*Update* To update the information below, we're told that the “Canon EOS R Pro” would receive a development announcement in January 2019, ahead of an official launch at a later date.
We're told that a “full-blown” Canon EOS R camera will be coming sometime in February 2019 ahead of CP+, which begins on February 28.
There are obviously no specifications for the camera this far in advance of an announcement, but we've been told on a couple of occasions that a new EOS R camera would be coming in the first quarter of 2019.
More to come…
Flagship so that will be a mirrorless 1D style camera, so unlikely to be high mega pixel.
Based on the observed certification, there is a FF 26 MP mirrorless camera coming. If this holds and it is a "flagship" I would imagine this is more of a high fps/action/sports body. If so, that would imply Canon thinks they have the goods to make AF match that use, which is good news.
Then, considering the rest of the market (Sony/Nikon) seems to be going high MP/lower MP with their bodies. So, a "flagship" could be a high MP body.
Canon's processors are not known to be the most powerful or speedy chips out there, this is why (well partly) that there are so many limitations with video shooting.
Canon needs multicore CPU's or simply double up like on the 1DX II. I think we will definitely be seeing more than one CPU in any mirrorless that has real video specs and/or high speed stills capture.
Canon salesman, more likely. Sharper lenses don't magically create more megapixels on a sensor that's smaller than the 5DS/5DSR. Sure, if you're using a sub-standard lens on the 5DSR and comparing it with the EOS R you're going to get poor results. But a good lens on the 5DSR will outperform any option on the EOS R right now in terms of resolution.
Just shrinking the die means they can put twice as many transistors in the same amount of space, or they can choose to just keep the number the same and have less power consumption and greater speed. A combination of the two could lead to some very speedy digic processors.
However the size of the actual sensor is hard to shrink at this point, because of the physics of light. That's why stacking the processors can be so effective, as the signal can travel backwards to a corresponding set of transistors to do initial processing rather than through a trace all the way to the edge of the board before being handed off to those transistors. (this is different than Foveon's use of the term 'stacked', where they actually stack the sensors themselves to capture more light).
I'd lean towards a 1dx style camera in 1st quarter with dual digic or a new die-shrunk digic 9. I hope we will see one of their new stacked sensors in production soon though because that opens the door to 4k/8k 120fps. Right now they can't do that because of the heat and bandwidth. Smaller processes for the stacked portion of the die and external digics mean more bandwidth + less power which equals less heat.
I'm a CPU/GPU guy and its refreshing to see someone else that understands the subject. :D
Actually yes... unless you are saying that DXO is wrong. A lot of standard lenses cannot deliver enough sharpness to fully utilize sensors more than say, 18 mp. First gen L glass was probably designed to resolve about 18 mp on a FF sensor. I don't know what the current line of L lens resolve to but I am sure a lot of the updates are due to the demands of higher pixel count.
Thanks for your insight. It is hard to believe Canon is still around 35-40nm. The new iphone is 7nm! I think the 1dx has two digic processors and also an AF processor/sensor, so it wouldn't be hard to believe a pro level R camera having multiple processors to accomplish these advanced tasks.