Canon Mark III Live View and seeing the image while the picture is being taken

Once I have pressed the shutter button is it possible to continue to see the image during long exposures? Viewing the image at this time would allow me to do some dodging when I do not have a Grad ND filter with me.

Nope. There's no "oven light" while the cake is baking.

You can't electronically read out the image from the sensor at the same time as it's being created, so that's not doable. At least not today.

apersson850 wrote in post #18241758
You can't electronically read out the image from the sensor at the same time as it's being created, so that's not doable. At least not today.

Not ever really, well not unless there was a paradigm shift in electronics technology that was greater in significance and magnitude than the introduction of the semiconductor junction diode. Even with the introduction of semiconductors the the underlying principals of active electronic circuit design remained pretty much the same, although with the original bipolar transistors they do work in a different manner to the older thermionic valves. In a valve a voltage is used to control the output, while in a transistor it is the flow of a current. The introduction of the Field Effect Transistor was a move back to the principals of valve technology, in a solid state package, apart from the operating voltages you can literally transpose an FET for a Triode Valve in just about any circuit configuration I can think of. The big advances that semi-conductors brought was initially miniaturisation, and then integration of components into a single device. A development that has proceeded at a geometric rate ever since, although it does now seem to be tailing off slightly, at least as far as computing and Moore's Law is concerned.

Given the constraints of current technology, which requires separate data collection and read stages you could potentially use a semi-silvered mirror to split the light path, and use a second imaging sensor, and I suppose that by some clever optics it could be possible to reduce the size of the sensor, you couldn't just use a smaller sensor, as that would affect the Fov seen on the viewing screen. The other problem will be to actually achive a useable signal from the secondary sensor, since the light going to it will be removed from the main sensor, so effectively reducing the sensitivity of the main sensor. It seemingly can be done, as it is available in Sony's range of DSLT cameras, that use a pelli mirror and feed it to the sensor of an EVF, rather than the OVF that Canon used on one of their film cameras. Of course you are not actually looking at the image as it forms in a long exposure. There would be one other possible way of keeping the screen on during a long exposure, and that would be to shoot a whole bunch of very short exposures, and to then integrate the results in post. The problem here is that you would have to extend the exposure by the combined length of all of the read cycles, depending on the required read time, and the desired VF frame rate this could be a very significant percentage of the exposure time. It is this inter-frame work required for reading and processing the sensor data that effectively limits the possible frame rates when shooting video, which should give some indication of what would be possible. This approach would have one big advantage in that it could seriously improve the DR of the final signal. Although each individual exposure is limited by the sensor saturation point, and the 14 bit ADC, you would not be so limited by the integrated result, just use a larger data container for the result, and keep adding the results together, which is all mathematical integration is, adding lot of very small slices of something together. As long as no single exposure becomes over exposed you would end up with more DR the longer the total exposure time was.

The problem with all of this is that it would be difficult and expensive to engineer, the resulting EVF image is still likely to be very very poor, to unusable in very low light, even with a very, by current standards, clean high gain sensor. given that to make any long exposure work you have to solidly mount the camera so it won't move the camera manufactures don't have any incentive to try to develop this type of function, because there is virtually no customer demand for it. I do note however that Leica do use the fact that the Messsucher, combined viewfinder/rangefinder for the non German speakers, system in their M series cameras do not black out during or after the exposure being made as a plus point in advertising and promotional materials.

Alan

I'll start thinking about a dual-port imaging sensor, which will allow readout simultaneously with image acquisition.
You'll hear about it in my patent application, and as soon as I get my patent, you'll probably start hearing more about me...

During long exposures no, but with Magic lantern, there is a hidden option to shoot full res RAW with electornic shutter without any shutter sound, shutter wear or livelivew interruption, basically turning it into mirorrless.

apersson850 wrote in post #18241912
I'll start thinking about a dual-port imaging sensor, which will allow readout simultaneously with image acquisition.
You'll hear about it in my patent application, and as soon as I get my patent, you'll probably start hearing more about me...

If you could just manage to read the charge off the sensel, without affecting the charge at all I think you would be a very popular person. Then you could run real Multi ISO systems for the whole sensor, wouldn't it be nice if you didn't need to shoot too fast to read an exposure out at all the full analogue gain settings one after the other. That really would give you a serious DR boost, and with no loss of vertical resolution as you get with ML's DualISO function. Whats more you could use almost any shutter speed and aperture combination that you wanted for any shot, well at least within the limits of the highest/lowest ISO values.

I suppose if you can solve the non destructive read it would then become feasible that you might then do a read out while you are still doing a store, but even if you could how useful would that be? You are generally doing a long exposure because the is very little light arriving at the sensor in a time unit, so you need to use a lot of time units to integrate the number of photons arriving at the sensor. So even if you can make a reading of the sensel charge non destructively, while the sensel is still recording data, initially the results are going to be pretty difficult to make use of, as by definition very few photons will have arrived to be converted to charge. It would literally be like watching a traditional silver print (or Polaroid) developing in the dish, with the disadvantage that the signal should grow in a linear manner, which may make adding a tone curve to the building image a little more awkward, as you may well want to have the curve be adaptive over time.

An interesting thought experiment, but I'm not really into chip design and fab, or the specifics of even current sensor design. It's 22 years since I graduated with my BEng in Electronic Engineering, and the university I attended was not big on semiconductors; UEA Class of 95.

Alan

It wasn't a fully serious proposal. But it would have been cool to be able to take HDR photos in real time, within one exposure, for example.
My education in IC design is also rather antique by now.