Canon patents faster Live view with phase detect AF

[adakshi]

Here is the link
http://www.photographybay.com/2009/02/1 ... view-mode/" onclick="window.open(this.href);return false;

Now with movement of main mirror they have three options.
Normal optical view with phase detect AF
Live view with phase detect AF with mirror facing down
Live view for video with no mirror in the lightpath.

Genius.

Thanks
Ayan

[David Kilpatrick]

It just looks as if they are using the Pellix type pellicle mirror to create an AF lightpath without a moving mirror for Live View.

Seems redundant to have the AF module in the base of the camera. There is no technical reason why the AF sensor (phases detection) can not be incorporated into the viewing prism, and measure phase from small wedge/prism marks on the screen - which would also function visually. There is also no technical reason why the AF sensors can not double as exposure metering sensors. A normally orientated pellicle mirror could then provide optical finder, live view simultaneously as needed with silent operation and complete effective dust sealing of the sensor.

Downside - with any pellicle mirror, there is a small overall light loss. The remaining 97% of light must be split two ways, typically 70/30 to film/sensor and AF/finder. There is therefore a 2/3rds stop cut in the brightness of the viewfinder, which could be why Canon prefers a flipped pellicle mirror (the 2/3rds cut is in the light to the AF cells). At present, this already happens, about 1/3rd of the light only is used for AF cells. There is also a 1/3rd cut in light to the sensor, which means an effective ISO loss. Finally, with a pellicle mirror there is a slight loss of resolution and contrast - the action is a little bit like a weak AA filter. This is not evenly distributed as the distance of the mirror skin or glass is not constant.

These compromises would be present in the proposed Canon design.

David

[David Kilpatrick]

No I wouldn't. I would have dealt with in exactly the same way. Facts are facts, and you can't have a pellicle mirror or beamsplitter without making choices about where to send how much light. At least with AF systems the option exists to send very little light to the AF, most to the sensor. I suggested using pellicle mirrors for digital SLRs years ago, and I mean a decade ago and more. So have many other people.

In the late 1990s I suggested a design to Hasselblad using the EL type body with a digital sensor placed in the base opposite the focusing screen - a camera where the mirror would be 45 degrees one way for viewing, lifted up for film exposure, or turn 45 degree the other way (like the Canon mirror) to make a digital shot. Seems quaint now to think in terms of needing film AND digital. And today, suggesting a flat digital sensor laid in the bottom of the camera (uncovered, this assumed leaf shutter lenses) would be mad because we are all fully aware of the dust issue. Back then, the sensor cleaning and dust thing had not yet arrived.

David

[adakshi]

It just looks as if they are using the Pellix type pellicle mirror to create an AF lightpath without a moving mirror for Live View.

Seems redundant to have the AF module in the base of the camera. There is no technical reason why the AF sensor (phases detection) can not be incorporated into the viewing prism, and measure phase from small wedge/prism marks on the screen - which would also function visually. There is also no technical reason why the AF sensors can not double as exposure metering sensors. A normally orientated pellicle mirror could then provide optical finder, live view simultaneously as needed with silent operation and complete effective dust sealing of the sensor.

Downside - with any pellicle mirror, there is a small overall light loss. The remaining 97% of light must be split two ways, typically 70/30 to film/sensor and AF/finder. There is therefore a 2/3rds stop cut in the brightness of the viewfinder, which could be why Canon prefers a flipped pellicle mirror (the 2/3rds cut is in the light to the AF cells). At present, this already happens, about 1/3rd of the light only is used for AF cells. There is also a 1/3rd cut in light to the sensor, which means an effective ISO loss. Finally, with a pellicle mirror there is a slight loss of resolution and contrast - the action is a little bit like a weak AA filter. This is not evenly distributed as the distance of the mirror skin or glass is not constant.

These compromises would be present in the proposed Canon design.

David

Ok I am not sure I have understood everything correctly, how is it possible to incorporate AF sensor in prism, won't it block light.

Also, in 2nd paragraph, why would I have effective ISO loss if I have mirror up while shooting in phase detect LV. I can understand VF might be dim in general. and scenario might be different in video mode. sensor will get less light in mirror down LV mode, but in that case canon can have AF. In mirror up video mode, they'll have max light but no AF.

Thanks
Ayan

[David Kilpatrick]

Ok I am not sure I have understood everything correctly, how is it possible to incorporate AF sensor in prism, won't it block light.

Also, in 2nd paragraph, why would I have effective ISO loss if I have mirror up while shooting in phase detect LV. I can understand VF might be dim in general. and scenario might be different in video mode. sensor will get less light in mirror down LV mode, but in that case canon can have AF. In mirror up video mode, they'll have max light but no AF.

Thanks
Ayan

AF sensors in the prism would be no different from metering sensors in the prism - or the eyepiece of the prism itself. They would require a rather strange looking focusing screen, with blocks looking like split image focusing aids for every sensor position.

The Canon design is for a semi-silvered mirror - a pellicle system - using phase detect LV. In video mode, or contrast detect LV, the full light level would reach the sensor but in phase detect LV mode probably only 70% of the light would reach the sensor. The benefit of the Canon design is that existing AF modules and the design of the bottom part of the darkchamber would remain the same as with current models, low cost. Only the mirror mechanism would need to be changed, and could presumably provide 'silent' operation using the mirror with the prism finder (still as a pellicle mirror). What I do not see in the diagram is any explanation how conventional viewing through the prism will work. That still calls for a twin-mirror instant return assembly. In total, three mirror elements would need to be included, a normal twin-mirror system folded up to the focus screen when the pellicle mirror was brought into position for phase detect KLV shooting.

David

[stevecim]

With the results some cameras are getting with at ISO1600/3200 would a pellicle mirror be a good option? at a first glance the pellicle mirror could be a good match with Sony's backlit sensor.

I remember seeing Canon/nikon film cameras doing +10 fps using fixed pellicle mirror allways wondered why they where not used on DSLR, I just assumed the mirrors where very expensive