When we use MTF to measure the imaging performance of a lens, we normally have the sagittal and tangential curves.
What kind of picture problem will be formed if the contrast of sagittal and tangential curves differs?
in the above graphics , the tangential and sagittal curves at zero to 20 degrees are close together, while the curves differ a lot at 30 degrees.
Lets say at zero to 5mm from the image center tangential and sagittal curves are close together and at 6mm from the center they differ a lot.
Imagine a barcode |||||||| on paper as object with a pin in its center. Now you can turn this barcode around the pin into other orientations /// … or \\ or ===.
is the pin 0-5mm from the center , the barcode will look very similar in all directions.
at 6mm however, it will change it‘s contrast .. the grey value distance between black and white .. with the rotation … it will become more clearer and more blurry, then clearer then more blurry.
if the tangential MTF if higher (i gues this is more likely, then the barcode is clearest when the lines of the barcode are tangents to a circle through the image center with that 6mm radius.
Imagine the barcode is a satellite, flying around earth (the sensor center). Then the satellit tries to always show the same side to the earth… same here … if tangential MTF is higher
P1 P8 P7
P2 O P3
P5 P6 P4 In which point must the barcode have which rotation for best contrast? for example in P3 , this is ||||| P7: \\\ P6: /// Solution: P5: \\\ P4: 三 Of course we hope, that the direction of the barcoder does NOT change with direction, because if the barcode gets super fine and very pale, then in some rotationsit is just a grey mass no barcode visible. for such lenses you better ask what kind of applications the customer has … and if it sounds like barcode, better choose a diffrent lens. similar effect would occur with concentrical circles (((( )))), but these structures are really rare, you do not find them in industry. Yes is very common that lens Sagitial and tangitial seperate in the middle and edge of the field. Example: imagine threads of textiles, ||||||||||||||||||||||||| all over the image Luckily they never turn, you may say. This is correct, BUT .. The image will always look strange, because P5: \\\ P4: 三 another example : Other less suitable applications are Monitor mask inspections … this stuff looks too similar to barcodes. TFT masks and alike things. That fact that the contrast changes does not mean the lens cannot be used. Contrast still can be high enough. Just the risk is higher that it will not work and it is harder to sell, because Customers may have no ideas about optics, but they can easily detect contrast changes just by turning the object. That's why in the USAF 1951 Chart there are ||| and == patterns, by the way. see : How to read an USAF1951 target?