It is such a shame from a big company like Canon to know that there is a great lens out there..but it's a bit of a lottery as to wether you will get one!
very true..hence my purchase of a new (still great price though) 24-105 zoom..that has ended up being a pretty good copy anyway!
I am going to be looking for my 24mm lens for interiors / exteriors /cityscape / landscape work soon so I guess, if I go that way, I'd better get ready for some talking with Canon!
bathman wrote:
I revisited Hubsands page again and had a look at the MTFs: The ZF25 seems to be a good sample for the following reasons!
Speaking of the B Zone, the Canon is clearly better up to f4, they are on level at f5,6, probably the Zeiss is a bit better above.
In zone C the Zeiss' published MTF shows that wide open and at f5,6 the corners are not so sharp, see that dive at 20mm! Hubsands corner tests are from the last 250px, he had to find these more or less sharp corners up to f5,6
Zeiss may have sacrificed some qualtiy in the corners to get better performance in the center where crop cameras need the highest quality lenses to resolve up to 12mp and maybe more in the futur.
I guess we will not see better samples of the Zeiss but some may get a 24L which does not reach Hubsands lens due to Canons lousy CC. ...Show more →
bathman wrote:
I revisited Hubsands page again and had a look at the MTFs: The ZF25 seems to be a good sample for the following reasons!
Speaking of the B Zone, the Canon is clearly better up to f4, they are on level at f5,6, probably the Zeiss is a bit better above.
In zone C the Zeiss' published MTF shows that wide open and at f5,6 the corners are not so sharp, see that dive at 20mm! Hubsands corner tests are from the last 250px, he had to find these more or less sharp corners up to f5,6
Zeiss may have sacrificed some qualtiy in the corners to get better performance in the center where crop cameras need the highest quality lenses to resolve up to 12mp and maybe more in the futur.
I guess we will not see better samples of the Zeiss but some may get a 24L which does not reach Hubsands lens due to Canons lousy CC. ...Show more →
Here are the MTF's again: http://www.16-9.net/lens_tests/zf25_canon24/mtf_zf25.gif
Now, even at f2.8, in the "mid-field" area, the performance shown by the MTF is pretty flat from the center with only a minor drop off -- yet the test shows it sharp in the center and definately un-sharp in the "ZoneB", mid-field -- much more un-sharp than the minor drop off in the MTF (about 5-6%) would suggest.
As for the drop-off at 20mm, the theoretical MTF's for the Canon 24L show a much more significant drop off at 20mm than the Zeiss ZF. Here are the Canon MTFs: http://www.16-9.net/lens_tests/zf25_canon24/mtf_c24l.gif
For the Canon MTF's, the Blue lines are either for f8 or f5.6 and the black lines are wide open. Solid lines are radial measurements, dashed lines are tangential measurements. Thick lines are for 10 lines/mm, the thin (lower) lines are for 30 lines/mm. (Zeiss' MTF charts show line for 10, 20 and 40 lines/mm).
The tip of the corner on a full frame 24x36mm sensor is about 21.5mm from the center. The corner crop shown (the satellite dish) is less than that from the center (it is lower and in a bit from the actual corner) -- it is no more than 20mm from the center. At f5.6 the Zeiss performance at 20mm from the center, according to the MTF's should still be reasonably sharp (the actual MTF value at 40 lines/mm is still 60%, and it is 80% at 20 lines/mm -- and I can show you a number of Zeiss Contax lenses widely considered and shown to be sharp with similar values in the center). This is just one example in this test of where this lens does not appear to be living up to its measured MTF's and why this might be a bad sample of the lens.
This is getting deep into the detail. Yes, we should be out taking pictures instead of measurbating all this. But Zeiss publishes actual MTF test data for a reason. It gives us a a pretty good idea what to reasonably expect from a lens. When performance in use deviates from what the measured MTF's would lead us to expect, then we can reasonably surmise that there might be a problem with the lens. To reject this as a possibility is not sound reasoning.
Maybe you are right LotusM50, I am not able to compare Canons and Zeiss published MTF. My comments were based on what I saw in the Zeiss data sheet, and maybe I go to far there.
Can you or anybody explain me what these lines, marked green, have on the performance of a lens?
