why????

[lol]

given nobody makes such fast zooms, it must be difficult to get decent image quality out of them, even with a smaller (APS-C) sensor

Kinda - if you go to the slightly smaller sensor FourThirds world, they have two f/2 zooms: 14-35mm and 35-100mm. Sit down before you look at the price though.

I've also seen some long f/1.4 or was it f/1.8 zooms in C-mount (think it was a Sony 18-100mm) which was tempting me but someone else beat me to it. They cover an even smaller image circle than 4/3 though.

I think f/2 zooms in APS-C are very possible, but they would be of even shorter range than existing f/2.8 zooms and the price would be rather insane. So overall you might as well stick to the f/2.8 zooms or get the faster primes.

As for the long term survival of APS-C DSLRs, that depends entirely on what happens with mirrorless systems. Can they gain enough traction to displace APS-C DSLRs, leaving only full frame DSLRs taking a niche at the high end. Or will mirrorless fail to make an impact, where APS-C DSLRs will remain a sweet spot for a long time? We'll have to wait and see.

[Rocky]

why does no one make say a 16-50mm f2.0 or something similar for crop cameras sure it would be expensive but i think people would be willing to pay for it, i would pay 70-200mm is f2.8 is money for it.

On a side not canon should make an L-s range for good crop for lenses like this. i aim fairly certain within a few years full frame will be less popular with the ridiculous iso performance of  new crop cameras and the possibility of photo binning (don't ask me about it i'm a noob but basically you have option to reduce megapixels to get better iso performance)

The lens that you want  is an excellent walk around lens. However, it may be a monster in size and weight. just look at the 17-50 f2.8 EF-S. It is 4.4 inches long, using 77 mm filter and weights 1.4 lbs. If it was f2.0 and 16- 50 mm, it might be using 90mm filter or larger and weighed over 2 lbs easily.  The 16-35 already uses 83 mm filter. How many people will like to have a monster like tah tas a walk around lens and can be spotted from far away???

[Mt Spokane Photography]

why does no one make say a 16-50mm f2.0 or something similar for crop cameras sure it would be expensive but i think people would be willing to pay for it, i would pay 70-200mm is f2.8 is money for it.

On a side not canon should make an L-s range for good crop for lenses like this. i aim fairly certain within a few years full frame will be less popular with the ridiculous iso performance of  new crop cameras and the possibility of photo binning (don't ask me about it i'm a noob but basically you have option to reduce megapixels to get better iso performance)

The lens that you want  is an excellent walk around lens. However, it may be a monster in size and weight. just look at the 17-50 f2.8 EF-S. It is 4.4 inches long, using 77 mm filter and weights 1.4 lbs. If it was f2.0 and 16- 50 mm, it might be using 90mm filter or larger and weighed over 2 lbs easily.  The 16-35 already uses 83 mm filter. How many people will like to have a monster like tah tas a walk around lens and can be spotted from far away???

Since he is rerferring to APS-C, I assume he means EF-s, which would be about the size and weight of a 16-35, but it would likely cost more.

[Rocky]

and that would mean $30 per APS-C sensor (166 working chips out of a $5000 wafer), which sounds pretty reasonable, and tells me this format is not going away anytime soon

You must have been using the number from the white paper by Canon. the APS-C potential site is REALLY wrong. My estimate is that there may be only 75 sites for the 8 inch wafer for APS-C sensor. Let us assume the yield for APS-C is 75 %, then we will have 56 sensor per wafer, $5000 per wafer, that will be $90 per sensor. Assuming the yield for FF is 25% (1/3 of the yield  of APS-C, being pressimistic ). then each wafer will yield 5 FF sensor. that will be $1000 per sensor. That seems about right.  If wecompare the price of &D to 5D MkII, It is about $800 difference.
My personal opinion is that the Canon white paper is used to justify the high price of the FF body. The site of the FF sensor on the wafer may be 22 to 24. That will bring the yield up to be 6 sensor per wafer. and the difference in cost will be $770 instead of $1000.We are assuming that the wafer cost is $5000. Which is extremely high for a CMOS process. What if the wafer cost is $2500, than the difference  of FF sensor and APS_C snesor will only be $450 or even $370.

