Canon 7D on it's way out?

[briansquibb]

Re: Canon will end the 7D series life?

Think about it this way:
When the 7D came out, canon engineers were well aware that the 1D4 (which came out a month or two later, i think) would be the last of its kind.

So is that why they continued developing the APS-H and even demo'd a 120mps APS-H. I guess that is another myth that is not supported by facts : :

 

[foobar]

Re: Canon will end the 7D series life?

I was not talking about technology advances but commercial technology advances. It is easily possible to make a 1.5 litre engine develop 1000bhp. But it is a lot cheaper to develop amd build an 8litre engine to develop 1200bhp.

Wrong methaphor for the subject.

Personally I thisnk the APS-C sensor is coming quickly to the point of commercial technology limitations - ie it becomes cheaper to move to a larger, lower tech sensor to get a bigger improvement.

For sensors the easy (read cheap) way to get more from a sensor is to increase the size - and I believe we are getting close to that point

That's simply not how chip production works. In chip production (and that applies to all kinds of chips, not just image sensors), a bigger area means exponentially higher costs. It's not that important what is actually on the chip.

The price for a chip of a certain size only significantly drops every 5 years or so when the industry is changing to larger wafer sizes. The reason why other chips get cheaper and cheaper is because they get smaller (or you can fit more complex designs into the same space).

Image sensors do not benefit from these effects since their size is fixed.


But even if overall sensor manufacturing costs would be low enough for a rebel-priced FF camera, all other components are more expensive as well: The viewfinder prism, the microlens array, the AA filter and also not to forget that wide-to-standard lenses need quite a bit more glass compared to their APS-C counterparts, thus FF camera systems will always be more expensive than ones with smaller sensors.

Computer technology is ahead of sensor development (in its timeline) - but there hardly been any significant progress in PC 'power' for the last 3 years. The push has been in multi core - ie 2 cores go faster than 1, rather the speed the single core. Even the entry PC's now have a minimum of 2 cores, midrange have 4.

That's because today's quad-core chips fit in the same die size as single-core chips did a few years ago.

So is that why they continued developing the APS-H and even demo'd a 120mps APS-H. I guess that is another myth that is not supported by facts : :

In my opinion:
- The 120mp APS-H sensor was just a technical exercise (so far there has been no indication that Canon is planning to put it into a product)
- The reason Canon went with APS-H on the 1D for so long was partly because it was the largest size that could imprinted in one shot, as stated in Canon's full-frame white paper
- APS-H was never a mainstream solution for Canon. It was always limited to a single camera model and (unlike APS-C) they never produced a single lens targeted at that crop-factor.

 

[briansquibb]

Re: Canon will end the 7D series life?

In my opinion:
- The 120mp APS-H sensor was just a technical exercise (so far there has been no indication that Canon is planning to put it into a product)
- The reason Canon went with APS-H on the 1D for so long was partly because it was the largest size that could imprinted in one shot, as stated in Canon's full-frame white paper
- APS-H was never a mainstream solution for Canon. It was always limited to a single camera model and (unlike APS-C) they never produced a single lens targeted at that crop-factor.

I dont think Canon would spend so much effort (and money) on a sensor that was planned to be dropped, although there may have been a subsequent decision.

APS-H was their mainstream pro solution - rather than the consumer APS-C.

I suspect the EF lens continued was because of the migration from 35mm film that ran in parallel. The APS-H sensor got the sweet point of the image as well as a useful 1.3 crop

It is a sensor that could easily replace the APS-C to improve performance quickly and easily.

Brian

 

[Meh]

I dont think Canon would spend so much effort (and money) on a sensor that was planned to be dropped, although there may have been a subsequent decision.

They might have but we can only speculate what the intention was. As was pointed out, APS-H made a lot of sense from a manufacturing perspective and therefore was the best compromise on yield and cost. It made economic sense to do it at the time and very well may have justified the R&D costs.

It is a sensor that could easily replace the APS-C to improve performance quickly and easily.

Not really but depends what you mean by "replace" I suppose. Can't use EF-S lenses on an APS-H body and AF sensors in an APS-C body are also much smaller so really can't say APS-H can replace APS-C. Two different classes of bodies I think.

