Sensor hygiene....

[sandollars]

POTAYTO, POT-OTTO. sheesh......

 

[Kolia]

Aha, so it is indeed a "Sony thing." That's kinda cool. What disadvantages might there be?

Yes, that is a feature that was added to ADD a disadvantage... lol

Anything with an EVF or running in Live View needs to expose the sensor in order to show anything on its display. So when you do hit the shutter button, the front curtain (it's what your Front Curtain Sync function is talking about) needs to go up and cover the sensor before it can go down to start the exposure. The rear curtain follows whenever it needs to to get your exposure as usual.

So not wasting time lifting that curtain makes for a faster response time over all. Just as if you didn't need to move your mirror out of the way before exposing the sensor.

 

[Gavjenks]

Yes, that is a feature that was added to ADD a disadvantage...

Uh... what?

"Not wasting time lifting a curtain" is an advantage. I am asking whether the digital front curtain system has any inherent disadvantages. As in, negative side effects that result from the decision to use a "digital front curtain." They don't list any on the website, but since it is their own company, of course they wouldn't. So I'm asking here.

One possible disadvantage off the top of my head is that this method would involve more data processing, since it needs to use a fancier algorithm to store and sort and throw away data repeatedly as the rear curtain advances, whereas the traditional method only has to clear the data once. This may or may not lead to issues with requiring a faster CPU or larger buffer (because you need to be storing stuff waiting to be written to the card as usual PLUS all this extra intermediate data getting sorted through as the shutter flies) or something that might make the camera cost more money.

 

[Kolia]

Dumping the sensor data one line at a time or all at once shouldn't cost anything more in term of CPU work.

Any sensor will dump everything prior to taking the picture.

From the user's perspective, the end result is a faster reaction time when you hit the shutter button.

 

[Gavjenks]

Dumping the sensor data one line at a time or all at once shouldn't cost anything more in term of CPU work.

I misunderstood how it worked. Yes, it makes sense that this would require no additional processing.

However, now I am curious about why they can't do electronic rear curtain as well. They have a section that attempts to explain why they can't on their website, but I don't see why that's a good explanation. "If we were to clear the pixels row by row going up like a rear curtain, it would just delete the image we just took" Well yes, duh, but the obvious thing to do would not be to CLEAR the rows, but to SAVE the rows going up one by one, in place of the rear curtain.

So what I'm proposing is:

1) Start clearing rows of data one by one. Move at a speed equal to the rate that your system is able to save a row to a buffer computationally.
2) Wait for a delay equal to the exposure time.
3) Begin saving rows at full saving speed (which is the same speed you cleared them at, see #1). No need to clear them, because once you've saved it, you don't care if it keeps collecting more light.

Ta da! correctly exposed image, with no mechanical shutters.




The only reason not to do that would be if the speed of saving data from one row is slower than the speed at which a mechanical shutter flies the distance of one row. Is that the case?

 

[Gavjenks]

Actually, come to think of it... if the computer were capable of transferring one sensor-full of data to a buffer at or faster than the flash sync speed of the camera, then there would be no point in even having a shutter installed on the camera at all, because dual electronic curtains could go faster than mechanical curtains could.

And since the shutter is usually the first thing that breaks in a camera, its lifespan could be much longer...

I assume that things like cellphones already do something like that, so are we only held back by data dump speeds? Or something else?