David J Heinrich
Forum Enthusiast
1. Slightly larger sensor size?
Elsewhere, Rriley said that the patent specs for 4/3rds said that the image field as a max diameter of 25mm, but we currently use 21.6mm; thus, the the sensor area could be increased from 225 sqmm to 300sqmm. However, it isn't clear that the lenses are acceptably sharp in those outer areas. There is no point in increasing the sensor size if those last bits of the image circle aren't sharp. Then you just have a larger-sized image with soft corners. And you've lost some reach (although you can crop to get it back).
2. Circular sensor? or a square sensor large to circumscribe the entire diameter of image circle. (whichever is more cost-effective).
We are no longer using film; why are we sticking to film concepts of square aspect ratios? Sure, we aren't going to usually use a circular image as our final print. But so what? Lenses give a circular image. Why not work with that?
Think about the advantages this offers. (1) You can choose your own aspect ratio. (2) No more need to tilt the camera. (3) For some shots, i.e., with fish-eye lenses, you may want a circular aspect ratio.
3. Tilt-shift sensor?
This would be a killer feature. It would effectively turn all 4/3rds lenses into tilt/shift lenses on the E4. Not much tilt is needed at all on a small format like 4/3rds to achieve dramatic results in terms of shifting the plane of focus. Simply have a few controls to tilt he sensor about it's mid-points this way or that, and provide an option to lock it in place at any location. Make sure to have a way to return it to perfectly parallel.
4. HDR built in?
Again, why not? Why do we have to take multiple shots for HDR (multiple mirror flips). It's absurd. The sensor should be able to dynamically report back to the camera body, "ok, here's the image so-far, with 1/1000th of an exposure...oh, here's an update now at 1/500ths, oh here's 1/250ths."
The way it would work is the mirror goes up one time, and you program a long exposure (say 1/60 of a second total), with captures from 1/1000th of a second to 1/60 second, cummulatively, so you have a picture at:
1/1000th
1/500th
1/250th
1/125th
1/60th
The sensor is only exposed for 1/60th of a second total; it just dynamically reports those cumulative results back to the body. The result is a really simple set of HDR shots, with less fiddling around, and less chance for camera shake. You could even program this as an option around the + - exposure compensation. I.e., the 5EV shutter bracketing. So when you say you want a 5EV bracket around your target exposure, rather than taking 5 separate shots, it can take 1 cummulative shot, which is parsed out into 5 photos.
5. A new way to think about exposure
Better: Or maybe, even better yet, rather than a cummulative shot, with the sensor exposed and periodically reporting back to the body, do a series of non-cummulative shots with the sensor reporting back to the body, and have a digital image-alignment processor in the camera merge them. This way, each individual exposure -- while under-exposed -- doesn't have as much possibility of shake. But it's still taken with 1 total 1/60th click, just reporting back to the camera at many intervals; if there was any shake not corrected by the in-body IS, that is aligned by camera firmware afterwards.
So, for a 1/60 second exposure, instead of having 1 image sent back to the camera, you'd have 16 1/1000th second images sent back, which could be aligned in-camera:
0/1000 (1/1000th exposure)
1/1000 (1/1000th exposure)
2/1000 (1/1000th exposure)
3/1000 (1/1000th exposure)
4/1000 (1/1000th exposure)
5/1000 (1/1000th exposure)
6/1000 (1/1000th exposure)
7/1000 (1/1000th exposure)
8/1000 (1/1000th exposure)
9/1000 (1/1000th exposure)
10/1000 (1/1000th exposure)
11/1000 (1/1000th exposure)
12/1000 (1/1000th exposure)
13/1000 (1/1000th exposure)
14/1000 (1/1000th exposure)
15/1000 (1/1000th exposure)
16/1000 (1/1000th exposure)
Even if Olympus can't make the dedicated hardware to do a proper alignment, it can still do a simple cummulation of all of them and give you 16 separate 1/1000th second images, which you could align later on on your computer.
I wouldn't think it would be too difficult to get the sensor to report back at 1/1000th second intervals.