 

[jrista]

I wouldn't write off APS-H nor APS-C yet. Both have value, from a cost perspective (cheaper to make, higher profit margin, easier to sell) as well as a crop/reach and framerate perspective. Full Frame is great, but it has its limitations. Three obvious limitations for FF are its much thinner DOF, limited readout rate (i.e. limited max FPS), and its limited reach. When it comes to wildlife, bird, sports, airplane, and pretty much any other form of action photography where you need greater reach, adequate DOF at wide apertures (which can often be thinner than you might prefer), and high frame rate...APS-H and APS-C are going to have value.

If Canon can achieve 14fps with their new FF sensor, what could they achieve with APS-H or APS-C? Previously, their APS-H sensors could achieve almost double what their FF sensors could...so it doesn't seem outside the realm of possibility that they could offer a 24-26fps 16mp APS-H professional sports body, or possibly a 28-30fps 14mp APS-C one. Apply the new 1D X sensor tech on a smaller scale, pack in a couple Digic5+ processors, and I think we could see amazing things.

There is more to it than just purely cost...nothing is ever really that simple. I see APS-C having a longer life than APS-H (if it isn't already relegated to the Canon museum), however I do see continued life for cropped sensors.

 

[5dMkii]

If Canon can achieve 14fps with their new FF sensor, what could they achieve with APS-H or APS-C? Previously, their APS-H sensors could achieve almost double what their FF sensors could...so it doesn't seem outside the realm of possibility that they could offer a 24-26fps 16mp APS-H professional sports body, or possibly a 28-30fps 14mp APS-C one. Apply the new 1D X sensor tech on a smaller scale, pack in a couple Digic5+ processors, and I think we could see amazing things.

Sorry, but you are wrong... it's not about the size... it's about the pixels and data throughput... See, 1Dx has only 18MP for reason. One - sensitivity and two the data amount it produces. D800 can do only 4fps for reason. Not that Nikon wouldn't want but currently there is no processing power and fast memory media which could process this sheer amount of data.

Count with me. The best computer SSD's can achieve what - 350MB/s write speed? And they have some clever controllers. Now - what is the size of D800's RAW? 80MB? x10? 800MB/s for 10fps? Only dreaming currently. Unless they would deploy some serious parallel processing and storage this is out of reach with current tech.

 

[jrista]

If Canon can achieve 14fps with their new FF sensor, what could they achieve with APS-H or APS-C? Previously, their APS-H sensors could achieve almost double what their FF sensors could...so it doesn't seem outside the realm of possibility that they could offer a 24-26fps 16mp APS-H professional sports body, or possibly a 28-30fps 14mp APS-C one. Apply the new 1D X sensor tech on a smaller scale, pack in a couple Digic5+ processors, and I think we could see amazing things.

Sorry, but you are wrong... it's not about the size... it's about the pixels and data throughput... See, 1Dx has only 18MP for reason. One - sensitivity and two the data amount it produces. D800 can do only 4fps for reason. Not that Nikon wouldn't want but currently there is no processing power and fast memory media which could process this sheer amount of data.

Count with me. The best computer SSD's can achieve what - 350MB/s write speed? And they have some clever controllers. Now - what is the size of D800's RAW? 80MB? x10? 800MB/s for 10fps? Only dreaming currently. Unless they would deploy some serious parallel processing and storage this is out of reach with current tech.

First off, make sure you read fully. I know its related to pixel count (which is the reason I stated 16mp APS-H or 14mp APS-C)...but its more than that...its the design of the sensor, the design of the readouts (how much parallelism, how the wiring literally connects to the pixels, etc.) the speed of the image processors, the amount and speed of buffer memory, etc. Not to mention the fact that the 1D X has more MP than the fasted of its predecessors, which did 10fps @ 16mp (1D IV) and 10mp (1D III), so obviously having a "lower" pixel count is not a critical factor. The point I was trying to make was by using the same TECHNOLOGY that made the 14fps 1D X 18mp sensor possible, it should be possible to make a higher FPS sensor with denser pixels in a smaller package where physical distances are smaller (something Canon noted was important with their 120mp APS-H prototype which achieved a whopping 9.5fps in testing!) You would gain cropping power/extended reach over the 18mp FF sensor, with high FPS...a win/win for action photographers.

