moondigger wrote:
Koivulehto wrote:
1. It would be very straightforward to implement 2.0 crop in software, i.e. to calculate the mean value (or even something a bit more sophisticated) of 4 pixels, or

That's not a 2.0x crop. It's averaging a 2x2 matrix of pixels into a single large pixel, sometimes referred to as binning. It wouldn't result in a 2.0x crop factor -- the full image circle would still be used, and the only difference between an image taken in "regular" mode and an image taken in this mode would be that this mode is lower resolution (5 megapixels) with (possibly) better low-light performance. There would be no telephoto effect at all. It would also be a lot more work for the CPU to handle, as it would have to capture all the pixels then perform calculations on each 4x4 matrix prior to writing to the card. Seems counterproductive if the idea is to increase frame rate.

2. It might alternatively be feasible to implement 2.0 crop directly in the sensor, i.e. to calculate the mean value on the sensor chip.

This would avoid the CPU overhead problem, but again isn't actually a crop.

Sorry about not thinking clearly after a hard day at work (it's now evening here ...). No beer this time, so I am running low on excuses.

So this leaves us no 2.0 crop, and thus no connection what so ever with the rumor. Nevertheless, the alternative 2 (= binning in sensor) would have some benefits:
- This might make some sense for high speed mode, like 8.5 fps (or whatever the mirror & shutter can do), and even a higher optional speed (like 16 fps) if the mirror is let to stay up after the 1st image & the shutter is capable for this speed. That would result in dark viewfinder during the burst. Of course, this kind of high speed implementation is not at all probable, but might be possible in theory.
- Better low noise - at least this binning principle has been used in microphone etc. sensitive pre-amplifiers to reduce their noise.
- Some existing or announced P&S might already use this binning IIRC

For the actual 2.0 crop as suggested in the rumor, I can't think of any concrete reason why it couldn't be any other crop instead. I mean, if Canon would decide to modify the sensor chip to collect only the center part of the image in high speed mode, they could include exactly as many pixels as they like.

i would be incredibly disapointed if canon did use a 2x crop mode such as nikons d2x..just looking through a 1.6x crop is claustrofobic.

Everybody here thinks combining the two camera would be easy for Canon and makes perfect economic sense for Canon.

Here are the facts but I could be wrong:
1. FF sensor are much more expensive than either 1.3x or 1.6x sensor. How much none of us know.
2. Canon 1D MkII image size of the sensor is 36,834,048 bytes.
3. Canon 1D MkII needs to move 36,834,048 bytes off the sensor at 8.5 times / second because 8.5 fps. So the memory needs to be pretty darn fast. I think the PC4200 can easily handle this because it can handle 525MB/sec.
4. Canon 1D Mark II's buffer needs to be big enough to hold 22 of these images so it needs to hold 810,349,056 bytes.

Design facts:
1. Canon CPU must be fast to process the images. It needs to apply proper sharpening, color adjustments, etc. The CPU cannot heat up, cannot have a fan and must be small enough to fit inside a cameara.
2. Canon Memory must be small and fast too. At the same time, it also need to run cool.

Now putting those things into account and using a 5D sensor, can Canon make it at $4500 price point or less? How about $6000? At what price point would PJs not buy the new camera? If they don't, what other options do they have?

Expanding on the binning concept, why not something like this:

Combined 1d-II and 1Ds-II (let's call it the 1Dc for combined), with 24 megapixels total.

For PJ use, one could elect to have the camera spit out a 6 megapixel JPG (24 / 4) from the binning used to create the lower-res, but lower-noise file.

If the camera could do THIS at 8 fps, then it could surely ALSO do 8 fps in the full 24 megapixel mode. After all, the camera still has to READ all 24 megapixels off the sensor before it can do any of the binning.

DavidP wrote:
If the camera could do THIS at 8 fps, then it could surely ALSO do 8 fps in the full 24 megapixel mode. After all, the camera still has to READ all 24 megapixels off the sensor before it can do any of the binning.

