FingerPainter
Forum Pro
- Messages
- 12,593
- Solutions
- 37
- Reaction score
- 13,643
No, it will be less than 1.4x as high. How much less depends on how sharp the lens is. An infinitely sharp lens would produce images 1.4 times as sharp from a 60MP sensor compared to a 30MP sensor. A not particularly sharp lens might not see very much improvement at all.
Image resolution is measured linearly, in lp/ph. It is the result of digitising the analog image cast by the lens. Lens resolution is often measured in lp/mm.
When you crop from 61MP to 15.25MP you are cutting out 1/2 the horizontal and 1/2 the vertical dimension of the image. So the number of line pairs cast by the lens on the retained part of the image is cut in half. It is digitized at the same rate (the pixel pitch didn't change) so you should expect to have 1/2 the lp/ph in the cropped image compared to what you had in the uncropped image.
However, if you kept a centre crop, and the lens is sharper in the centre than at the edges, you'd retain a bit more than half the resolution. Perhaps 55%. In contrast, using a zoom and not cropping probably yields something closer to 80-90%. Your suggested approach therefore loses 2.5-4.5 times as much resolution
Why? A 2x teleconverter gets half the lens resolution cast on the sensor, but doesn't halve the pixel rows as well.
In one case the information per pixel changed, in the other case it didn't.
As above.
Why would you assume there is additional noise? An image doesn't look noisier because it has more noise. It looks noisier because it has a lower Signal to Noise Ratio (SNR). An image with a lower SNR almost always has less noise than an image with a higher SNR but it also has an even smaller signal than the image with the higher SNR. If image A has half the SNR of image B, then image A will look noisier than image B. Image A will probably have about half the noise of image B and about 1/4 the signal of Image B. That's what you get when you crop to 1/4 of a 61MP image.
Or to put it another way, noise isn't something separate that exists alongside of pixel values. instead it is a relationship between pixel values. For any given pixel it is a relationship with all the other pixels in what you are looking at. If you change the number of pixels looked at, you change the set of relationships a pixel has even though you don't change the value of the pixel itself.
As you increase the number of pixels included in an image the effect of the noise of any one pixel (or of any specific set of pixels) on the noisiness of a whole image is reduced. The effect of the noisiness of a small set is drowned our by the greater signal and total noise of the larger set.
The noisiness of any sub-part of an image is always greater than the noisiness of the whole image. even though the pixels in the sub-part do not change,. How noisy a viewed image looks depends on the SNR of what you are looking at, which depends primarily on how much light was captured in what you are looking at.
A sub-part of an image has less light captured in it than was captured in the whole image, so the sub-part has a lower SNR, so the sub-part looks noisier than the whole image.
Last edited:
