(aborted?) Sony 10.66 MP APS (D100, etc.) sized sensor

  • Thread starter Thread starter Ron Parr
  • Start date Start date
It's not an excuse. You can't make anything 100% perfect. And a silicon waffer is not an exception. There are allways defects on the waffer and if the IC is bigger there's an higher probability of lowering the yield. Does that raise the price 1%? 10%? 20%? 100%? I don't know but it's a fact than besides having less chips per waffer you will have more an higher probability of a defect.

But I think that someday DX will max out and even if at that time FF is confined to pro market the camera manufacturers will turn to FF for an advantage to keep bodies selling. Look what they have acomplished. A body used to last sometimes 10 years now people keep a body for 2 years or less. I believe they would like to keep this momentum foing.
you can't make 100% perfect silicon waffer (not contaminated) so a
bigger chip will allways be more sensitive to random defects
besides the smaller number per waffer difference
Stop making excuses for the chip makers please. Unless they can
show us the real numbers, no one has any idea what the actual yield
is. A lot of us are assuming that the yield is very bad because
cameras with full frame sensors are so expensive. The yield for
Kodak's sensor is good enough that its full frame digital is
selling for nearly half the price of the Canon. And Canon's sensor
has lower megapixel count (less chance for random defect that would
result in a bad chip), and its CMOS sensor has fewer circuits
because of its simpler design. Therefore Canon should get better
yield than Kodak. Yield isn't the reason why full frame sensors
are so expensive. The lack of competition is.
I would still prefer a FF sensor. But I don't know if will happen.
FF will allways be better but Dx sensors can win based on a
price/good enough combination.
It will happen. It is a matter of time. Competition can bring it
about much more quickly. Without AMD, we may still be using much
slower chips and paying higher prices for processors.
 
Well I agree with Anastigmat. We need inovation at the top end.
Just because 'you' don't need or understand other's need for
progress doesn't mean there shouldn't be any.
I am most certainly not against 'progress' and it is presumptuous of you to assume that I am. First, 'progress' is a very ill-defined word. Second, I do understand that other photographers may 'need' full-frame DLRs - eveyrone has different needs and requirements depending on their circumstances and their principal photographic subjects. I have stated this many times. What I object to is the manner in which many posters to these forums fail to understand my requirements and just because they do not need small, light and fully weather-sealed equipment they dismiss the views of those who do.

The prevailing assumption that a bigger sensor is, de facto, a good thing may be true for some people but it is not true for everyone. Some people argue that there are no advantages to a smaller sensor, but this is blatantly wrong. Smaller sensors do have real advantages, and they have real disadvantages. I just want these facts to be acknowledged - it is far from all one way traffic in favour of full frame.

There is no reason why full-frame based DSLRs and APS or 4/3rd based DSLRs should not co-exist to meet the different requirements of users, but many here seem to want to kill off the latter entirely and force us all down the full-frame route.
If you like small sensors (29% of the 35mm format) and you think an
E-1 is significantly smaller than a full-image-circle system then
go and buy one.
I do not like small sensors per se, but I like the consequences of using small sensors in terms of equipment size and cost.
I think it's way too big a compromise for the
barely noticible size advantage.
Well I think exactly the opposite - the huge extra bulk of full frame is a disadvantage that is not justified by the barely noticeable image quality advantage (either less noise at high ISO or greater resolution). For example, see:
http://www.luminous-landscape.com/reviews/cameras/8mp-alternatives.shtml

And, moreover, this advantage of full-frame will become less & less significant as sensor technology improves.
It's also not cheap.
I agree that the E-1 did seem expensive when it was first launched but now here in the UK it costs less than both the Canon 10D and Nikon D100, for what is a better featured body (speaking as someone who has owned both a 10D and a D100).
I actually
suspect Oly could have made it much smaller but it wouldn't look as
good for marketing reasons.
I actually agree with you that I also think that Oly could have made the E-1 smaller, but I don't know about your "marketing reasons" claim - you may be right, who knows?

I have pointed out before in these forums that the E-1 4/3rd sensor is actually very close in size to the old 110 film format (only a mm or two difference) and so, in principle, I would have thought it would be possible to produce a 4/3rd DSLR similar in size to the tiny Pentax 110 film SLR which was referenced in this thread and which I owned twenty years ago.

The real saving in weight & size comes in the lenses. At wide-angle the smaller image circle allows the lenses to be smaller (for example, compare the new reduced image circle lenses from Nikon, Sigma and Pentax with their 35mm equivalents), whilst at telephoto the FOV multiplication factor is a huge advantage (assuming that the smaller sensor has a similar total pixel count to the full-frame alternative so that cropping images from the latter would not give equivalent results since the resolution would be reduced unacceptably; the increased image noise at high ISO on the smaller more densely packed sensor is, in my opinion, a small price to pay for the advantages gained).


