Rishi Sanyal wrote:
toughluck wrote:
Rishi Sanyal wrote:
Your setup with 2/3" sensor, 50mm focal length, F4.5, with a min focus distance of - and I'm giving you the benefit of the doubt & suggesting you shot at the closest focus distance possible - let's say 0.5m, gives you a DOF of 2cm.
The iPhone image, shot at about 2cm w/ a 13mm F15.5 equiv lens has a DOF of .08cm, 25x less than yours. In fact, as I'm composing the frame it's wildly changing as the camera-subject distance changes from handshake & subject/photographer movement. Not only is it difficult to compose the frame, it's amazing AF can even keep up. Naturally, I have no control over what's in focus, and manual focus it out of the question.
With that much needed perspective, and taking into account the much smaller sensor on the iPhone, are the results surprising? Or, in context, are they perhaps impressive, if you assess the detail at the focus plane and consider the roughly 25x less depth-of-field than your example?
(Please correct me if I'm off on my calculations or assumptions)
Wow, why so defensive? I wasn't attacking you, I was pointing out how phone cameras definitely didn't break the laws of physics and they definitely are not better than advanced digital compact cameras, let along interchangeable lens cameras, despite all the claims from phone makers.
I wouldn't say I was being defensive, but I did find the title of the thread a bit hyperbolic, and your glossing over of the fact that the focal plane was at the eyelashes, not the iris, well, odd, as has been pointed out.
First of all, I mentioned that the iris in my shot would be slightly out of focus and there may be some motion blur.
Replying to your comment in the gallery (not here), yes, I shot my wife's picture straight on and you shot from an angle. I clearly said that the iris in my picture is somewhat out of focus, but still shows some detail. Your picture shows lack of detail even within the focus plane.
The eyelash at the focal plane (yes, there's one very small eyelash that appears sharpest) looks sharp to me at 100%, keeping in mind I don't expect any lens to perform its best at its minimum focus distance. I don't understand why in your comments you keep talking about the lack of detail in the iris - clearly the iris is not what's in focus. Had you seen how dramatically the composition and distance-to-subject was changing as I was trying to frame the shot (at those close distances / magnifications), you'd probably appreciate that I had no control over what would be in focus.
Also, keep in mind that nearly all browsers enlarge images to 200% on HiDPI displays when you click on '100% zoom' in our gallery viewer. If you download the image and view it at 100% in Photoshop or Preview on, for example, a recent HiDPI 13" - 16" MacBook Pro, it looks very sharp.
Focal distance in EXIF data shows 25 cm distance (shortest possible).
I used this DOF calculator:
https://www.vision-doctor.com/en/optical-calculations/calculation-depth-of-field.html
I input 3.4375 µm CoC diameter (equal to pixel pitch of 5 mpix 2/3" sensor), 50 mm focal length, f/4.56 and 250 mm working distance for my shot.
iPhone 13 Pro data: 4.032×3.024 mm sensor (1/3.27" tube size). Pixel pitch equal to 1 µm, so diagonal pixel count is 5040, and the diagonal length is exactly 5.04 mm, which means focal length equivalent is 8.585× nominal. For 13 mm equivalent focal length, the actual focal length is 1.514 mm.
For your iPhone shot, I input: 1.575 µm (equal to pixel pitch of the iPhone UWA camera sensor if it had 5 megapixel instead of 12), focal length 1.514 mm, aperture f/1.782, 20 mm working distance.
The resulting total depth of field for A1 is 0.78 mm and for iPhone is 0.98 mm. If I drop the CoC diameter to 1 µm, the DoF is 0.63 mm, but then we'd be comparing pixel sharpness, not sharpness of a resolution-independent medium like print of full screen. Even if you wanted to argue for per-pixel sharpness, DoF is in the same order of magnitude and nowhere close to 25× difference.
Thank you for that DOF calculator, as the one I was using clearly gave me some incorrect numbers. This is the one I used:
https://www.dofmaster.com/dofjs.html
... where if you enter a 2/3" sensor with 50mm focal length, F4.5, 0.5m subject distance, it indicates a DoF of 0.02m or 2cm, which is nearly an order of magnitude off! (It should be 3mm, so given my 0.5m assumption, had I used a correct calculator, I would've said 4-5x less DoF). Also, last night I was digging in the EXIF of your file for a subject distance but couldn't find it, which is why I made that assumption, but now I see the 0.25m distance in parentheses in the depth-of-field field.
So I agree with your calculations, though if we're really being technical here, I'd say that since you're drawing your conclusions from comparing the images at 100% magnification, you should actually use a CoC of 1µm for the iPhone.
But even then, as you say, the the DoFs are comparable, and the differences probably have to do with a number of things: quality of optics, optical performance at minimum focus distance, higher SNR on a 2/3" sensor, less noise reduction on the Minolta... but one can only guess without setting up a controlled experiment.
-Rishi
I'm actually going to try my hand at a controlled experiment. I have several different cameras that I want to check to find out whether detail in the iris is actually possible with small sensor cameras at all, or if diffraction will destroy all detail before it ever reaches the sensor.
If detail is lost in noise reduction, then computational features should show details at least sometimes. If it's lost because the sensor is too small and the lens is already beyond its diffraction limit, no amount of computational trickery will ever improve the result.
