[lytrytyr]

Sigma AF 30mm f/1.4 EX DC (Pentax K) Review / Test Report - Analysis

"The borders are only fair at wide-open aperture but in the extreme corners there isn't much resolution left."


http://www.lenstip.com/190.4-Lens_review-Sigma_30_mm_f_1.4_EX_DC_HSM_Image_resolution.html

"The maximum aperture range is basically useless"

22 lp/mm at the edge of the MFT frame.

---

Last edited by lytrytyr; 07-27-2012 at 11:32 AM.

[Digitalis]

The poor resolution at full aperture off-center may be making it harder for the PDAF system.

This displays a complete misunderstanding of how PDAF systems work - PDAF works by seeing contrast between light and dark at moderate spatial frequencies, the resolution of a lens has little to do with how well a lens will AF. This is also why flare tolerance is such an important factor with lenses - because flare kills contrast, and with an AF lens like my sigma 100-300mm f/4 - focusing in strongly backlit situations can be a real pain, even with the lens hood attached.

Lack of contrast is also why the Canon 50mm f/1.0 was discontinued - it had such low contrast at f/1.0 that the AF accuracy was being compromised because of the lack of contrast from the lens - the Canon 50mm f/1.2L addresses this somewhat successfully by having a higher level of inherent contrast - but overall the 50mm f/1.2L is not exactly the highest resolution super-speed lens in the market. Resolution and Contrast are diametrically opposed,it is a very careful balance to be maintained. You can make a lens with extremely high contrast (The DA 16-45mm f/4 springs to mind) but it will in all probability have average resolution, you can make an extremely high resolution lens that has low contrast (Minolta 50mm f/1.7 springs to mind). Zeiss make lenses with high resolution - Leica makes lenses with high contrast - this is a factor that is most apparent in RF lenses.

[lytrytyr]

PDAF works by seeing contrast between light and dark at moderate spatial frequencies

At which frequencies?

[Designosophy]

My Sigma 30 f/1.4 has good contrast wide open - better than my FA 50 f/1.4, but I have found that the center AF point is most reliable. The left-most cross-type points are pretty bad, but the right-most are better. I can't use the far right and left points at all. With the FA 50 f/1.4, I can rely on all of the AF points pretty well. Is the contrast you're talking about something other than the contrast I see in photos?

[lytrytyr]

My Sigma 30 f/1.4 has good contrast wide open - better than my FA 50 f/1.4, but I have found that the center AF point is most reliable. The left-most cross-type points are pretty bad, but the right-most are better. I can't use the far right and left points at all. With the FA 50 f/1.4, I can rely on all of the AF points pretty well. Is the contrast you're talking about something other than the contrast I see in photos?

Since Digitalis appears to have misinterpreted the post
where I hypothesized about the off-center PDAF difficulties with the Sigma 30/1.4,
it may be helpful to clarify the word "resolution" first
(contrast is also involved).

There are at least three uses of the term "resolution" that are relevant.

"Scientific" resolution is basically the same as "spatial frequency,"
a quantity measured in line pairs per millimeter or picture height.

"Photographic" resolution is what gets measured in MTF graphs.
It is specified by a contrast level at a particular spatial frequency.
Photozone and Optyczne/Lenstip present the frequency for a 50% contrast level.

"Perceived" resolution is what was tabulated by Yoshihiko Takinami:
http://www.takinami.com/yoshihiko/photo/lens_test/index.html
The distinction between that and "photographic" resolution
is very apparent in Takinami's listings for the FA*24/2 and M100/2.8.
Takinami rates the former very highly, even off-center, while Photozone does not.
Takinami suggests the latter has poor "perceived" resolution (on film),
as opposed to the high "photographic" resolution
documented by the official MTF curves I posted in my review at
SMC Pentax-M 100mm F2.8 Reviews - M Prime Lenses - Pentax Lens Reviews & Lens Database

In post #46 of this thread, I was using "resolution" in the sense of "photographic" resolution,
while Digitalis seems to have thought I meant "scientific" resolution.

My hypothesis was that the off-center point AF difficulties with the 30/1.4
are due to low "photographic" resolution at the spatial frequencies needed for the AF system.

Designosophy, the contrast you are seeing in your photos
is probably more like the "perceived" resolution, at least at low spatial frequencies.
I do not have personal experience of either the Sigma 30/1.4 or FA*24/2,
but the Photozone tests suggest that their "photographic" resolutions behave similarly.

[Digitalis]

My hypothesis was that the off-center point AF difficulties with the 30/1.4 are due to low "photographic" resolution at the spatial frequencies needed for the AF system.