Then you can go take pictures and as your lens purchases will lock
you into the 4/3 sensor you won't feel the need to comment in
threads dealing with the exciting advances and price drops of 1.3x
and FF sensors. At least by then the E1 will be priced more
competitively with it's other consumer format brethren.
There you go again - it is wrong to automatically regard a smaller format as a "consumer format". This is sheer arrogance. You sound just like some old adherent of large format film critcising the new-fangled 35mm film format as being only good enough for 'consumers'.
I find it highly amusing the way the E-1ers pop up every time
sensor size is discussed.
Why am I an "E-1er"? I don't own an E-1. My main camera for the lats 15 months has been a Canon 10D! But, unlike some, I am not blind to the advantages of the E-1 and the 4/3rd system.
I imagine that in 12 or 18 months when
these threads are discussing the latest large sensors and their
wonderful high ISO performance, superb dynamic range, enlargement
capability etc etc those E1-ers will be strangely quiet, locked
away in their E1 forum enclave. The forums will be all the worse
for it. I miss them already.
This is absolute nonsense - obviously any advances in sensor technology (leading to "wonderful ISO performance, , superb dynamic range, enlargement capability etc etc") that are applicable to full frame sensors will be equally applicable to APS and 4/3rd sensors, and vice versa. Just because a sensor is smaller does not mean that it must use older technology. Indeed the improvements to which you refer will actually make the smaller sensor more attractive compared to the larger because the latter will hit the law of diminishing returns (in terms of image improvements that are actually visiible) before the former.

Terry.
 
All other things being equal, a larger sensor would do better, but
a new 10-11 MP APS sized sensor could still do better than an old 6
MP APS sized sensor.
Thank you. A larger sensor is indeed better if all else is equal.
Hence you can make a even better chip than your 10-11mp APS sized
sensor by scaling it to full frame.
But do you necessarily need it? I for one would happily swap my 14MP full frame camera for a 10-12MP DX camera if it offered similar resolution in a much smaller and lighter package. There would no noticeable difference in print quality at A3 and compact is much more important to me than absolute cost, yet you seem unable to accept that people might find this an perfectly fine compromise.
Erm... The technology for making such sensors at prices most
people would be comfortable paying probably doesn't exist,
That is just an excuse. There is no real world numbers to support
that claim.
I understand that there are some Nikon folks who have an interest
in rationalizing reduced size sensors as a way of explaining away
Nikon's lack of options. If it entertains you to taunt them, go
ahead.
I am not taunting them. I am pointing out the fact that since
Canon has a full frame sensor, Nikon cannot compete against Canon
if it is true to its word and never stray from the 1.5x format.
Nikon would instead be competing against the Olympus E-1 instead if
it never produces a full frame digital SLR camera.
Looking at the Canon Nikon lineups I'm struck by the way that the two manufacturers don't necessarily go head to head all the time.

For example on http://www.warehouseexpress.com the following prices are listed:

D2h £2300 1DMk2 £3500

D2x £3500 1Ds £4800

Now most people seem to accept that as a camera the D2h is slightly better than the 1DMk2 but that the Canon has the superior image resolution and noise performance. However, they are clearly not competing on price.

Again, although a lot of people might prefer the D1x over the 1ds ergonomically, it simply cannot compete on image resolution. On the other hand it is quite a bit cheaper.

It looks to me that Nikon is handling the resolution disadvantage they have in their flagship cameras by positioning them differently on price. This simply avoids the problem in the short term. Yes they have ceded the high end to the 1Ds but then again Canon doesn't have a £3500 pro model to compete on price with the D1x. Likewise with the D2h - Canon doesn't have a pro body to compete on price with this.

This actually gives the consumer more choice - higher resolution or lower price. Either way they get fine handling and rugged top quality bodies.

Are Nikon just being clever here in covering up their lack of competitiveness, I don't know...
 
In photography, quality is limited by the amount of pixels you can
print out and most of all see . The human eye has no infinite
resolution and so the maximum number of megapixels you "need" can
be calculated, assuming normal viewing distances. Guess what the
number was I read recently in a magazine: 5.88 megapixels.
So this is where we are already with current technology. Resolution
increases to 8 or even 12 megapixels will certainly be nice as we
get some headroom to crop and perform noise reduction, but any
further resolution increase won't help an uncropped picture much.
Many if not most cameras sold these days have zoom lenses that allow them to go beyond the standard field of view. This clearly indicates that people want the ability to capture more than what they can see with their eyes. Optical zooms are one relatively heavy and mechanically complex way to do this. Higher resolution sensors are another.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
It shouldn't polarize people. There is virtually nothing that a
small sensor can do that a large sensor cannot do better.
I generally agree, but small sensors will always have their place no matter how cheap big sensors get. Small sensors permit smaller cameras and there will always be people who are willing to trade quality for size. Also, small sensors are better for macro shots because you can get more DOF and one never seems to have enough DOF in macro shots.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
We all know that large sensors have lower yields. But there is no
real numbers to show exactly how low.
If you estimate the number of chip cripping defects per wafer, then the rest follows from mathematics. The only questions are the number of defects and what assumption you make about the distribution of defects. Defects typically are not distirbuted uniformly, which is actually a good thing.
Some people are guessing
that it is so dreadfully low that camera makers have to charge us
$8,000 for a camera that contains one such chip. I say the yield
is not so low because Kodak sells a full frame digital camera for
about half that price.
Both manufacturers are likely using experimental/bleeding edge fabrication techniques that involve different tradeoffs. This techniques also have development costs that must be amortized. Finally, as I've said before, Canon's higher quality has several consequences: They might have higher standards, which means more rejects and lower yields, or they might just have better technology and the luxury of charging more for their superior product.
The high price of the full frame Canon
therefore has more to do with a lack of competition than
manufacturing cost. The same can probably be said about the full
frame Kodak camera. Hence full frame cameras becoming the future
standard in digital cameras is not as hopeless as one may be led to
believe by those people who are making excuses for the chip makers.
The, "therefore," in the above paragraph is really precious. After criticizing others for guessing (when in fact they are generalizing from well known facts about semicondcutor manufacturing), you make your own guess and then proceed to make strong statements that follow from this guess as if they are facts.
The megapixel count probably has very little to do with it.
Yes it does. Everybody makes a big deal about big sensors having
more circuitry and therefore more chance that one of these circuits
may be bad, thus ruining the whole chip.
I have not heard any well informed people making such claims.

