An interesting data point on Canon's sensor costs

[david_littleboy__tokyo__ja]

What Kelly said: the _fixed_ costs in getting a sensor to market are enourmous. My bet is that Canon actually made a per-camera profit on the original 1Ds but lost their shirts over the life of the product. It was a loss leader, and a brilliant one: lots of folks (me included) might have looked more closely at the Nikon and other low-end dSLR offerings (the lack of a spot meter irritates) were it not for Canon's apparent commitment to full frame.

 

My take is that the original 1Ds sensor was never intended to be anything other than a loss leader and proof of concept.

 

However, the 1Dsmk2 sensor is another story. It really looks as though Canon ought to be able to turn a profit with that sensor in a US$3,000 body. The only thing stopping them is the desire to not compete with themselves: a $3,000 EOS 3 class camera with that sensor would steal market from both the 20D and the 1Dsmk2, both of which are probably higher-profit products than a $3,000 EOS 3 class dSLR would be.

 

The monkey wrench in the works here is the Sony 12MP sensor. Sure, when the balanced reviews come in, we'll see that the noise even at ISO 100 is enough to limit dynamic range and that ISO 800 is a mess, but if Nikon puts that sensor in a D70 class body, Canon's in serious trouble: for that many pixels a lot of folks can forgo radical contrast masking (which is the only place the ISO 100 noise will be a problem) and use a different camera for low-light work.

 

So Canon needs an answer, and I'm hoping that'll be a $3,000 EOS 3 class dSLR with the 1Dsmk2 sensor...

 

By the way, I hope someone does a (sensible, fair, unbiased) comparison between the 17-40 on the 1Dsmk2 and the 12-24 on the 12MP Nikon. The 12-24 will need to provide 1.5x the on-sensor resolution (more accurately, an MTF pattern shifted by 1.5x to the right) to provide the same image quality, and that might be hard.

 

In other words, the people who think that the cropped sensors "crop away the bad parts" are forgetting that they need 1.5x the performance to match that of the larger sensor.

[ron c sunshine coast,qld,a]

Please excuse my grumpiness above-it mostly comes from seeing alot of false info being thrown about.

<P>There is actually some good info here,i'll go through it later 'cause i'm no good with even simple numbers today : )

[mark u]

Some points answered:

 

Camera weight: You can buy a Nikon D70 that weighs 21 oz, or a D2X (well, soon maybe) that weighs 2.4 lb for the same crop factor (both excluding batteries and lens). There's no reason why a less heavily engineered full frame DSLR shouldn't weigh in at much the same as current 1.5/1.6 crop DLSRs. Canon and Nikon produce (full frame) film SLRs that weigh around 13 oz. - the same as an EOS IX Lite (APS film SLR).

 

Full frame sensor vignetting: This is much more to do with sensor optics and their interaction with lens design (particularly with very fast or wider angle lenses). Improve the design of these, and the problem will be manageable. We have already been told by Canon that the next unifying 1 series body will be full frame.

 

Sensor stitching: the stitching is done by moving the mask over the wafer, not by physically welding two sensor halves together. Positional accuracy in the order of a few nanometers when doing this is now achieveable according to Canon, so yields are barely affected by the procedure.

 

Replacement sensor cost: Although it helps set an upper boundary, I would guess that the $500 quote is a price and not a cost (i.e. it includes a profit margin). Moreover, it is very likely that instead of replacing just the sensor, the entire main board would be replaced, which includes most of the other more expensive components.

[jon_austin]

Finally, a sane, rational and quantitative discussion of a subject in which many of us are highly interested.

 

I expect to replace my 20D in about 4 or 5 years with a new, Canon-brand, full-frame dSLR for less than US $2K.

[chiswick_john]

The AA filter infront of the sensor is very,very expensive - that is one of the reasons why the Kodak full frame DSLR is so cheap compared to the Canon - it does not have one.

[tuan_tran]

Easy math guys assuming that the cost of both bodies are about the same in price.