That's what I was getting at previously when I said binning would impose more CPU overhead than just processing the full resolution anyway. If Canon wanted to save CPU cycles as a way to offer higher framerates at lower resolution, they'd have to implement the lower resolution as a crop, not by binning. In that scenario, they would ignore the outer areas of the sensor and just process the center 6 or 8 megapixel block, as if the camera actually had a smaller sensor. This would introduce a crop factor in the same way that using a smaller sensor does, but it would be user-selectable. 22 or 24 megapixels at 3-4 frames/second, or 6 - 8 megapixels at 8+ frames per second.

For various reasons I doubt Canon will combine the actual cameras that way, and think it more likely the 1.3x crop will go away in favor of a full-frame 12-13 megapixel camera that can handle 8 fps. If they used a non-moving pellicle mirror instead of a standard mirror, maybe they could even push the frame rate to 10 or 12 fps.

tuantran wrote:
Everybody here thinks combining the two camera would be easy for Canon and makes perfect economic sense for Canon.

While I think such a scenario is possible, I don't think it will happen or that it makes economic sense to do so. However I do think the 1.3x crop will be eliminated in favor of a full-frame camera.

1. FF sensor are much more expensive than either 1.3x or 1.6x sensor. How much none of us know.

We may not know the exact numbers, but we do know a full-frame camera can be profitably sold for under $2000. Whatever the specific requirements are for CPU and memory at 8 megapixels/8 fps, we already know Canon can handle it because that's what the current 1DMk2 does with the DIGIC II. It is reasonable to expect that the DIGIC III will allow greater throughput. 12 or 13 megapixels at 8 fps does not seem unreasonable to me.

4. Canon 1D Mark II's buffer needs to be big enough to hold 22 of these images so it needs to hold 810,349,056 bytes.

Well, it doesn't have to hold quite that much -- it's flushing to memory at the same time the pictures are being taken. But in any case the 1DMk2 is already handling this load.

1. Canon CPU must be fast to process the images. It needs to apply proper sharpening, color adjustments, etc. The CPU cannot heat up, cannot have a fan and must be small enough to fit inside a cameara.
2. Canon Memory must be small and fast too. At the same time, it also need to run cool.

Agree, and again I'll point out that this is already being done with the 1DMk2.

Now putting those things into account and using a 5D sensor, can Canon make it at $4500 price point or less?

I don't think there's any question that they can. 12.7 megapixels means approximately 50% speedup and memory increase; if they can't do that 18 months after the 1DMk2 hit the market for the same price, then Moore's law falls apart. However in the computer world (my area of expertise) Moore's law continues to hold true - an approximate 100% speedup/capacity increase every 18 months. Meeting a 50% target should be relatively easy.

Pricing isn't as closely tied to actual production costs as many people think it is. Most retail electronics sell at a 400% markup or more over the cost to produce them; some itmes sell at a 1000% markup over the production cost. Now it's true that advertising, distribution, middle-men, retailers, etc get a piece of the pie, but when Canon can offer a $600 rebate on a 5D already selling for $700 under MSRP, you know there's plenty of margin on the camera. I believe a 1DMk3 based on the 5D sensor as we've discussed here could easily retail for $4000 and make big profits for Canon's DSLR division. Heck, a year after its introduction I wouldn't be surprised to see it for under $3000 on one of Dell's stackable coupon + double rebate offers.

So Russ, I get it that you're pretty confident that the 1D Mark III will be full frame and under $4500?

If you're right and if you're in San Francisco, I'll buy you dinner at Boulevard.

Are the following items 400% markup?
1. Hard Drives
2. PS3, XBox360
3. DVD Players
4. LCD monitors
5. Compact Flash Cards
6. Plasma TV, LCD TV
7. Laptops, Desktops

I live in Silicon Valley and have friends who can get me some of the above products (HP, Microsoft, Samsung, Panasonic, Western Digital, Lexar, Sandisk and Sony) direct from the manufacturers and most of the time, I get like 10% off the retail price and those are employee discount.