The problem with big chips is that they have a large area and, thus, a bigger chance of hitting a bad area on the wafer. This would be true regardless of what's actually in the mask.
So the more megapixels
the sensor has, the more likely it is to fail to pass inspections.
You haven't made a case for this yet. I can imagine a few arguments for how this might be true if your assumption is that the critical defects are coming from transmission lines failing and taking out entire rows. A chip with more pixels will have more transmission lines.

If discards are determined by the percentage of the total image that is performing below some standard, then it really doesn't matter since this will be determined by the percentage of bad areas in the substrate regardless of the number of pixels into which it is divided.
Thus it follows that Canon should get better yields than Kodak. At
the very least Canon should get about the same yield as Kodak since
their sensors are identical in size.
You need to make a lot more assumptions to push this argument through.
Nice strategy! Just claim that the yield is so low that full frame
cameras will never become affordable.
This is another one of your straw men. I certainly haven't claimed this.
Then claim that someone is
uninformed if they question the claim that yield can be so low.
I'm hypothesizing that you're not that informed on these issues because you keep saying things that aren't quite right.
Claim that yield is low because large chips have more circuits,
I never claimed this. In fact, this is the kind of statement that you have made which has led me to question your understanding of the process.
and
then deny the possibility that the Kodak chip (which have more
pixels and circuits than the Canon chip) could have a higher yield
than the Canon chip.
I find myself again puzzled about whom you think you're arguing with.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
Assuming you're talking about 35mm film derived SLR bodies with
reduced size sensors: The reason for pursuing improved reduced
size sensors is cost.
Yes, it is. It costs a lot of money to buy chips from a chip
maker. It costs a lot less if you make your own chips. Canon
makes its own chips. Therefore it has the lowest priced DSLR camera
on the market.
This is basically a string of non sequiturs.

It may, in fact, cost a LOT less to purchase sensors from another manufacturer than to make them yourself. This will depend upon many factors:
  • The relative expertise in manufacturing
  • The relative expertise in design
  • The investment in equpment made by both parties
  • The volume at which the chips are made and the existence of other customers.
  • The cost of capital for the different parties
  • Labor costs for the different parties.
It's much less expensive for Nikon to buy D70/100 chips from Sony than for Nikon to make them themselves. This is because Sony has expertise in manufcturing CCDs, while Nikon does not. Sony has CCD fabs, while Nikon does not. Moreover, Sony plans to sell this chip to other parties allowing them to spread of cost of development and of running the factory over several customers.

