I have been dreaming about a FF 1D mark IV camera ever since it was in the rumors, but it turned out to be a "minor" update to the 1D III, similar to the 50D to 40D upgrade. Nevertheless, now I begin to think if a possible 1Dv camera will finally bring us the longtime awaited revolution. So let's discuss if such a camera can be designed.
See, now Canon has two 1-series bodies in their Lineup. Some people say it's redundant to do so since Canon can very possibly combine these two cameras into one while maintain the utility of two. To build upon that, I wonder if the following is doable from an engineering point of view.
To make a FF camera that has a very high resolution (say 32mp) just like the resolution that the possible 1Ds IV will obtain. Make the camera capable of shooting at full resolution at a low frame rate (say 5fps) to maintain its utility equivalent to an 1Ds camera - this is not something new to do. Now, let's also use the same FF sensor without cropping, reduce the image output to a lower resolution (say 16mp, just like to shoot lower rez images on your camera by adjusting image size) and increase the frame rate to 10FPS to simulate the performance of a "FF" 1D IV - I wonder how hard this can be.
The current processors have proven to be fast enough to handle the data at this resolution and speed above. The idea of the movie mode of a FF DSLR (5D II) is to use the same FF sensor (without cropping) to achieve 24FPS or more at a much lower resolution compared to its native resolution. So can we use what we currently know how to do to make a camera that is both a FF high rez camera at low speed and a FF high speed camera at low rez? If so, i guess the merging of the 1-series bodies will make sense.
Ben
PS: Daniel, help!... =)
Life's too short to spend time trying to make scene of what Canon should, could or would do... go spend some quality time with the family, and save the brain aneurism for the Canon execs that try to make these decisions - if they can make a buck off it, you better believe they'll do it.
Benjamin: To make a FF camera that has a very high resolution (say 32mp) just like the resolution that the possible 1Ds IV will obtain. Make the camera capable of shooting at full resolution at a low frame rate (say 5fps) to maintain its utility equivalent to an 1Ds camera - this is not something new to do. Now, let's also use the same FF sensor without cropping, reduce the image output to a lower resolution (say 16mp, just like to shoot lower rez images on your camera by adjusting image size) and increase the frame rate to 10FPS to simulate the performance of a "FF" 1D IV - I wonder how hard this can be.
AFAIK, every camera on the market handles lower resolution by downsampling the sensor data, but is still reading the whole sensor every time. I think to accomplish what you're suggesting would involve reading every other pixel (i.e. the odds from this row, the evens from that row), and I don't know if that'd be feasible (or feasibly able to double the speed). So...I don't think your idea offers the flexibility you suggest. :(
Benjamin:I have been dreaming about a FF 1D mark IV camera ever since it was in the rumors, but it turned out to be a "minor" update to the 1D III, similar to the 50D to 40D upgrade.
This is off topic, but personally, I consider both the 1D4 and 50D to be major upgrades. The 30D and 1D2n are examples of what I consider to be minor upgrades. Everyone is entitled to their opinion, of course.
Benjamin:Nevertheless, now I begin to think if a possible 1Dv camera will finally bring us the longtime awaited revolution.
I don't think it will be the 1D5, but I do think the camera you're dreaming of will come eventually.
Benjamin:I wonder if the following is doable from an engineering point of view.
There are at least three difficulties:
As for the mirror, Nikon has proven that they can do 11 FPS on a full frame mirror in a $4,000 body (D3). Unfortunately, the rest of the camera isn't fast enough to match that, so you have to switch to 1.5X mode to get 11 FPS (or go down to 9 FPS and use the full frame).
The next problem is reading the sensor fast enough. On the 5D2, the mirror assembly is cheap, so it can't hit more than 4 FPS. The sensor is probably faster than that; maybe 6 or 8 FPS; but nowhere near fast enough for 30 FPS. Even if it could read out the entire sensor at 30 FPS, the electronics (DIGIC) wouldn't be fast enough to handle it.
If your sensor can't hit 11 FPS, there are several workarounds:
Skipping is the method Canon uses to create 30 FPS video on the 5D2. Instead of reading out the whole senosr, it skips 2 out of 3 rows. This results in terrible aliasing artifacts.
Cropping is the method Nikon uses to get the 9 FPS up to 11 FPS on the D3s: it crops the angle of view to 1.5X. That allows the sensor to go much faster. This is also the method used to create 5X and 10X live view.
On-chip binning is the best compromise. It connects multiple pixels together (e.g. four) and reads them out together. It also causes aliasing artifacts, but not as bad as the ones you get from skipping.
After you do something to get the sensor fast enough, then you have to process all that data. The 1D series cameras use multiple DIGIC processors to process the raw data, create jpegs for camera previews, etc. Moore's Law helps here. Downsampling a high MP image to a low MP image is actually very difficult. Doing it in a way that doesn't cause artifacts is even more difficult, as Canon's highly flawed sRAW implementation shows.
Benjamin:So can we use what we currently know how to do to make a camera that is both a FF high rez camera at low speed and a FF high speed camera at low rez?
Here's what I think a good "1D and 1Ds" combination camera would be:
I don't think that's what the 1D IV will be. I'm guessing more along the lines of 48 MP, 5 FPS, and video.
Ahh, I thought it was easy... Maybe just because I don't know enough to know it's hard.
But anyway, fantastic explanation! Really appreciated!
PS: I would call 30D and 1D II N "minor, minor upgrades"... hahah =)