Re: Sony announces A7R III and FE 24-105mm f/4 OSS lens today!
DannyBurkPhoto wrote: mjm6 wrote:
I'm not him, and this may not be what he is getting it, but this is how I see it:
This isn't pixel shift to increase the image resolution or file size (pixel dimensions), it's de-bayer-filtering the pixels, so this is really completely different from a stitching approach. The file size remains the same, but the ability of the file to hold onto detail increases because the data is collected for R, G, and B at ever location, rather than being approximated from nearby pixels.
It would be possible to do that if they were to shift by 1/2 a pixel and then make a set of shots. It could feasibly be done by shooting a single 1/2 pixel shifted to the left and down, for example. However the much better implementation would be to shoot four with a full pixel shift over to de-bayer and then shoot another set of four 1/2 pixel shifted, for a total of 8. I think this is what Olympus does to double their resolution.
If you do a lot of copy work or art reproduction, this might produce a substantial improvement in sharpness from a 42MP file (assuming you have the optics to pull this off). I'm not sure how useful this will be for much of anything else the way they have it implemented.
Ill see if I can find more data
I'm slightly familiar with the vague concept of pixel shift, but not from actual use. I was hoping for more actual MP resolution from a new camera, but this could be interesting for those times when it's usable.
I'm not convinced that stitching multiple frames isn't the best way to gain resolution vs pixel shift. If the pixel dimensions of an image are the same as one gets from A7RII, that image still needs to be interpolated upward to reach a larger size for printing, whereas multiple stitched images may be large enough with no interpolation required. Am I understanding correctly that, in theory, a pixel-shifted image from A7RIII can create a larger print size yet retain the same amount of detail as an image from A7RII, similar to what it might have been from, say, a 70 MP single frame capture, even though it has to be interpolated to reach that larger print size?
This is not exactly new but what essential it is doing is taking four shots and each pixel gets multiplied by 4 than which will add bit depth to the file and color data as well. than takes that data and merges it together . Its more than this and overall better than stacking as stacking only adds more elbow room for enlargement this actually improves the pixel data itself. Thats how I understand it
I could possible saying this in the wrong way so excuse me if its definition is slightly off
Oct 25, 2017 at 01:00 PM
Previous versions of GMPhotography's message #14229285 « Pre-orders open! Sony A7R III and FE 24-105mm f/4 OSS lens! »