Canon has decided to make a long term investment in CMOS sensor manufacturing and this makes sense because it builds upon their expertise in lithography equipment. In the case of the 10D/300D sensors, they have bit the bullet and made a big investment in manufacturing. It makes sense for them to amortize this invesment and pursue their long term strategy, so FOR CANON, it is probably a wise economic decision to manufacture their own sensors at this time and they probably enjoy lower per unit costs than Nikon pays on the D100/70 sensor - although even this isn't clear. CMOS is supposedly cheaper, but Sony has huge economies of scale at their huge CCD/TFT fabs.
Problems with yield and limits on reticle
sizes can create a steep price curve when moving beyond roughly APS
size.
I am sure that big chips have lower yields, but no one has given us
any real numbers to tell us exactly how low. Therefore there is no
real numbers to justify the high costs of the cameras with big
sensors.
It's interesting how you view the lack of numbers as harmful to the conclusions drawn by your critics, but little hindrance for yourself.
I sure hope that Canon finds a way to produce a 1.3X or larger
sensor at a reasonable cost, but given the challenges I'm not going
to criticize people who are focusing on reduced size sensors.
They have found a way. It is called CMOS. CMOS chips are cheaper
to make than CCD sensors. Another way they found is to cut out the
middle man so they can get chips at a much lower cost than if they
buy it from somebody else.
The thing that you are ignoring here is the fact that nearly all computing devices have been CMOS for decades but that people have avoided making large chips because of the high costs and difficulties in manufacturing such chips.
BTW, I should mention that the relatively low cost of the Kodak
24x36 sensor is an anomaly in the sense that they have been
partnering with fabs (first Tower, now somebody else) that are
experimenting with chip stitching as a way work around reticle size
limits. (They don't stitch finished chips, but stitch the images
on the wafer.) It remains to be seen if these tecniques will be
effective in the long run or if they come at a price in terms of
quality. It seems that they are using a larger line width process
which gives them more slack but also has price in terms of quality.
I'd say the jury is still out on Kodak's experiment.
In the absence of real numbers on yield, your speculation is rather
pointless.
There's not much speculation there. There are many facts from which you should learn.
The size limit on pixels is based on physics.
Not really.
Okay, so chip sensors do not have to obey the laws of physics?
Your argument must obey the laws of physics.
Thank you. A larger sensor is indeed better if all else is equal.
Hence you can make a even better chip than your 10-11mp APS sized
sensor by scaling it to full frame.
This has been a point that everybody here as agreed upon from the very beginning,
Erm... The technology for making such sensors at prices most
people would be comfortable paying probably doesn't exist,
That is just an excuse. There is no real world numbers to support
that claim.
How tiresome. The difficulties in making large chips is a well documented fact. Sensors might be easier to make. That part is speculation.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
Hi Peter

I have no idea where this maximum number of pixels needed = 5.88 comes from but unless it is limited to some special kind of academic case it seems pretty meaningless to me.

I prefer to work on the more practical idea that what you "need" is enough pixels to max out the printer you are using at any given print size.

So for me for instance, I can't really see that much difference in detail between say a 12 inch wide and a 15 inch print from my Epson 1290 using an image from my 6MP D100 (for most subjects anyway) but I can see a distinct difference between 10 x 8 inch prints and 19 x 13 inch prints.

If I shoot on my 14MP Kodak instead, the resulting 19 x 13 inch prints are just as good as the 10 x 8s from the D100 - well they are after they have been post processed to deal with all the horrible Kodak colour artifacts!

What this means for me in practice is I can regard the 6MP as routinely sufficient for A4 size prints and acceptable for bigger prints whilst the Kodak is excellent all the way up the maximum size my printer can produce.

That's good enough for me as I have no intention of printing bigger than that (unless I get tempted by an Epson 4000, anyway...)

I don't see any obvious advantages in increasing pixel count much beyond 16MP or so for the current DSLR designs because the Kodak and the 1Ds are already challenging lens designs.
Exactly. The issue is that it's quite difficult to claim to have
insights into what "enough" will be 5-10 years down the road when
we don't know what the price/performance tradeoffs will be and
when, despite the bluster around here, none of us really knows what
Joe consumer's preferences are.
Well, for sure, the price/performace tradeoffs will shift and
therefore might influence what is considered to be "enough" in 5 or
10 years. However, in many fields there are some objective criteria
which describe the "enough" point. In these cases people just enjoy
the sinking prices. For example with audio, CD quality is
"enought". There are few humans who really can hear "better"
(though many claim to do so). And tests have shown that for many
people a good MP3 cannot be distinguished from a CD, so the CD
quality for a lot of people even is better than "enough". That is
one of the reasons, why all new audio formats "better" than CD have
failed so far.
In photography, quality is limited by the amount of pixels you can
print out and most of all see . The human eye has no infinite
resolution and so the maximum number of megapixels you "need" can
be calculated, assuming normal viewing distances. Guess what the
number was I read recently in a magazine: 5.88 megapixels.
So this is where we are already with current technology. Resolution
increases to 8 or even 12 megapixels will certainly be nice as we
get some headroom to crop and perform noise reduction, but any
further resolution increase won't help an uncropped picture much.
Peter
 
