jvanhuys wrote: ↑

Sun May 28, 2023 10:17 pm

Fascinating concept this pupil ratio. I must confess I've never heard of it before.

Just wondering why manufacturers don't elaborate on this more.... hmmm.

Thanks for sharing the concept.

There is some more information at http://www.photomacrography.net/forum/v ... =29&t=8895

Some lenses do mysteriously do better than others at f/22 [...] I think the most likely explanation is that they have a high pupil ratio, which results in a lower effective aperture at a given magnification, compared with an ideal thin lens (which has a pupil ratio of 1). So this is also physics. A...

FotoChris wrote: ↑

Tue May 16, 2023 11:04 am

[...] and on the other hand for M43 at 20MP the pixels are less than 3microns.

3.347 μm pixel pitch, according to my calculations at https://savazzi.net/photography/micro-four-thirds.html

[...] I could see an application in architecture and a limited application for landscape and special cases of product photography (with available/continuous light) but other than that it's a hard sell. The Olympus implementation of sensor shifting, at least in the OM-1 (I don't remember if also on ...

Doing a little thinking about the Olympus pixel shift and associated algorithm. Oly-pixel-shift.jpg The squares represent a typical Bayer layout. The circles represent the 8 successive positions of the blue sensel while shifting. Adjacent sensels of course shift in the same way. This means that each...

Yes. But each quarter pixel is shot eight times, at least twice with each Bayer filter (four times with green). Granted the values obtained in each single shot are not independent of the values of the neighboring shots, but with knowledge of the neighboring values, some of that dependence can be re...

Lou Jost wrote: ↑

Mon May 15, 2023 10:20 am

But every quarter pixel is photographed eight times.

Each shot overlaps 50% of the preceding shot (area-wise, assuming a sensel is square).

There are two Olympus pixel-shift modes, 80 Mpixel and 50 Mpixel. I mentioned them above; not really two different modes but two encodings of the same image, one with spatial redundancy reduced or removed. ...because a half-pixel shift, at best, can only overlap 50% of the neighboring Bayer filter ...

However, I do believe that a sensor-shifted 80 Mpixel image created by a 20 Mpixel sensor contains less information than an image recorded by a native 80 Mpixel sensor. Right, Olympus agrees also, downsizing their pixel-shifted jpgs to 50Mp if I recall correctly. But remember that the 80Mp sensor u...

[...]So I think it is not necessarily true that the Oly shifted image has a lower information density than the original unshifted image. You left out the added color information in the shifted image. We completely agree in this. I do believe that sensor-shift in the Olympus adds a considerable amou...

Pixel-shifted images are simply less information-dense than non-shifted images. One could perhaps try to verify this last statement by compressing pixel-shifted and non-shifted images with a lossless algorithm, and check whether pixel-shifted images compress to a higher extent than non-shifted imag...

[...] Just out of curiosity I re-sizes the 2x high-res image to 50% of its original size and then resized it to 200% and then applied a slight sharpening filter. And it looks exactly like it did before. It depends on the algorithms used for both operations of resizing. In addition, pixel shift quad...

Thanks for posting your comparison. I'm wondering why the target images are different sizes for each lens even though the pretense is that all are shot at 1:1. What am I missing? Added a new series shot with the 30 mm at 1x. I left in place the original series (shot at approx. 1.2x) because it show...

Thanks for posting your comparison. I'm wondering why the target images are different sizes for each lens even though the pretense is that all are shot at 1:1. What am I missing? Good point, there are indeed differences. In the case of the 30 mm, it seems I made a mistake when setting the magnifica...

Yes...the 60 is quite good at f/3.5 so it might have a resolution advantage. It took some time, but I got around to it: https://savazzi.net/photography/olympus-macro-lenses.html I included the Olympus 30 mm macro as a bonus. The page is not yet linked to the rest of the site, and the contents may s...