I've read in Canon literature that the more "in line" the solid and dashed lines are for a given line pair (you've marked in green the line pair that measures 40 lines/mm) the more natural/pleasing the bokeh will be.
StevenPA wrote:
I've read in Canon literature that the more "in line" the solid and dashed lines are for a given line pair (you've marked in green the line pair that measures 40 lines/mm) the more natural/pleasing the bokeh will be.
Right, I've always thought of the sagital bars (solid lines on the MTF) forming something that looks like the spokes on a bicycle wheel if fully drawn out. The tangential bars (dashed lines) would basically form concentric circles.
But John, maybe you can answer this. How does image quality differ if the line pairs for a given frequency (20 lines/mm, for example) are "in line" with each other and close together versus if they follow the same "in line" pattern but are farther apart? Is this a bokeh thing?
Edit: poor phrasing.
Edited by StevenPA on Feb 08, 2007 at 12:36 AM GMT
Perhaps the use of the X for MTF measurements seems to be able to favor the lens designers, IOW, inadvertantly bias the results towards good readings. The 90 degree orientation of the grid target on hubsand's tests are appropriate for neither sagital nor tangential measurements, so comparing his grid to the MTFs is a bad comparison.
Sure, the grid is a real life example, and perhaps the Canon really shines in this particular case, but I would like to see what each lens does with an test chart of actual X's as in Van Walree's examples. This may solve the discrepency between what we are seeing vs. the MTF curves.
Here is another theory to consider: Regardless of whether the targets consists of x's or +'s, hubsand's test shots are being recorded on a sensor run through Bayer demosaicing. This is hugely different than testing with SPUR or TechPan film. The pixels are arranged in geometric fashion, where film grain is quite random. This may account for why the grid looks crappy in one test -perhaps the geometric pixel orientation was a-skew just a hair relative to the orientation of the target +'s for the Zeiss tests, but was much closer in geometric pixel orientation for the Canon tests. It may then be mere chance that the Canon tromped over the Zeiss in hubsand's grid close-up.
Thanks, John. Yes I know this link just posted it elsewhere , so what do you think does this explain the poorer but not so bad performance of the Zeiss at f5,6?
Zone B here in question shows that pic with the chair and not only the grid is blured, also the rest of the image seems to be a bit OOF or general unsharpness, however you want to call it
Assuming the C24L MTF is f5.6, I've separated out and superimposed the ZF25 and C24L's 10lpmm and 30lpmm data. Because Zeiss publish 20 and 40, I've split the difference to arrive at a 30lpmm figure that compares exactly with the Canon.
In so doing, I've discovered that Canon's figures are based on a slightly smaller image circle than Zeiss' MTF data, which runs out to about 23mm (beyond the frame). Hence the apparent truncation of the Canon's chart (in red). http://www.16-9.net/mtf_zf25_24l.gif
Assuming the Canon data is f5.6 (which we don't know for sure) a perfect ZF25 resolves detail in the corners (Zone C / 17-22mm) rather better than the 24L, but is less sharp in Zones A (0-7mm) and B (8-16mm). Somewhat confusingly, this is diametrically opposite to what we find in RL.
bathman wrote:
About your thought on grids and MTF.
Zone B here in question shows that pic with the chair and not only the grid is blured, also the rest of the image seems to be a bit OOF or general unsharpness, however you want to call it
Which brings us to the initial conclusion- The Canon was an exceedingly good sample and the Zeiss was a dud. Even if Zeiss and Cosina is using Denning statistical sampling or whatever, human error still exists.
hubsand wrote:
Assuming the Canon data is f5.6 (which we don't know for sure) a perfect ZF25 resolves detail in the corners (Zone C / 17-22mm) rather better than the 24L, but is less sharp in Zones A (0-7mm) and B (8-16mm). Somewhat confusingly, this is diametrically opposite to what we find in RL.
Thanks Hubsand.
Now, here in this thread is something we can learn, how about making a link on your 16-19.net to this thread?
is there any (practical) way to build actual MTF's from hubsands ZF25 ? comparing those MTF's to the ones published by zeiss ... wouldn't that tell us for sure whether Mark got a dud ? but i don't know how you build actual MTF's.
, so what do you think does this explain the poorer but not so bad performance of the Zeiss at f5,6?