I'll be there to and I'll buy the drinks and take a pictures of you all with the new 1D Mark 3
I hope your right because that's exactly how I picture the next 1D

I guess I can take another person with me too

We'll make it a photo adventure with our new Mark III. Sweet.

tuantran wrote:
So Russ, I get it that you're pretty confident that the 1D Mark III will be full frame and under $4500? If you're right and if you're in San Francisco, I'll buy you dinner at Boulevard.

It's a guess, though I think it's a reasonable possibility. Of course I've been wrong before and I'll be wrong again.

Unfortunately I live on the wrong side of the country to accept the dinner invitation, even if my guess is correct.

moondigger wrote:
full-frame 12-13 megapixel camera that can handle 8 fps. If they used a non-moving pellicle mirror instead of a standard mirror, maybe they could even push the frame rate to 10 or 12 fps.

I'd go for one of those in a 1-series body. Not so much for the increased fps, but for the QUIET and no viewfinder blackout of the pellicle mirror. And hopefully they could increase the signal/noise ratio just enough to offset the light loss of the pellicle mirror, so we'd end up with images about the same noise level of the 1D-II and 5D.

moondigger wrote:
DavidP wrote:
If the camera could do THIS at 8 fps, then it could surely ALSO do 8 fps in the full 24 megapixel mode. After all, the camera still has to READ all 24 megapixels off the sensor before it can do any of the binning.

That's what I was getting at previously when I said binning would impose more CPU overhead than just processing the full resolution anyway. If Canon wanted to save CPU cycles as a way to offer higher framerates at lower resolution, they'd have to implement the lower resolution as a crop, not by binning. In that scenario, they would ignore the outer areas of the sensor and just process the center 6 or 8 megapixel block, as if the camera actually had a smaller sensor. This would introduce a crop factor in the same way that using a smaller sensor does, but it would be user-selectable. 22 or 24 megapixels at 3-4 frames/second, or 6 - 8 megapixels at 8+ frames per second.

For various reasons I doubt Canon will combine the actual cameras that way, and think it more likely the 1.3x crop will go away in favor of a full-frame 12-13 megapixel camera that can handle 8 fps. If they used a non-moving pellicle mirror instead of a standard mirror, maybe they could even push the frame rate to 10 or 12 fps.

Didn't anyone notice the alternative binning on the sensor chip in the posts yesterday? I am still not claiming that it would be probable in a Canon 1D series, but with a built-for-purpose sensor it would be possible, and it would only need to read 1/4 of all pixels from the sensor chip to the Digic III or whatever processing chip.

tuantran wrote:
Are the following items 400% markup?
1. Hard Drives
2. PS3, XBox360
3. DVD Players
4. LCD monitors
5. Compact Flash Cards
6. Plasma TV, LCD TV
7. Laptops, Desktops

Many of them, yes. The component items (hard drives, RAM) don't tend to be marked up as much. However remember that I said 400% over the cost of building a given device, not 400% over what it can profitably be sold for.

Remember how much DVD players cost in the late 90s? I don't mean when they first came out, when they were $700 and up. I mean after they had been established and people were really getting interested in them. Prices for a name-brand DVD player like a Sony or Panasonic were around $300. People felt like they got a great deal if they found one on sale for $250. Does anybody think it actually cost any more than $50 to build one? I can assure you it didn't. In fact I'd be surprised if it ever cost Sony more than $30 to build a DVD player -- parts and labor included.

Now I believe the cost of building a camera body is much higher than a DVD player. It wouldn't surprise me if the cost of building a single 5D is in the low hundreds of dollars. But that's a far cry from $4500.

I live in Silicon Valley and have friends who can get me some of the above products (HP, Microsoft, Samsung, Panasonic, Western Digital, Lexar, Sandisk and Sony) direct from the manufacturers and most of the time, I get like 10% off the retail price and those are employee discount.