I appreciate your attempt to understand but - have you ever seen the physical size of the AF points on a pentax DSLR AF module? after a while of use one quickly learns that the AF points are considerably bigger, and arranged differently than what appears in the viewfinder. Lens resolution in either your "scientific" or "photographic" categories are nothing more than insignificant variable as far as AF system is concerned. The AF system isn't looking for small details at MTF 5% it is looking at contrast differentials 100X bigger than that.

[lytrytyr]

have you ever seen the physical size of the AF points on a pentax DSLR AF module? after a while of use one quickly learns that the AF points are considerably bigger, and arranged differently than what appears in the viewfinder.

From Autofocus - Wikipedia, the free encyclopedia :

"AF sensors are typically one-dimensional photosensitive strips (only a few pixels high and a few dozen wide)"

[Digitalis]

"AF sensors are typically one-dimensional photosensitive strips (only a few pixels high and a few dozen wide)"

now you are just cherry picking facts and using them to support your own claims, and people who do that often have very little knowledge on what they are talking about. You haven't seen an AF module from a modern DSLR? most disappointing - because if you did you would have a better Idea of how far off the mark you are.

Every DSLR AF module has more than one or two pixels on it, the pixels are typically very large so that they are reliable even under low levels of light. AF arrays are found most often in rows of pixels arrayed in a pattern defined by the manufacturer - cross type pixel arrays are sensitive to both horizontal and vertical contrast - there are also double cross arrays that are able to focus on details running at an angle across the pixel array and single strip arrays are only sensitive in one direction - the AF sensors aren't single pixels they are large groupings of pixels.

here is an image of the AF module from the Canon EOS 1DX
the active area of the AF module is approximately 16.7mm X 11mm
the total resolution of the module would be around 0.8Mp with a pixel pitch around 50μm

[creampuff]

...I hypothesized about the off-center PDAF difficulties with the Sigma 30/1.4,

...I do not have personal experience of either the Sigma 30/1.4 or FA*24/2

That's all that people need to know.

[lytrytyr]

That's all that people need to know.

Ha Ha.

But no help in explaining the difficulties that Rondec and Designosophy have reported.

[lytrytyr]

now you are just cherry picking facts and using them to support your own claims

No, just trying to bring clarity to your comments about the size of AF points.

[Digitalis]

No, just trying to bring clarity to your comments about the size of AF points.

And all i'm saying is that the AF points are bigger than a single pixel, and the fact that AF sensors cannot resolve high frequency detail - because they don't need to. So whether a lens has high resolution in the corners or not there will always be enough information for the AF to work.

My Sigma 30 f/1.4 has good contrast wide open - better than my FA 50 f/1.4, but I have found that the center AF point is most reliable. The left-most cross-type points are pretty bad, but the right-most are better. I can't use the far right and left points at all. With the FA 50 f/1.4, I can rely on all of the AF points pretty well. Is the contrast you're talking about something other than the contrast I see in photos?

The answer to that question is obvious - de-centering, and its effects of field curvature on your particular copy of the Sigma 30mm f/1.4. Field curvature is a real pain to work around, because it is a characteristic that changes with the focus distances used as well as the F stop the lens is used.Vignetting can have a bad effect on AF - however that is more problematic for slower lenses like the DA15 mm f/4 where AF sensors cannot get enough light.

[snake]

Thus far, I have not seen too many examples where the FA31 outright beats the Sigma 30 1.4. In fact, some poor examples I've also seen from the FA31, so it seems, much like every other lens, highly dependent on the photographer behind it.

To be honest, I'm tipping in the favor of the Sigma 30. Since I have an FA35 already, it's a moot point. Tehy turn up on the used market for cheap all the time, but it's not so much a FL difference to move away from the FA35 and F2 us fine enough for most low-light conditions, yet with good sharpness from wide open.

[Digitalis]

Thus far, I have not seen too many examples where the FA31 outright beats the Sigma 30 1.4.

I have seen enough shots from the 30mm f/1.4 to know it is has strengths, but corners sharpness isn't among them - and the bokeh from it is a real mixed bag too because it has some pretty horrendous coma which isn't helped by the fact that it commonly has a poorly polished aspherical element in it's design.

[gtl]

My Sigma 30mm copy colors are good enough for me. Its little soft from near corners with big apetures but thats ok and understandable. But AF is not turstable at all and it makes pics keeper rate really very low but when focus is ok pics are nice. I have been using my sigma 30mm with K100D and K5 bodies and bo0th have same problem especially in low light. Maybe ur copy of this lens is better...

I have heard of sigma lenses requiring calibration on camera bodies. Why not look into that with your local sigma service centre?