Second, the fastest computer SSD's are over 1GB per second when you factor in PCI-E slot variants:

1.9GB/s r/1.7GB/s w: http://www.newegg.com/Product/Product.aspx?Item=N82E16820227764
1.0Gb/s r/ 900MB/s w: http://www.newegg.com/Product/Product.aspx?Item=N82E16820227760
975MB/s r/875MB/s w: http://www.newegg.com/Product/Product.aspx?Item=N82E16820227745

Flash memory can be EXTREMELY fast, if you have the right interface for it. On the memory card front there are the new 125mb/s QXD memory cards, the QXD interface (or perhaps it was the SDXC interface) is currently rated up to 300mb/s, and who knows how far they'll go in the future...600mb/s, 800mb/s? You would only need 800mb/s assuming you actually save an UNCOMPRESSED 36.3mp D800 NEF, compressed they are around half the 74mb of an uncompressed one. For a 16mp APS-H sensor at 24fps with properly compressed images, you would need maybe 300mb/s write speed, which is technologically possible with QXD/SDXC today.

 

[Viggo]

Why would you need more than 12 fps? Then you need highspeed video instead.

And the other problem which is much bigger is, how the heck to you AF between 20 fps? TO make a motor that flips the mirror for 20 fps isn't a problem.

This also a cost issue, who would pay 9000 usd for a aps-c with 20 fps?

 

[briansquibb]

Why would you need more than 12 fps? Then you need highspeed video instead.

And the other problem which is much bigger is, how the heck to you AF between 20 fps? TO make a motor that flips the mirror for 20 fps isn't a problem.

This also a cost issue, who would pay 9000 usd for a aps-c with 20 fps?

+1

Interesting to see the frame rate DROP from the Nikon D700 at 8fps to the D800 at 4fps. Rather changes the character - from a sports camera to a studio camera.

I wonder if this will become a trend?? I feel the between 6 and 8 fps is adequate for me

 

[funkboy]

In my opinion, if Canon produces 3 new bodies in 2012 (1DX, 5DX and 5D mark III), we won't see any 7D mark II this year

Don't forget that we're supposed to be seeing a Ciné DSLR this year too; the new product pipeline is certainly pretty loaded...

 

[funkboy]

Or, if you prefer, yes, it means that a 7DII is coming soon, just as surely as the upcoming storm we're expecting in the Northeast US was caused by a butterfly flapping it's wings in China.

That butterfly had better not tweet about its accomplishments if it doesn't want to get itsself on a terror watchlist...

 

[jrista]

Why would you need more than 12 fps? Then you need highspeed video instead.

And the other problem which is much bigger is, how the heck to you AF between 20 fps? TO make a motor that flips the mirror for 20 fps isn't a problem.

This also a cost issue, who would pay 9000 usd for a aps-c with 20 fps?

Why would it cost $9000? One of the most significant cost factors is sensor size. It costs 10-20x as much to make a FF sensor than an APS-C sensor, so I'd figure a high speed APS-C action camera would cost only as much as a FF, and then only because of marketing, not necessity.

As for mechanics, flipping a mirror that fast IS a problem, which is why Canon can only do 14fps in mirror-lockup mode. I assume the same thing for an APS-C sensor. You could do clever things with AF as well...you probably wouldn't need to AF between every single frame since its almost real-time...ever other would be just as effective as a 1D X at 12fps.

 

[jwong]

Why would you need more than 12 fps? Then you need highspeed video instead.

And the other problem which is much bigger is, how the heck to you AF between 20 fps? TO make a motor that flips the mirror for 20 fps isn't a problem.

This also a cost issue, who would pay 9000 usd for a aps-c with 20 fps?

Why would it cost $9000? One of the most significant cost factors is sensor size. It costs 10-20x as much to make a FF sensor than an APS-C sensor, so I'd figure a high speed APS-C action camera would cost only as much as a FF, and then only because of marketing, not necessity.

As for mechanics, flipping a mirror that fast IS a problem, which is why Canon can only do 14fps in mirror-lockup mode. I assume the same thing for an APS-C sensor. You could do clever things with AF as well...you probably wouldn't need to AF between every single frame since its almost real-time...ever other would be just as effective as a 1D X at 12fps.

I agree that future performance will be limited more by physical constraints rather than signal processing. The cost between a FF and a APS-C sensor might be a few hundred bucks, but the curtain/shutter mechanics would have to be redesigned to support faster frame rates (assuming same linear speed over a smaller sensor size). It might be easiest to start designing it using the 1DX as a starting point. If so, the price wouldn't be that different from the 1DX.