Companies aren't in the business of providing super amazing discounts to their employees. If employee discounts were as significant as you think they should be, there would be too much incentive for employees to buy tons of stuff and sell it at a profit. Invariably when a company provides its employees with really significant discounts, there's a limit on the amount they can get. I once qualified for a huge discount on Apple Macintosh computers, but I was limited to one at the special price per year. Even so, one of my coworkers bought the most expensive Mac at the time (a II fx) and sold it at a profit to a mutual acquaintance. Imagine if those prices were available with no limit...

The point I'm trying to make is that I sometimes see comments from people, usually when the Dell stackable discounts happen at the same time as double or triple rebates, that they can't understand how Dell or Canon could make a profit with such steep discounts and rebates. The answer is deceptively simple -- the cameras and lenses actually cost much less to build and distribute than the MSRP.

Here's a different example. Canon sells three 50mm lenses at three very different price points. 1.8 for $70, 1.4 for $300, and 1.2 for $1500. Anybody think the 50/1.4 actually costs 4 times the 50/1.8 to build? Anybody think the 50/1.2 actually costs 21 times as much as the 50/1.8 to build?

I have no doubt that the 50/1.2 costs more to build than either of the other lenses, especially if you consider the recent design/development costs that they have to make up. But they're all just so much glass, plastic, and metal. They're not paying the line workers who build 50/1.2s more than they're paying those that build 50/1.4s or 50/1.8s. In fact they're probably the exact same workers. They're not paying more for polycarbonates to use in one lens than any other. Grinding special elements costs more, and fluorite costs more than glass for lenses that have it, but there's just not enough of a difference to account for the pricing difference.

Successful companies charge what the market will bear. It might cost them $90 to build a 50/1.2 (just a wild guess, don't quote me) -- but if people will pay $1500 for it because it's an L and for the improvements in the focus motor and a half stop of light, that's what Canon will charge.

moondigger wrote:Pricing isn't as closely tied to actual production costs as many people think it is.

True, but don't forget other costs. R&D, packaging, advertising, overhead (employees), etc. So yes, a $6000 camera could cost just $1500 to make when you add up the parts. That does not give a real picture. Those other costs of business can't be ignored. So the money made over production cost gets eaten up elsewhere.

Most retail electronics sell at a 400% markup or more over the cost to produce them; some itmes sell at a 1000% markup over the production cost.

All depends on how you define 'produce'. Even thin I seriously doubt these figures. Apple, the envy of the electronics industy for its high margins is not even that high. I think they are 70% at best, 40% at the low end. There was a company that did a cost analysis on the iPod and found the $200 nano had parts alone that cost $70 even with big volume discounts. Add in additional manufacturing expenses and packaging (more than most people think) and the iPod came would be about a 200% markup.

Now it's true that advertising, distribution, middle-men, retailers, etc get a piece of the pie, but when Canon can offer a $600 rebate on a 5D already selling for $700 under MSRP, you know there's plenty of margin on the camera.

You are assuming too much. They could be breaking even after rebate or even losing m oney. Remember, buying a DSLR is investing in a brand. Canon wants you on the Canon bandwagon. That way you buy their other items...lenses, flash, and other bodies.

I believe a 1DMk3 based on the 5D sensor as we've discussed here could easily retail for $4000 and make big profits for Canon's DSLR division. Heck, a year after its introduction I wouldn't be surprised to see it for under $3000 on one of Dell's stackable coupon + double rebate offers.

Again, speculation all based on assumptions of canon's pricing and profit strategy. Facts are Canon 1 series cameras have always been a premium and people still buy them. We don't know how much Canon makes on anything they sell. And we have no idea what this next camera will look like.

What does this all mean...this post, your post, and this entire thread is a waste of time and energy. dang it...

Anyone here can look at the SEC filing and will see that the profit is actually 15.5% for the camera division of Canon.

