Lines per Picture Height?

[xpatUSA]

On another site, camera reviews quote resolution in "lines per picture height" (LPPH) and offer a table comparing the subject camera with others. This means that two sensors of different heights but the same pixel pitch and equal camera optical performance would show different values of LPPH.

As an example, the Sigma SD14 sensor height is about 14mm, the Nikon D50 about 16mm but they both have 7.8um pixel pitch. That would be 1795 vs. 2051 LPPH, respectively. Does the D50 really have more absolute resolution, ignoring "Foveon vs. Bayer" considerations?

What about full-frame vs. APS-C? etc. etc.

Of course, these differences would have only a small effect on "real-world" photographs but what does this Forum think about the validity per se of these units?

[Inkanyezi]

Line pairs per picture height is a real world comparable measurement for evaluation of the impact in an image, just because it is not related to pixel pitch or megapixel count, but shows how many line pairs that may be stacked into the image height in the picture at a particular contrast. It does away with megapixels and sensor sizes but makes a direct comparison between the images you get out of the system. Whether it is relevant to you is another thing.

When I look at test results, I tend to look only at the one that I find most important to me, dynamic range. All other parameters have long ago passed the point where I would object to any camera. Depending on preferences and understanding of the figures, your preference might be another, but generally, resolution, just as megapixels, in my book makes little difference.

Still, a declaration of the MTF adjusted to image height instead of just millimetres, may show a significant difference between cameras, and it might even be valuable when you are considering different size sensors as FF versus medium format.

[ajohnw]

The line pairs per picture height are often mentioned as mtf which is entirely incorrect. Where results mention the term and give a little more information MTF50 is mentioned. This the optical resolution in line pair where pure black and white lines has fallen from 100 to 50% contrast. Imagine a graph which in simple terms has 0 to 100% contrast on the vertical axis and line pairs per mm on the horizontal. The mtf isn't far of a straight line from 100% to 0 contrast to some line pairs per mm figure that depends on the F ratio of the optic, faster the better. I've lied here for simplicity. The graph is near a straight line if the horizontal axis is in units relating to the line pairs per mm where the contrast is 0. All we would see is a plain grey and no lines at all. This MTF curve is then common to all optics, the F ratio sets the zero point.

Pass one the switch from lpm to lpph as lpm always used to be used. Magazines even published resolution targets for these for free in the past to allow people to test their own lenses. The effective closeness of the lines was altered by taking the shots from a certain distance. The change is probably due to the ease of measurement.

To answer your question given lpph this has to be related back to the pixel pitch on the sensor to be meaningful in terms of a cameras' resolution - the sensor has such and such a height and the lens can image so many line pairs on it. It's a woolly area given anti aliasing filters and bayer masks. Full colour resolution needs a block of pixels, black and white doesn't but then there is the effect of the anti aliasing filter. There is also the fact that in some ways the lpph figure assumes diffraction limited optics. Camera lenses wont be in practice. Any defects in that area reduce contrast which in turn reduces resolution. The measurement technique used in tests in many ways is only a method of comparing lenses and can't really relate to resolution only approximate it. Best way to actually measure resolution is to download a resolution target off the web and photograph it. Usually several of them to get figures in the centre and edge.

-