[biz-engineer]

Knowing how a lens will render without buying the lens can be useful. I've always had a hard time getting anything out of looking at the MTF charts provided by lens manufacturers. Usually, MTF chart are provided for lens apertures wide open, I didn't find it very useful originally. But, I now think that I nailed it, having the lenses , what I see in pictures seems to correlate pretty well with the MTF charts provided by Ricoh. For instance the D-FA 28-105 and D-FA 24-70 render corners and bokeh differently, and this can be guesstimated by looking at their MTF charts (if my conclusion isn't wrong.. of course).

Below are links to the MTF charts provided by Ricoh for the D-FA 28-105 and D-FA 24-70 lenses:
HD PENTAX-D FA 28-105mmF3.5-5.6ED DC WR / Standard-Angle Lenses / K-mount Lenses / Lenses / Products | RICOH IMAGING
HD PENTAX-D FA 24-70mmF2.8ED SDM WR / Standard-Angle Lenses / K-mount Lenses / Lenses / Products | RICOH IMAGING

The 28-105, e.g at 28mm , shows corners where image features look like smeared, this is very eye catching on a large print.
The 24-70 shows corners where the contrast of image features is softened, less attention catching on a large print.
While the DFA28-105 seem to be sharper than the 24-70 outside the center, the corners of the 28-105 are eye catching in large prints, and the 24-70 produces a more pleasing bokeh.

That difference can be seen in the MTF chart for the 30lp/mm lines.
There are two types of lines in the MTF chart: the S lines (Sagital), and the M lines (Meridional).

The S lines at 30lp/mm represent the detail contrast parallel to image circle radius.
The M lines at 30lp/mm represent the detail contract perpendicular to image circle radius.

So, when both S and M lines drop in outwards direction from the image center, detail contrast drop by the same amount is both S and M direction, which produced softened image feature.
When the S line and M lines gap increases in outwards direction from the image center, the amount of detail contrast drop isn't the same in S and M directions, which shows as if image features are being smears toward the edges of the image circle. I could observe this with the DFA28-105 and the 24-70, but my conclusion correlated pretty well with what I can observe on the DFA15-30 , DFA70-200 2.8 and DFA 150-450 (especially about how bokeh looks like). I haven't checked for the DFA 70-210.

I could confirm this finding by reading the MTF charts published by Ricoh, and knowing how my lenses render images.
Is my conclusion just by accident or not?

I know some people hate charts. But here , maybe is the chance to reconcile both image aesthetic and lab tests.

[stevebrot]

Combining the charts for a dozen examples of a given lens might give one a starting point for pin-pointing a design as opposed to a manufacturing weakness, but it is probably easier to simply glance at the charts from Ricoh and various data-driven reviews followed up with digesting hands-on reviews and examples such as those on the lens clubs and lens example databases here in PF.

BTW, I like Roger Cicala's personal approach to MTF analysis:

Roger Cicala: why I don't use an MTF bench to test my own lenses: Digital Photography Review


Steve

(...only tests new purchases and then only for obvious defects...)

[biz-engineer]

Combining the charts for a dozen examples of a given lens might give one a starting point for pin-pointing a design as opposed to a manufacturing weakness

Yes, that was my point: having an idea of how lens designs may render just by looking at the MTF charts. If I look at the Pentax D-FA *50 or D-FA*85, their charts are comparable to the MTF charts of Zeiss primes, S and M line pairs remain joint together from center to corner.

[JensE]

@stevebrot Thanks for the reference to Roger's artcile. It put a grin on my face that I have independently come to use (and recommend) the same kind of test target for 3-D testing as this master of lens evaluation . Harvested grain fields are perfect and provide a little more contrast than grass, but depending on season I use whatever is available. Instead of synthetic pinpoint light sources, I tend to wait for a (partially) clear night sky though :-)

@biz-engineer In general, I also find it very useful to correlate measurements and visual observations. It definitely helps me to understand what I (dis-)like about a particular lens and thus to know what to look out for. I've shared some visual analysis recently backed by measurements on macro lenses to understand what I've observed - it has been received with mixed reactions from basically "Oh, that's why!" to "Waste of time, I'd rather take photos!" - so it's likely just for us less artsy or more nerdy folks or those at the physical limits of photography.
Thanks for sharing your view on diverging sagittal and meridonial resolution on those lenses above, I think it's well-founded and not just accidental. It is likely only visible when there is a very pronounced difference. Otherwise other characteristics like the transition from in-focus to out-of-focus areas will dominate the visual impression, but I immediately recognize a number of my lenses where it is very visible.

[Madaboutpix]

Interesting points made in this thread. I have my own history of bothering, or not bothering too much, about MTF charts. Back in the 1980s, when I started my photographic journey, I used to study the MTF charts in the test magazines almost religiously, and developed a reasonable skill of interpreting them, at least for my layman's purposes. Over time, I read about complicating factors like sample variation and, likely not before the onset of the digital era, of various kinds of misalignments (decentering, bad spacing, tilt). Robert Cicala definitely deserves a lot of credit for educating photographers in that respect, even if his pragmatic approach is probably lost on much of the gear-sniffing, measurebating DPR crowd out there. Most recently, he helped me to understand field curvature deeply enough to get more and renewed fun out of my DA15. I also learned what an advanced RAW converter can do to mitigate various shortcomings of lenses, and I learned to appreciate the importance of human sensory evaluation of lenses, and that my personal impressions and tastes were "valid" enough in their own right and actually mattered. And I learned to appreciate "character" in a lens's rendering, without falling into the other extreme of completely dismissing lab testing, MTFs, and the like, and putting up with just about any objective or subjective flaw in a lens.

When I was still in my "mostly-lab-test-led" phase, I probably wouldn't have touched a lens like the DA10-17 fisheye, which cannot exactly be said to excel in lab tests. I'm glad that I gave it a chance anyway, as it turned out to be one of the most fun lenses in my bag, and one that gives me images that I happen to like a lot. I kinda dig what its designers, among them the much-revered Jun Hirakawa, tried to and did achieve with it.

[house]

The absolute MTF values alone are
therefore not a sufficient criterion for
predicting the subjectively perceived
image quality. The curves must be
assessed appropriately and the viewing
conditions in each case must be taken into
account.
It has been shown in many experiments
with test subjects and many different
images that there is a fairly useful
correlation between the subjective quality
assessment and the area under the MTF
curve.

Severe astigmatism is indeed very noticeable and very ugly. For my use case a it's also noticeable when there is a sudden change in the mtf. It seems a gentle slope is less noticeable than super sharp until a steep drop near the edges. The smooth drop can have lower actual resolution but look less objectionable.