Canon's SEC Filing 3rd Quarter

tuantran wrote:
Anyone here can look at the SEC filing and will see that the profit is actually 15.5% for the camera division of Canon.

Canon's SEC Filing 3rd Quarter

Yes, but that doesn't tell us what any particular individual item costs to build and sell, and losses are factored in too. Paying lawyers to file patent applications, other legal fees, etc knock the profits down. There's a wide range of margins depending on the product too. A 50/1.8 might have 40% margin, while an 85/1.2L might have a 500% margin or more.

Yes, but that doesn't tell us what any particular individual item costs to build and sell, and losses are factored in too. Paying lawyers to file patent applications, other legal fees, etc knock the profits down. There's a wide range of margins depending on the product too. A 50/1.8 might have 40% margin, while an 85/1.2L might have a 500% margin or more.

Yes, but as a company, you have to include the lawyers, regulation etc... as well as some loser products when you are pricing your merchandise. If overall the profit for the Canon camera division is 15.5%, I could tell you for sure they are not selling their gear with a 400% markup.

chez wrote:
If overall the profit for the Canon camera division is 15.5%, I could tell you for sure they are not selling their gear with a 400% markup.

You can't tell us any such thing "for sure" unless you work for Canon making such decisions. We're all just speculating here but I believe I have plenty of evidence to back my speculations.

Obviously they have different margins (markups) on different pieces of gear. I'm fairly certain they're charging at least 400% markup on the exotic lenses like the 50/1.2 and 85/1.2, if not more.

One can't draw conclusions about the markup on any single item that a company sells based on their SEC filings, unless they specifically talk about that one single item in the filing.

The point I'm trying to make is that I sometimes see comments from people, usually when the Dell stackable discounts happen at the same time as double or triple rebates, that they can't understand how Dell or Canon could make a profit with such steep discounts and rebates. The answer is deceptively simple -- the cameras and lenses actually cost much less to build and distribute than the MSRP.

Well, look at the price differences worldwide for the 5D. The initial MAP was $3200 in the US, but it sold for significantly more than that in Europe, and more in the UK than on the Continent (all this I gather from what people have reported here on FM).

The consistent price (barring regional rebates, seasonal store sales, and other promotional gimmicks) has fallen modestly in the US, but significantly on the Continent...yet only very slightly in the UK.

This suggests to me that there was always a lot of room between the wholesale price Canon Inc is charging the regional Canons and the MAP. I'll bet the wholesale price hasn't even changed since the camera was introduced.

Wholesale price of the 5D is about 2600 as of last time I checked. I strongly doubt it costs Canon much less than $2000 (optimistically speaking).

There is margin in electronics, but the most margin I have seen is for stuff like speakers (a $400 speaker costs $200 at wholesale prices) and home theatre (generally have a 33% margin)

I'm sure it's nice to think that it costs Canon almost nothing to make a 5D and they are simply guoging customers. If you really want to get good prices, get a part time job selling cameras. The reps will give you vouchers for great deals. My coworker got a killer price on a Sony HDMI reciever (lower than wholesale...about %40 off) by taking a quiz on Sony products and ordering directly from Sony.

I sure hope it has IS built into the body!

In the real world, you need to put everything into account. You need to pay the engineers that designed the lens. You need to pay the leasing cost for the building that the engineers are in. You need to buy them equipment and software. Then you need to pay the testers and manufacturing team to setup the facility. This can go on. The 50L doesn't magically get designed overnight without these cost associated with it. When you add the fact that not many will buy that lens, then it would take much longer to recoup the cost.

If 1 engineer worked on that lens for say three years and Canon is paying $100,000/year for that engineer, how many lens would then need to sell before they recoup $300,000? How many lens if they were selling half the selling price?

ok, I got to page 10 and I can't go on reading all the rest of this...