The depth of field advantage is beneficial to landscape photographers too but the difference between 35mm and 1/2 frame is rather less than what you see from small digicams so that one is pretty neutral for me.
It shouldn't polarize people. There is virtually nothing that a
small sensor can do that a large sensor cannot do better.
I generally agree, but small sensors will always have their place
no matter how cheap big sensors get. Small sensors permit smaller
cameras and there will always be people who are willing to trade
quality for size. Also, small sensors are better for macro shots
because you can get more DOF and one never seems to have enough DOF
in macro shots.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
You are telling me that I will lose the argument because I have too
much supporting evidence? That is certainly possible but I would
rather take that chance than to have too little supporting evidence.
NO! Rather that you keep switching the point of attack back and
forth between different supposedly compelling points which suggests
you don't entirely have the confidence they are compelling!
I am merely answering the points that my opponent is raising. If I am switching, then I do it in response to my opponent's constant switching. Therefore it is my opponents who have no confidence in their own arguments. I have absolute confidence in my arguments.
It shouldn't polarize people. There is virtually nothing that a
small sensor can do that a large sensor cannot do better. Even the
supporters of the small sensor would agree.
I do not agree!! I have a 1/2 frame and a full frame camera and
whilst I now use the Kodak almost exclusively, I still recognise
the drawbacks of the full frame. The fact we are having this debate
demonstrates the polarisation...
The drawbacks of the full frame? I think you mean the drawback of the Kodak camera, including its bulk and slow write speed. I presume. Full frame cameras need not be bulky if, say, Pentax is making the camera. Other "drawbacks" of the full frame sensor are in fact attributable to the camera or the quality of the sensor, not the size of the sensor. The same "drawbacks" would exist if you scale back the Kodak sensor to APS size. The same drawbacks will not exist if you scale up the APS sized sensor in your camera, so that the pixel size remains the same. About the only drawback is the large files and slower write speed. But for most people this is rarely a problem if the camera can write at a speed of 1-2 frames per second.
Canon will likely start selling a 1.3x
8mp camera at a price just a couple hundred dollars more than the
D70. Prices will drop to $1k or below by this time next year.
Again you may be right but this doesn't have the ring of informed
analyst, merely wishful thinking.
It is not wishful thinking at all. The D70 is eating Canon's market share big time. Canon's market share went from a lofty 80% to less than 40% in a single month, the month the D70 was introduced. Canon's market share has rebounded slightly but it still trails Nikon 46 to 52% according to June sales figures. The D70 outsells both the 10D and Digital Rebel. Both of these have 1.6x sensors. To take back market share, Canon has to make a camera that is noticeably better than the D70, since a lower price has not helped the Digital Rebel outsell the D70. Canon has an advantage in having large sensors over just about every other camera maker except Kodak. If Canon puts a 1.3x format sensor in the 10D replacement, and improves its write speed (not a difficult thing to do since the 1d MkII is quick), then this camera would be better than the D70. Even naive consumers can see a difference in quality since the new Canon will have more megapixels. Canon can put the new 10D in a smaller body with more plastic to differentiate it from the 1D MkII. Such a camera nevertheless would not outsell the D70 if it is priced at $2,000, as some posters insist Canon must do because of the price of the sensor. Since Canon makes its own sensor, it can make the sensor quite cheaply since it does not have to pay high profit margins that other chip makers charge other camera makers. A 1.3x 8mp 10D MkII selling for slightly more than the D70 is possible. When Nikon drops the price of the D70 to respond to Canon's challenge, it will squeeze Digital Rebel sales further unless Canon drops the price of the 10D MkII to around $1K in response. Hence it is likely that a 1.3x format camera will be selling for $1k this time next year.

Wishful thinking? Not at all. It is a prediction based on current market conditions and existing products. Canon may not be willing to do what I said it may do. But if it does not, canon may need to be content with being second place to Nikon in market share, since the D70 continues to lead the Digital Rebel in sales according to the latest figures.

In contrast, those who claim that Nikon will not make a camera with a full frame sensor are engaging in wishful thinking. Nikon cannot compete unless it can offer a product competitive against Canon's larger sensors.
Secondly, a huge factor in favour of small sensors is that cameras
and lenses will be possible.
That is just nonsense. Why are full frame sensor cameras and lenses not possible? Both are here and now.
To echo your own comments, as soon as
the technology is there, competition will drive down the size of
cameras and APS will always be capable of being smaller and lighter
than full frame.
Most consumers do not want smaller cameras and lighter cameras if it means sacrificing image quality. That is why the APS film format died. It has a smaller negative. Your argument that APS sized sensor cameras can be smaller is based on a comparison between the Kodak and D70, but it is an invalid comparison. If Pentax makes a full frame camera available later this year, it will almost certainly be smaller than the D70 in size and certainly smaller than the Kodak. If the camera shrinks any smaller than the *ist D, there is the chance that some photographers may not be able to handle it comfortably. A camera the size of the *ist D with the image quality of a full frame sensor is a combination that no 1.5x format camera can beat.
 
It shouldn't polarize people. There is virtually nothing that a
small sensor can do that a large sensor cannot do better.
I generally agree, but small sensors will always have their place
no matter how cheap big sensors get. Small sensors permit smaller
cameras and there will always be people who are willing to trade
quality for size.
I agree. The APS sized sensor will eventually find its way into point and shoot cameras, where compactness is more important than image quality. The sensors in these types of cameras are hitting the ceiling on megapixel count because noise has become noticeable at ISO settings higher than 200. A camera that would help a point and shoot compact shoot at ISO 800 or even 3200 would be a strong competitor if most cameras in this category tops out at ISO 400.

The APS sized sensor will also be used in bottom of the line digital cameras, somewhat analogous to cameras like the Nikon N60 in the film world.
Also, small sensors are better for macro shots
because you can get more DOF and one never seems to have enough DOF
in macro shots.
That is unfortunately not true. Depth of field depends on the focal length lens being used and the aperture being used. If you use the same lens and the same aperture, then depth of field remains the same regardless of the size of the sensor. Using a smaller sensor is like cropping a negative. You cannot increase depth of field by cropping a negative or a slide.
 