Elsewhere, Rriley said that the patent specs for 4/3rds said that the image field as a max diameter of 25mm, but we currently use 21.6mm; thus, the the sensor area could be increased from 225 sqmm to 300sqmm. However, it isn't clear that the lenses are acceptably sharp in those outer areas. There is no point in increasing the sensor size if those last bits of the image circle aren't sharp. Then you just have a larger-sized image with soft corners. And you've lost some reach (although you can crop to get it back).
2. Circular sensor? or a square sensor large to circumscribe the entire diameter of image circle. (whichever is more cost-effective).
We are no longer using film; why are we sticking to film concepts of square aspect ratios? Sure, we aren't going to usually use a circular image as our final print. But so what? Lenses give a circular image. Why not work with that?
Think about the advantages this offers. (1) You can choose your own aspect ratio. (2) No more need to tilt the camera. (3) For some shots, i.e., with fish-eye lenses, you may want a circular aspect ratio.
3. Tilt-shift sensor?
This would be a killer feature. It would effectively turn all 4/3rds lenses into tilt/shift lenses on the E4. Not much tilt is needed at all on a small format like 4/3rds to achieve dramatic results in terms of shifting the plane of focus. Simply have a few controls to tilt he sensor about it's mid-points this way or that, and provide an option to lock it in place at any location. Make sure to have a way to return it to perfectly parallel.
4. HDR built in?
Again, why not? Why do we have to take multiple shots for HDR (multiple mirror flips). It's absurd. The sensor should be able to dynamically report back to the camera body, "ok, here's the image so-far, with 1/1000th of an exposure...oh, here's an update now at 1/500ths, oh here's 1/250ths."
The way it would work is the mirror goes up one time, and you program a long exposure (say 1/60 of a second total), with captures from 1/1000th of a second to 1/60 second, cummulatively, so you have a picture at:
1/1000th
1/500th
1/250th
1/125th
1/60th
The sensor is only exposed for 1/60th of a second total; it just dynamically reports those cumulative results back to the body. The result is a really simple set of HDR shots, with less fiddling around, and less chance for camera shake. You could even program this as an option around the + - exposure compensation. I.e., the 5EV shutter bracketing. So when you say you want a 5EV bracket around your target exposure, rather than taking 5 separate shots, it can take 1 cummulative shot, which is parsed out into 5 photos.
5. A new way to think about exposure
Better: Or maybe, even better yet, rather than a cummulative shot, with the sensor exposed and periodically reporting back to the body, do a series of non-cummulative shots with the sensor reporting back to the body, and have a digital image-alignment processor in the camera merge them. This way, each individual exposure -- while under-exposed -- doesn't have as much possibility of shake. But it's still taken with 1 total 1/60th click, just reporting back to the camera at many intervals; if there was any shake not corrected by the in-body IS, that is aligned by camera firmware afterwards.
So, for a 1/60 second exposure, instead of having 1 image sent back to the camera, you'd have 16 1/1000th second images sent back, which could be aligned in-camera:
0/1000 (1/1000th exposure)
1/1000 (1/1000th exposure)
2/1000 (1/1000th exposure)
3/1000 (1/1000th exposure)
4/1000 (1/1000th exposure)
5/1000 (1/1000th exposure)
6/1000 (1/1000th exposure)
7/1000 (1/1000th exposure)
8/1000 (1/1000th exposure)
9/1000 (1/1000th exposure)
10/1000 (1/1000th exposure)
11/1000 (1/1000th exposure)
12/1000 (1/1000th exposure)
13/1000 (1/1000th exposure)
14/1000 (1/1000th exposure)
15/1000 (1/1000th exposure)
16/1000 (1/1000th exposure)
Even if Olympus can't make the dedicated hardware to do a proper alignment, it can still do a simple cummulation of all of them and give you 16 separate 1/1000th second images, which you could align later on on your computer.
I wouldn't think it would be too difficult to get the sensor to report back at 1/1000th second intervals.