The thing I love about Canon digitals is that we all have choices. I personally love the 1.3 crop: I get a little wide, and I get a little long. I'm not so much a fan of my 1.6 crop, even though I'm a telephoto shooter. However, there are others who use 600mm lenses with only crop sensor bodies. Then there are people who like ultrawides, or who grew up using a 50mm when it actually looked like a 50mm in the resulting print. With Canon, we can buy whichever one fits our style of shooting.

That's the thing - none of us are the same. I persoanlly see myself as getting only 1.3 and ff sensors in the future. Some of you will stick to 1.6 crop. Some will move to 1.3 crop. Others will go for film-sized sensors. And there are the people who have 2 or 3 different sized sensors.

We don't ALL have 1.6 crop cameras, from lowest end to highest end.

I'm damn glad we don't.

Oh, one thing about PJ's - I dunno about what papers you guys work for, but the final images that are usually submitted are under 1MP for the papers I work for. I don't need more MP - I need better DR, better high ISO, maybe a bigger buffer at the highest quality settings. What I don't need is a more expensive camera, which a full frame that has a crop mode would have to be.

I expect all threee sensor sizes to stick around for at least another generation or two. Putting a full-frame sensor into a body where the majority of users will never use it is just a bad idea - it will cost more (to manufacture, as well as for us), which will lose them sales. Places that buy these cameras in batches, like large newspapers, won't like that, and you can be sure Canon will hear about it. Again, just because you like FF does not mean that it's right for everyone.

MPerdomo wrote:
Wholesale price of the 5D is about 2600 as of last time I checked. I strongly doubt it costs Canon much less than $2000 (optimistically speaking).

There's no way it costs them more than a few hundred dollars to build each one (parts & labor). Tack on money to recoup R&D costs, plus distribution costs, plus advertising costs, and amortize the amount it will cost them to do free (warranty) repairs at the expected defect rate, and you'll arrive at the total cost figure for Canon. I seriously doubt it's more than $1000. Then they sell them to dealers for whatever wholesale cost is appropriate. $2600 might be accurate, or (more likely, IMO) that's what the price list said when they first came out. Now it's probably less. But even if it isn't, that doesn't mean $1600 of pure profit for Canon. They're offering $600 double rebates right now, which means ~$640 comes out of the profit column to cover the rebates and the contract with the rebate redemption contractor. Now the profit is down to under $1000, on a high-margin item.

I'm sure it's nice to think that it costs Canon almost nothing to make a 5D and they are simply guoging customers.

I never said Canon was gouging customers. Successful companies charge what the market will bear for a given item. Companies that don't die off. There's a huge markup on the 85/1.2L compared with the cost of producing one, for example. But that's offset by other, lower-margin items.

Remember, I'm not comparing the retail cost of an item to the wholesale cost, which is what you seem to be focusing on. I'm comparing the retail cost of an item to Canon's cost to produce it, which is a very different thing.

If you really want to get good prices, get a part time job selling cameras. The reps will give you vouchers for great deals. My coworker got a killer price on a Sony HDMI reciever (lower than wholesale...about %40 off) by taking a quiz on Sony products and ordering directly from Sony.

That's how I got my big discount on the Apple computer (mentioned in a previous post) -- I got 100 on a "test" of my Macintosh computer knowledge and got about 45% off. I guarantee you that Apple (and Sony, in your example) didn't sell me that computer at less than it cost them to build it and get it to me.

About a year and a half ago on a thread somewhat like this one, I said I wouldn't buy a full-frame DSLR until the price dropped below $2000. I can't find the post in the archives now... it's like a needle in a haystack and the FM search leaves a lot to be desired. In any case, if I remember correctly I said it would happen within a couple years, and that furthermore the price of a full-frame DSLR would drop below $1000 within 5 years. People said I was crazy, there's no way that could happen, etc.

So less than two years later Dell had a sale and the 5D (which didn't exist at the time of my speculation) was available after rebate for $1785. I seriously doubt it will take another 3 years before a sub-$1000 full-frame DSLR will be available. MSRP may not be below $1000 within three years, but street price will be.

keep em rollin ;-)