Unfortunately for the both of you, depth of field remains the same for people who use the same lens and the same aperture, even if the sensors in their cameras are different. A 400mm lens will have the same shallow depth of field no matter what camera it is mounted on. The small sensor is equivalent to involuntary cropping. Take the negatives you have and crop them. The cropped images won't give you more depth of field.
The depth of field advantage is beneficial to landscape
photographers too but the difference between 35mm and 1/2 frame is
rather less than what you see from small digicams so that one is
pretty neutral for me.
Also, small sensors are better for macro shots
because you can get more DOF and one never seems to have enough DOF
in macro shots.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
Unfortunately for the both of you, depth of field remains the same
for people who use the same lens and the same aperture, even if the
sensors in their cameras are different. A 400mm lens will have the
same shallow depth of field no matter what camera it is mounted on.
The small sensor is equivalent to involuntary cropping. Take the
negatives you have and crop them. The cropped images won't give
you more depth of field.
With a smaller sensor, you use a shorter focal length lens to achieve the same field of view and therefore get greater DOF for the same composition. This is basic stuff.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
Also, small sensors are better for macro shots
because you can get more DOF and one never seems to have enough DOF
in macro shots.
That is unfortunately not true. Depth of field depends on the
focal length lens being used and the aperture being used. If you
use the same lens and the same aperture, then depth of field
remains the same regardless of the size of the sensor. Using a
smaller sensor is like cropping a negative. You cannot increase
depth of field by cropping a negative or a slide.
You forgot that with a smaller sensor, you use a shorter focal length lens to achieve an equivalent field of view.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
Some people are guessing
that it is so dreadfully low that camera makers have to charge us
$8,000 for a camera that contains one such chip. I say the yield
is not so low because Kodak sells a full frame digital camera for
about half that price.
Both manufacturers are likely using experimental/bleeding edge
fabrication techniques that involve different tradeoffs. This
techniques also have development costs that must be amortized.
Finally, as I've said before, Canon's higher quality has several
consequences: They might have higher standards, which means more
rejects and lower yields, or they might just have better technology
and the luxury of charging more for their superior product.
Yes, they have the better technology and thus the "luxury of charging more." That has been my point all along. Most of the apologists claim that yield is so low that Canon would lose money unless it charges so much per camera. Absent real numbers to show actual yield from the chip makers, these arguments amount to little more than excuses these people are making on behalf of the chip makers, who actually know the actual yield but who are not telling us.
The high price of the full frame Canon
therefore has more to do with a lack of competition than
manufacturing cost. The same can probably be said about the full
frame Kodak camera. Hence full frame cameras becoming the future
standard in digital cameras is not as hopeless as one may be led to
believe by those people who are making excuses for the chip makers.
The, "therefore," in the above paragraph is really precious. After
criticizing others for guessing (when in fact they are generalizing
from well known facts about semicondcutor manufacturing), you make
your own guess and then proceed to make strong statements that
follow from this guess as if they are facts.
If you read your previous paragraph, you can see that you have reached a similar conclusion, since you claim that Canon may have the luxury of charging more because it has the better product. But since I said it, you are compelled to disagree, even though it is in agreement with your own conclusion.
The megapixel count probably has very little to do with it.
Yes it does. Everybody makes a big deal about big sensors having
more circuitry and therefore more chance that one of these circuits
may be bad, thus ruining the whole chip.
I have not heard any well informed people making such claims.
Go read the Extreme Tech article on digital sensors, it makes such a claim. More circuits means more chance that there is one bad part that ruins the whole chip in the big sensor.
The problem with big chips is that they have a large area and,
thus, a bigger chance of hitting a bad area on the wafer. This
would be true regardless of what's actually in the mask.
So you are claiming that the manufacturing process is perfect, and that the only way a defect can occur is if the material is defective?
So the more megapixels
the sensor has, the more likely it is to fail to pass inspections.
You haven't made a case for this yet. I can imagine a few
arguments for how this might be true if your assumption is that the
critical defects are coming from transmission lines failing and
taking out entire rows. A chip with more pixels will have more
transmission lines.
That is strange. You are claiming that defects in the material can result in chip failure or a chip not passing inspection, but a manufacturing failure because some of the circuits failed to be properly formed cannot happen? Both kinds of defects would cause part of the chip to not function. And even if the defect is in the material, a small sensor packing parts in higher density is going to result in a larger percentage of its parts failing than the larger sensor with less densely packed parts. So a reasonable person may conclude that the large sensor is more tolerant to defects in the material than a small sensor, if a small number of bad parts may not result in a bad chip.
 