[Canihaspicture]

Canon uses 300mm wafers

[bvukich]

Haven't they moved up to 12"/300mm wafers yet?

I haven't been able to find any current pricing,  the only reference to 12" wafer pricing I could find was from 2003; and at that time they cost about $200USD, and 8" were $30-40USD.  12" were relatively new at that point, and I would assume (perhaps incorrectly) that the cost per in^2 of 8" vs. 12" has since reversed.  Although I'm sure actual prices have gone way up.  The feature widths back in 2003 were 2-4x the sizes used today, so probably needed less perfect wafers.

I had always been under the impression that the cost of the actual silicon though, while not trivial, was a relatively small part of the equation.  It's the $1B+ that a fab costs, along with labor, litho equipment, masks, design/engineering, etc. that truly set the cost to get a good die out the door.  And the fact that a fab is basically a fixed capacity manufacturing facility, only compounds that.  You can't just move faster to increase output, you have to increase yield to have any gains.

[Rocky]

Canon uses 300mm wafers

I am glad that someone confirms that Canon is using 300MM wafer. I have been suspecting it for a while but I have no way to find out. If Canon is really using the 300mm wafer, then the price gap between APS-C and FF sensor will even be smaller due to: 1. the cost per unit area of finished wafer is smaller. 2. The ratio of site between FF and APS-C is closer due to the ratio between perifferal partial site to whole site is smaller. May be the difference in final cost can be down to around $200.

[Rocky]

why does no one make say a 16-50mm f2.0 or something similar for crop cameras sure it would be expensive but i think people would be willing to pay for it, i would pay 70-200mm is f2.8 is money for it.

On a side not canon should make an L-s range for good crop for lenses like this. i aim fairly certain within a few years full frame will be less popular with the ridiculous iso performance of  new crop cameras and the possibility of photo binning (don't ask me about it i'm a noob but basically you have option to reduce megapixels to get better iso performance)

The lens that you want  is an excellent walk around lens. However, it may be a monster in size and weight. just look at the 17-50 f2.8 EF-S. It is 4.4 inches long, using 77 mm filter and weights 1.4 lbs. If it was f2.0 and 16- 50 mm, it might be using 90mm filter or larger and weighed over 2 lbs easily.  The 16-35 already uses 83 mm filter. How many people will like to have a monster like tah tas a walk around lens and can be spotted from far away???

Since he is rerferring to APS-C, I assume he means EF-s, which would be about the size and weight of a 16-35, but it would likely cost more.

17-50 EF-S and 16-35 EF is already the same length and weight. I was using the 17-50 EF-S fiter size of 77mm
as starting point and use  the 18-55 EF-S (F3.5) 58mm filter size and the 16-35mm Filter size (77 mm) to guestimate the filter size. If between f3.5 and f2.8 (both EF_S) with increase of 19mm in filter size, I  guest an increase of only 13 mm between f2.8 and f2.0 is very stingy.  I have not yet taking the increase in viewing angle between 16mm and 17mm into  account. With increase in the size of optical elements, the weight will increase also. It fact I may have already under estimate the size  and weight of the 16- 50mm APS-C f2.0 lens.

[Edwin Herdman]

I don't see Canon (or any other manufacturer) ditching APS-C size sensors simply because they can fit many more of them on the wafer.  This gives them more leeway to deal with increased reject rates and simply put out more for the same cost.  Even if the larger wafer sizes could be produced more cheaply than the smaller ones, the cost effectiveness of larger vs. smaller chips is always (absent some strange, rare geometric configurations between similarly-sized chips where more of a smaller chip can't be produced than of the larger) going to be in favor of the smaller chips.  Cheaper wafers might bring down FF sensor costs but it'll bring down APS-C costs, too.

[Canihaspicture]

I think 4/3 will die an early death when manufacturers start going mirrorless APS-C.... I also believe DSLRs will always favor the full-frame for professionals and anybody who prefers high IQ. Full frame might also pull back some of the medium format market after the next round of products come out.

