in
your assumptions about my assumptions :) 

My view is strictly a theoretical one - viewing the camera
as a black box - not caring what you find within. Then the size
of the hole that lets in the light is the meassurement of the
amount of photons per second you can catch. And this meassure the
theoretically best you can do. The size of the hole also
determines the theoretical diffraction limited resolution.
Assuming a given FOF, the size of the hole also determines
the DOF. And - I repeat - all those are meassures of the
theoretically best you can do.

In practice, you need som kind of sensor and lens. Neither of
those might be able to perform near to the theoretical limit.

For reasonable sizes of holes and sensors you can perform rather
near to the theoretical limit. Then my statement also holds
for real stuff.

BTW - I have found no faults in your posts - it is just that
your posts does not match my original statement. Of course -
if you have a better camera you want to take better pictures.
I have never challenged that.


/Roland
Anonymous
February 27, 2005 10:14:34 AM

Archived from groups: rec.photo.digital (More info?)

Olympus can build great camera's...Just like Canon/or Nikon...Take a look at
the 8080...In its class ..There is no match...

--
_________________-
BOCH
________________
A+TECH
_________
"Chris Brown" <cpbrown@ntlworld.no_uce_please.com> wrote in message
news:vo3me2-hdi.ln1@narcissus.dyndns.org...
> In article <1pfe11pfi90olroismd3pjt4l0op7php7k@4ax.com>,
> rafe bustin <rafeb@speakeasy.net> wrote:
>>On Sat, 19 Feb 2005 00:28:21 -0500, Stacey <fotocord@yahoo.com> wrote:
>>
>>
>>> OTOH it seems canon owners spend a lot of their time trying to convince
>>>"someone" THERE choice is the best for EVERYONE? I never have figured
>>>that
>>>one out. Post the word "Olympus" (or even Nikon) and David/Rafe run into
>>>the fray telling everyone how bad they are while neither have ever even
>>>held one? Just seem odd to me..
>>
>>
>>Nah, it's more like, where the word "Olympus" appears
>>there's likely to be a foolish post with your name on
>>it, implying that you're the only smart digicam buyer
>>on this forum. You're our new George Preddy.
>
> That's possibly a bit strong. Stacey has only just "gone digital",
> apparently after a road to Damascus type experience, and is understandably
> very enthusiastic about her new purchase. We've all been there, haven't
> we?
>
> GP, on the other hand, is a net.kook in the best tradition of them.
!