Canon continues to work on sensor-based image stabilization, which is a feature we expect to see in the next major EOS R camera body release. Canon has been working on getting the in-body stabilization to work with its in-lens stabilization technology.
This USPTO patent takes things one step further, as Canon appears to be working to get three stabilization systems working together.
These three systems would be:
- In-body or image sensor stabilization
- In-lens image stabilization
- Electronic image stabilization
Canon News breaks down what this patent likely means:
So what's the goal of all this? The gist is they want to improve image stabilization when the various parts of the camera system are working together to stabilize video recording. What Canon is stating (I believe) is that as the shutter speed increases, the amount of shake that the optical / image sensor handles increases, and the frequency of shake that the electronic image stabilization corrects decreases. This leads to better image stabilization when using slow shutter speeds to shoot video.
There are two types of shake. Low frequency is when you are hand holding and there's a slight movement in a general direction, more like drifting. High frequency is when things are jittery and there are lots of small, but frequent movements. Like when I drink too much coffee.
Optical and Image sensor stabilization works great with high-frequency shake, electronic stabilization works best with low-frequency shake. The camera monitors the shake and determines which system should correct the shake being detected.
According to Canon, what can happen in low light situations and slow shutter speeds is something they call accumulated shake. Accumulated shake is what happens when the image sensor is actively capturing an image, motion blur during this time cannot be corrected by electronic image stabilization because the motion blur has already happened during the image sensor accumulation because the shutter speed is slow. Thus the term "accumulated shake".
Accumulated shake is most likely low frequency because that shake is handled by electronic image stabilization. Thus the problem. For slow shutter speeds, you can't rely on Electronic image stabilization to correct for low-frequency shake as much as you can with a fast shutter speed.
What Canon is stating (I believe) in this patent application is that they are changing it so that optical and sensor image stabilization is used for more low-frequency shake as the exposure time increases, and that how the vary image stabilization pieces interreact is determined by the shutter speed.
Whew.
If there is a patent it indicates this has been in the lab and in the works for some time now.
Is that only from the Electronic stabilization? I really don't like the warp look that electronic stabilization gives. I think IB or Lens stabilization is better.
It's mostly from the electronic stabilization although it's also there to some degree even with the IBIS alone but the magnitude is much less and not really an issue like with electronic stabilization. I suspect that to some degree it's simply a price you pay for roll correction.
I expect the rumored RF 1DX would have to have this, I wonder if the more "5D" camera will have it as well... Either way I'm betting we will see it by next year.
Possibly, but I'm not counting on it. If the rumored R 1Dx is truly a mirrorless variation on the 1 series theme it will be a stills oriented bombproof body. Canon doesn't generally put untested technology in the 1 series. In this case, I think it will make it into the 1 series mirrorless only if Canon is absolutely certain the body can be beaten and battered without damage to the stabilization system.
Suppose the rumour about the simultaneous launch of the RX and 1DX3 is true, do you think they'd throw it into the RX to make it look better on the spec sheet than its mirrored cousin?
This is a great explanation and now I have a proper term to describe this type of low frequency shake. "Accumulated shake" is way better than what I colloquially call the "spaghetti light effect."
If you watch the split video, the phenomenon on the right is what CR is describing. It's terrible and can't be fixed in post. Most manufacturers that only have EIS solutions in the action camera/360 camera space have patched this in firmware by pushing the shutter speed and ISO in low light to somewhat mitigate the effect, but many of those cameras are intended to be used in auto/semi-auto. If Canon can utilize all 3 systems to achieve gimbal like stabilization in good light and rely on a combination of IBIS/IS+EIS in low light, it would be novel indeed.
OIS, handles multi- pixel image shifts best in still photography, works well with video for big corrections.
IBIS, handles “a few pixels” and sub-pixel shifts the best in still photography, works well in video for small corrections.
Pixel shifting, terrible for stills, fantastic for very large corrections in video.
Note that Canon is already using a combo of OIS and pixel shifting in the last few Cameras that they released.
Note that Oly and Panasonic are using all three in their latest high end cameras.
It is inevitable that Canon will follow suit. The debate is not about which system is better because the combo of the three outperforms each of the choices. The debate should be about why they took so long to get there......
Like virtually all market leaders, Canon is a conservative company. They don't like mistakes and don't risk their reputation on new technologies until they are confident it will work and that there is market demand. Conservative approach may not be what pleases forum participants, but who can argue with their success?