[recon photography]

Canon's 17-55mm f2.8 is really fine for me but for the price you would have to be crazy not to get the sigma os or the tamron non vc version unless you either have heaps of money or are committed to crop sensor, maybe its just over priced in Australia, its $1,500 here vs $300 for tamron or $600 for sigma OS 

[Hillsilly]

Will M4/3 die an early death?  I'm not sure.

As a recent m4/3 buyer, I thing I can understand Canon's reluctance to enter the mirror-less market.  My initial thoughts on an EPL1 are overall  positive.  The picture quality is very good.  At ISO 200 in good light I doubt that you coud tell the difference between M4/3 and Canon's APS-C.  It also allows you to experiment with odd lens combinations - I've started to re-use all my old Minolta MC and MD lenses which has given them a new lease on life.  (They work surprisingly well).

But there a number of downsides.  Battery life , focus speed, and a minor time lag between real life and the image you see in the viewfinder.  Also, FF lenses just don't sit well.  They're too big.  If Canon was to try to enter the market with a camera that rectified these problems, they'd be using a bigger sensor, bigger battery, more processing power.  I think they'd still end up with a camera that was T3i sized.  In which case, what's the point?

Therefore, I think Canon probably have the right strategy.  From a useability perspective, mirrorless has no real benefits over the G12 / S95.  Picture quality would be better, but given that it would also have a lot of disadvantages to the lower end DSLRs, why enter the market with a new line that needs marketing, R & D expenditure and support when your current products are better performers? 

The sales figures of M4/3 show that they aren't big sellers anyway.  Its a very niche market.  The only buyers seem to be odd people like myself that just want to play around with new toys, want something small that provides quality photos and use unusual lenses (plus, my camera looks cool!).

Therefore, my prediction is that Canon won't release a mirrorless camera in the near future.  As such, M4/3 will survive for a long time to come. 

[Canihaspicture]

If they release a G12 successor with an APS-C to compete head on with the Fuji X100 I think they've got game.

[noisejammer]

Problems with Vignetting and edge & corner softness exist even in the most expensive lenses, they are much much less of an issue with crop bodies.

Umm... no. Vignetting and corner behavior are properties of the lens, not of the sensor.

Vignetting - If you take a crop sensor and install a lens intended for a ff camera, you will have less vignetting. If you select a medium format lens and use it on a ff camera, you have the same result. If you use an EF-S lens on an EF-S sensor, you get vignetting too.

Your comment on corner softness is not factually correct (but it demonstrates why autofocus is limiting.) I'm predominantly a Zeiss user and it's easy to show that "corner softness" is almost always a result of field curvature. If focused at the corner - say using live view - they are as crisp as the centre is normally. This means that the determining factor is usually how well your subject suits your lens' properties.

[neuroanatomist]

Problems with Vignetting and edge & corner softness exist even in the most expensive lenses, they are much much less of an issue with crop bodies.

Umm... no. Vignetting and corner behavior are properties of the lens, not of the sensor.

Flake clearly understands that.  "Problems with Vignetting and edge & corner softness exist even in the most expensive lenses," i.e. the problems are with the lens, but a crop sensor mitigates those issues.

Your comment on corner softness is not factually correct (but it demonstrates why autofocus is limiting.) I'm predominantly a Zeiss user and it's easy to show that "corner softness" is almost always a result of field curvature. If focused at the corner - say using live view - they are as crisp as the centre is normally. This means that the determining factor is usually how well your subject suits your lens' properties.

Sorry, but you're incorrect.  Corner softness is sometimes the result of field curvature, but from an optical design standpoint, it's simply more difficult to produce a lens that's as sharp in the corners as in the center, because light at the periphery of the image circle must be refracted more strongly.  If you look at the MTF data on a site like Photozone.de, you can see 'corner softness' represented quantitatively.  They take field curvature into account, as they state in their FAQ, "If a lens suffers from field curvature and/or residual aberrations (see below) this is taken into account - in this case the corners are measured independently from the center using different reference images."  So, they are doing exactly what you recommend, i.e. focusing on the corners when measuring corner resolution, and still coming up with less resolution than in the center for most lenses (including Zeiss lenses).