The Radical 1Ds Mark IV [CR1]

[NormanBates]

I don't think so.
It's better to increase pixel density and scale down afterwards. Remember, Bayer pattern sensor do deliver more details than a Foveon sensor with 1/4th pixel count.

Can you substantiate this????

I don't know about the Foveon sensor, but I know that increasing pixel density and scaling down afterwards is not a bad thing:
http://www.similaar.com/foto/mpix/mpix.html

 

[NormanBates]

maybe "modular" means they are implementing something that I have been proposing for some time: build a standard stills camera, then add functionality through different grips, as in:

* for fast shooters: a grip with an additional digic processor, doubling the fps count
* for heavy shooters: a grip with an ultrabig battery and either 4 card slots or a connector for an external SSD drive
* for videographers: a huge and very expensive L-shaped grip that also covers the back of the camera, and which includes:
- a fast video processor, that can record a series of reduced resolution RAW images through binning (not line-skipping) to an external SSD drive
- full-res HDMI output
- a big (5"), high-res (720p), swivel screen
- a couple of XLR inputs
- a headphones connector

 

[x-vision]

This new rumor makes a lot of sense.

The 1D/1Ds pairing is not sustainable anymore.

This pairing was working fine for low-res/hi-res cameras but will not work for hi-res/even-higher-res cameras.

To compete with the upcoming Nikon D4, the 1DV will have to have a FF sensor (likely 20-24mp) and sell for no more $5500.

With that, Canon will have a very hard time charging much more for the 1DsV - if it's basically the same camera as the 1DV but with more megapixels (50mp?) and slower fps.

To charge more for the 1DsV, they need to better tailor it for its intended market, which is the hi-res studio market served currently by MF.

Canon will most likely keep the EF mount and the FF senor (24x36mm) but the body format will change to better fit its intended use.

Maybe Canon will make it an interchangeable-sensor camera, so that it is more future proof.
Also, they may decide to make multiple variations of the 50mp sensor (say a B&W variation, a variation with 6-color filter, etc.), so a modular design makes a lot of sense.

 

[Artisttt]

maybe "modular" means they are implementing something that I have been proposing for some time: build a standard stills camera, then add functionality through different grips, as in:

* for fast shooters: a grip with an additional digic processor, doubling the fps count
* for heavy shooters: a grip with an ultrabig battery and either 4 card slots or a connector for an external SSD drive
* for videographers: a huge and very expensive L-shaped grip that also covers the back of the camera, and which includes:
- a fast video processor, that can record a series of reduced resolution RAW images through binning (not line-skipping) to an external SSD drive
- full-res HDMI output
- a big (5"), high-res (720p), swivel screen
- a couple of XLR inputs
- a headphones connector

This would be very cool! Instead of 1D;1Ds;5D;3D, just one good FF still camera with multiple grips.

 

[JW]

I think one thing we'll see is in-camera remote flash control and a new flagship flash (590 or 6XX EX.. maybe 6XX RX to indicate the receiving capability of the flash). I'm thinking it will be controlled by IR, RF, or maybe even something like bluetooth technology (it won't be w/ pre-flashes since the 1D obviously doesn't have a flash). Just a wild guess, but it's been rumored recently that there's a new flash coming. I'm thinking it will be more than the normal flash evolution.

 

[peejay]

All I can say is it better be worth the damn wait.

Get on with it Canon. MF Backs are looking awfully tempting these days...

 

[afrank99]

I don't think so.
It's better to increase pixel density and scale down afterwards. Remember, Bayer pattern sensor do deliver more details than a Foveon sensor with 1/4th pixel count.

Can you substantiate this????

It's obvious, isn't it?
Any Bayer pattern sensor definitely has FULL luminance resolution while chroma resolution is 1/4 of its nominal resolution.

Bayer 24MP: 1 luminance, 1/4 chroma
Foveon 8MP: 1/4 luminance, 1/4 chroma

And we all now, luminance resolution is much more important to the eye than chroma resolution, this is why JPEGs often use chroma channel downsampling without loosing too much quality.

In reality this means that Bayer sensor deliver resolutions comparable to a Foveon with ~2/3 of this resolution (16MP in this example).

 

[hsmeets]

So what’s so radical?

Don’t disregard the large sensor Canon showed off. Obviously it wouldn’t be that big, but there is something radical about the 1Ds4 sensor.

Just some random thoughts ranging from maybe to unlikely....

- Back illuminated type of sensor.
- No AA filter anymore? software selectable levels of AA filtering in post processing.
- Colored Microlenses that also act as bayerfilter
- Binning in other configurations as 2x2 pixels (different pixel shape?)
- Foveon like sensor (3 layers) or Fuji style sensor (big and small pixel pairs)
- No optical bayer filter, spectral sensitivity tweaked per pixel with doting of rare elements right onto the die.
- Binning of pixels for Video, global electronic shutter.
- AF integrated into Sensor to allow Phase detect during live view without translucent mirror, Foveon style implemented as a second layer under the pixels (or between pixel in the spacing room).
- Slide in sensormodule into a standard body (slide in from bottom or so).
- Bayer filter R-G-G-B changed to R-G-B and some other color like yellow/orange for better reproduction of skin-tones?

 

[Edwin Herdman]

Okay, time to have a little fun and predict the future. Check back in 2011 or 2012 and see how close I come.

Sensor somewhere in the 50-100mp range. (Not that difficult since Canon has already unveiled a 120mp APS-H sensor.)

Coming into this post I didn't realize it was in the 1.3x crop size. But it's still a one-off for now, lab grown means they can afford a lot more failed parts than they would in a realistic setting: They can just push up the pixel count, which normally would lead to an ever-increasing rate of defective parts (I dunno if it's geometric or what, but certainly a significant increase, to the point of being uneconomical) that fail because of the tight production tolerances required. Production tolerances that would be perfectly reasonable for current generation parts would, my theory goes, end up with misses on current tech. I'm sure they need an improved process over current commercialized technology just to make it, in fact. It's promising though. Of course, who knows there are any compromises they made for the sake of the pixel count which would hobble it in a production camera. I doubt it, but you never know. From my time watching the PC industry, you hear about fantastic new technologies that don't end up coming for years. Fiber optic connections between components on a mainboard have been talked about for years and they're still many years out from being developed, yet there are working prototypes all the same.

I do expect this to be coming out sooner than some of the pure research type stuff we hear about constantly on the PC side of things, like "real-time ray-traced Wolfenstein on a laptop!" (turned out the laptop was just streaming video from like four massively parallel servers) and the like. Things that won't be coming for fifteen years, if ever. The fact that this is most likely a complete system that just needs the camera's processor and other systems to be beefed up in parallel for a working camera means it'll be out sooner than these one-off designs we see so often.

Considering that the fabrication plants for CMOS technology are owned by third parties, and the state of the art in their current process size is consistent regardless of who is using these third party facilities, I don't expect this is a sizable advantage over other parties. Sony has been doing pretty well recently on the sensor front I think, and their relationship with Nikon probably has people in both companies worried.

Unless this materializes into a product before the competition, it's just a note from Canon saying "hey, we put money into R&D too! Isn't that cool?" It's probably a few years out, if not more. I think how soon the process can be commercialized, and other systems in the camera made fast enough to process 120MP images with any speed, is going to be slower than we would think.

It IS interesting to note that the medium format announcements lately haven't been much over 40MP. I think Hasselblad just put out a thirtysomething megapixel back. Seems to show to me that the MF manufacturers are stuck with the technological crumbs of the CMOS feast (poor metaphor).