If discards are determined by the percentage of the total image
that is performing below some standard, then it really doesn't
matter since this will be determined by the percentage of bad areas
in the substrate regardless of the number of pixels into which it
is divided.
That is not so. A small sensor will have a larger number of pixels affected by the same size defect than a large sensor.
Thus it follows that Canon should get better yields than Kodak. At
the very least Canon should get about the same yield as Kodak since
their sensors are identical in size.
You need to make a lot more assumptions to push this argument through.
Nice strategy! Just claim that the yield is so low that full frame
cameras will never become affordable.
This is another one of your straw men. I certainly haven't claimed
this.
If they can become so inexpensive that they will become the standard, then the APS sized sensor will not persist. You are denying that you claim the full frame will never become affordable on the one hand, and arguing that the APS sized sensor will become the standard. My argument is quite simple. The APS sized sensors are being used because of artificially high prices for full frame sensors, not because the APS sized sensor has any advantage other than price. When full frame sensors come down in price and the cameras built around them becomes affordable, the APS sized sensor will disappear from most DSLR cameras.
Then claim that someone is
uninformed if they question the claim that yield can be so low.
I'm hypothesizing that you're not that informed on these issues
because you keep saying things that aren't quite right.
You can prove that I am wrong about the yield by giving me the real numbers coming out of real factories making real big sensors. You have none, but you insist that my analysis is not right. Your argument therefore is based on your belief, not on facts. And your belief appears not to be based on facts.
Claim that yield is low because large chips have more circuits,
I never claimed this. In fact, this is the kind of statement that
you have made which has led me to question your understanding of
the process.
So, you are now arguing that the number of circuits have nothing to do with failure rate then. The defect is in the material only. Well, can you tell me why a full frame sensor, which is only about twice as large as an APS sized sensor, should have a failure rate that is much higher than 2x the failure rate of APS sized sensors? Twice the area means doubles the chance of encountering a fatal defect in the silicon wafer. That means the full frame should only fail at twice the rate of APS sensors during manufacturing. It is also true that fewer full frame sensors can be packed onto the same wafer. Let's be generous and assume that twice as many APS sensors will fit on the same wafer as would full frame sensors. That means the difference in yield ratio is 2x2 or 4x. Full frame sensors should only cost 4x as much to make as the 2x sensor. So, can the difference in yield account for the price difference between a Canon EOS 1DS and the Canon Digital Rebel? Of course not! It shows that even if you are correct, your conclusion is still wrong.
and
then deny the possibility that the Kodak chip (which have more
pixels and circuits than the Canon chip) could have a higher yield
than the Canon chip.
I find myself again puzzled about whom you think you're arguing with.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
Okay, so you claim that you did not make that statement. Since both chips are full frame, they should have identical yield ratios. If so, low yield can be eliminated as a reason for the price difference between the Kodak and the Canon full frames. My conclusion is that the Canon costs more because of a lack of competition. You have yet give us any facts that can falsify that argument. So you claim that I am uninformed. If you are so well informed, then please inform us why the Canon should have lower yield and perhaps exactly how low is the yield of the Canon photosensor. If you are unable to do that, then you are the one who is uninformed, not me.
 
There is no reason why full-frame based DSLRs and APS or 4/3rd
based DSLRs should not co-exist to meet the different requirements
of users, but many here seem to want to kill off the latter
entirely and force us all down the full-frame route.
Agreed.
Well I think exactly the opposite - the huge extra bulk of full
frame is a disadvantage that is not justified by the barely
noticeable image quality advantage (either less noise at high ISO
or greater resolution). For example, see:
http://www.luminous-landscape.com/reviews/cameras/8mp-alternatives.shtml
Whaaa? If this were tennis I'd be asking.. YOU CANNOT BE SERIOUS. You are joking aren't you? From that pic of the three cameras you think the other two (not FF but sized to accept a FF image circle) have "huge extra bulk"? And you think that a sensor less than 1/3 the size would have barely noticible differences in noise at high ISO or in resolution?

If you are serious then theres really no point in persuing this. That article is full of BS. If you cant see te difference in image quality in the pics then I don't know what. I suggest you post in the Canon 1D forum and tell the pros there they are stupid for not using a Powershot camera. They could live for a month on the savings - they'd have to, they'd have no income.
And, moreover, this advantage of full-frame will become less & less
significant as sensor technology improves.
I agree that the E-1 did seem expensive when it was first launched
but now here in the UK it costs less than both the Canon 10D and
Nikon D100, for what is a better featured body (speaking as someone
who has owned both a 10D and a D100).

I actually agree with you that I also think that Oly could have
made the E-1 smaller, but I don't know about your "marketing
reasons" claim - you may be right, who knows?