 

1Ds: Body + 3 * Sensor = $8000

1D: Body + Sensor = $4500

 

or

 

1Ds: Body = $8000 - 3 * Sensor

1D: Body = $4500 - Sensor

 

Substiting one for the other,

 

$4500 - Sensor = $8000 - 3 * Sensor

 

3 * Sensor - Sensor = $8000 - $4500

 

2 * Sensor = $3500

 

Sensor = $3500 / 2 or $1750

 

So

 

1Ds: 3 * ($1750) = $5250 for the sensor

1D: $1750 for the sensor

[mark u]

That math is not a very useful guide - besides, it implies that a 300D sensor costs the whole price of the camera. We know that Nikon has been making money from the D70 - it was explicitly stated that it was responsible for a useful part of their tripled profit in April-September 2004. We also know that overall the camera division of Canon made sales of 7.34 $bn, and an operating profit of 1.26 $bn, a margin of 17.1% (down from 19.3% in 2003). 69% of those sales were digital cameras of all types, with just 16% being film cameras and lenses (third party competition in lenses must be very strong), and 15% being videocameras. Management commentary on their financial results notes:

 

"as the digital camera market matures, we seek to further expand sales of high-value-added digital SLR cameras, which are differentiated from the competition by such innovations as our independently developed CMOS sensor. We also aim to expand our profits and market share in the compact digital camera segment. Here, we are focusing on further improving product features and strengthening cost competitiveness through such means as reducing the number of parts required and in-house production."

 

Camera division sales increased by 16.8% over 2003, yet numbers of digital cameras sold increased by 60% - a reflection of the severe price erosion and competition at the bottom of the market which is also commented in the report. The evidence from Canon's financial reports is pretty strong that they enjoy some very healthy margins on those elements of the DSLR market where they face little or no competition - which is no more than you would expect. I wouldn't be at all surprised if margins on the 1 series cameras accounted for over 40% of the selling price. Budding financial analysts can draw their own conclusions:

 

http://www.canon.com/ir/results/2004/rslt2004e.pdf

[mark u]

Another small observation in relation to Chiswick John's comment about the cost of AA filters: The cost of AA filters is probably much more closely proportional to their area than sensors are. The filter for a 1.6x crop DSLR therefore can't be much less than 40% of the cost of a full frame filter. Since the cost of the filter has to come out of the $500 not accounted for by the cost of a 300D body, which also has to cover the cost of the sensor itself and the other DSLR specific items (DIGIC etc.) and profit margins, that actually puts limits on the cost on the cost of a filter for a full frame sensor, even if you assume higher quality filters are used at say double the cost per square inch of filter.

[dogbert]

"I don't know how marketers get over this hurdle, but they did it with VCRs, CD players, Computers and Camcorders, all of which started out at 10x their current price and with 1/10 of the current capabilities."

 

It aint a hurdle and it aint a dillemna. It is simple economics that people by more when the price is lower. What we have seen with VCRs, CD players and the rest is just moving along a demand curve.

 

The big factor in the price fall of all of these was commoditisation of the various parts. Once the parts for making these things is available from multiple suppliers the prices drop. This happened once the technology stopped being proprietry and moved out of Japan and basically became open source. You can now get a CD player for about $25 in the US, when the original ones cost $3000 in 1986 dollars. Similarly digital 2 megapixel P&Ss are cheap because the parts have been commoditised.

 

DSLRs are a niche market and the more niche they are the higher the profits will be. Note from Canaon's statements that they are trying to make profit at the high end by differentiating their product (eg FF sensors) while at the low end they are trying to cut costs faster than their competitors - there is little scope to differentiation in the P&S market.

 

All of this suggests that the key to getting cheaper FF cameras will be competion from other manufacturers.

[tuan_tran]

--That math is not a very useful guide - besides, it implies that a 300D sensor costs the whole price of the camera.

 

I was assuming the bodies of the 1D and 1Ds are about the same in price. When you hold these constant and equal, it makes sense to me. The body of the 1D <> (not equal) 300D and the sensor of the 300D is several times less than the 1D.

[mark u]

Here's an alternative breakdown:

 

Body cost (excl. digital related - from EOS 1V less a profit margin) $1300 for both cameras.

 

Digital electronics and AA filter - $600 for 1Dmk II and $900 for 1Ds Mk II (allowing for extra cost of AA filter for larger sensor)

 

Sensor cost $750 for 1D MkII and $2250 for 1Ds MkII

 

Totals: $2650 for 1DMk II, and $4450 for 1DsMk II

 

Profit (including retail margin): $1850 for 1D Mk II (41% of retail) and $3550 for 1Ds Mk II (44% of retail).

 

There is no way Canon are selling these cameras at or below cost!