I have pointed out before in these forums that the E-1 4/3rd sensor
is actually very close in size to the old 110 film format (only a
mm or two difference) and so, in principle, I would have thought it
would be possible to produce a 4/3rd DSLR similar in size to the
tiny Pentax 110 film SLR which was referenced in this thread and
which I owned twenty years ago.
Here is where we agree totally. If the E1 were in fact as small as that Pentax 110 system (which I remember gazing at lovingly in the local camera shop window as a child) then I would very happily say what a fantastic contribution it to the available choices it is.
The real saving in weight & size comes in the lenses. At
wide-angle the smaller image circle allows the lenses to be smaller
(for example, compare the new reduced image circle lenses from
Nikon, Sigma and Pentax with their 35mm equivalents), whilst at
telephoto the FOV multiplication factor is a huge advantage
(assuming that the smaller sensor has a similar total pixel count
to the full-frame alternative so that cropping images from the
latter would not give equivalent results since the resolution would
be reduced unacceptably; the increased image noise at high ISO on
the smaller more densely packed sensor is, in my opinion, a small
price to pay for the advantages gained).
We'll agree to disagree then. Theres no way I'll buy into a camera/lenses with an sensor of only 28% when the bodies/lenses are fully 90% of the size.
There you go again - it is wrong to automatically regard a smaller
format as a "consumer format". This is sheer arrogance. You sound
just like some old adherent of large format film critcising the
new-fangled 35mm film format as being only good enough for
'consumers'.
Maybe you are right on your last sentence. If I were getting married again (no thanks) and the wedding photographer told me on the phone he was going to use a 110 camera I'd be saying no thanks and looking for someone else preferably with medium format. Would't you? Why would a pro carry something that has a 28% sensor?
I find it highly amusing the way the E-1ers pop up every time
sensor size is discussed.
Why am I an "E-1er"? I don't own an E-1. My main camera for the
lats 15 months has been a Canon 10D! But, unlike some, I am not
blind to the advantages of the E-1 and the 4/3rd system.
To me an E1er is an E1 apologist umm sorry advocate often rigorously defending Olympus and the E1 as though they have been brainwashed by the marketing hype. I think it's just a little bit of sensor-envy :) I've described it as a cult before.
This is absolute nonsense - obviously any advances in sensor
technology (leading to "wonderful ISO performance, , superb dynamic
range, enlargement capability etc etc") that are applicable to full
frame sensors will be equally applicable to APS and 4/3rd sensors,
and vice versa. Just because a sensor is smaller does not mean
that it must use older technology. Indeed the improvements to
which you refer will actually make the smaller sensor more
attractive compared to the larger because the latter will hit the
law of diminishing returns (in terms of image improvements that are
actually visiible) before the former.
Terry, the fact that we agree on the issue of the Pentax 110 size issue demonstrates that we both see the value of small sensors when used wisely. The only reasons for using small sensors would be to enable the smaller systems or cheaper systems. To me the E1 is too much of one and not enough of the other two.

Yes small sensors will improve in quality along with large sensors but at any point in time large sensors will outperform small sensors. Nothing too controversial there I hope.
 
Whilst I agree that the E1 is bigger than it could have been given the sensor technology I have a couple of observations:

1. It clearly borrows an awful lot from the design of the E10/E20 which were the same size but used the much smaller 2/3 sensor. Oly just seem to have a thing about this particular design and as a former E10 owner I can say it was very nice to hold and use....

2. But seriously, the E1 was initially marketed as a camera to compete with the likes of the D1x not the d100 or 10D so they needed to make it look good and impressive. It is pretty small in comparison to other pro targeted cameras. The next generation 4/3 cameras will likely be smaller.
 
I'm not so ignorant I need to be lectured on these basics, you know ;-)

Obviously the statement carried the implication you would be using an appropriate focal length for the format.

Anastigmat, it would be best if we tried to exchange information and inform opinions instead of trying to generate argument just for the sake of it wouldn't you agree?
The depth of field advantage is beneficial to landscape
photographers too but the difference between 35mm and 1/2 frame is
rather less than what you see from small digicams so that one is
pretty neutral for me.
Also, small sensors are better for macro shots
because you can get more DOF and one never seems to have enough DOF
in macro shots.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
Now your slight hectoring is revealing your own (partial) ignorance on this matter. There are more factors than just focal length and aperture to consider e.g. subject distance, final required print size and sensor size.

The sensor size plays a part because one variable in the DOF equation is the required Circle of confusion - and that is partly dependent on the image magnification which is determined by print size and sensor size.
It shouldn't polarize people. There is virtually nothing that a
small sensor can do that a large sensor cannot do better.
I generally agree, but small sensors will always have their place
no matter how cheap big sensors get. Small sensors permit smaller
cameras and there will always be people who are willing to trade
quality for size.
I agree. The APS sized sensor will eventually find its way into
point and shoot cameras, where compactness is more important than
image quality. The sensors in these types of cameras are hitting
the ceiling on megapixel count because noise has become noticeable
at ISO settings higher than 200. A camera that would help a point
and shoot compact shoot at ISO 800 or even 3200 would be a strong
competitor if most cameras in this category tops out at ISO 400.

The APS sized sensor will also be used in bottom of the line
digital cameras, somewhat analogous to cameras like the Nikon N60
in the film world.
Also, small sensors are better for macro shots
because you can get more DOF and one never seems to have enough DOF
in macro shots.
That is unfortunately not true. Depth of field depends on the
focal length lens being used and the aperture being used. If you
use the same lens and the same aperture, then depth of field
remains the same regardless of the size of the sensor. Using a
smaller sensor is like cropping a negative. You cannot increase
depth of field by cropping a negative or a slide.
 

Keyboard shortcuts

Back
Top