Standard lenses Compared; Canon 50mm 1.4 and 1.8, Minolta 50mm and 45mm MC and MD

<p>Robin,<br>

"My experience is that Canon's AF is as accurate as Live View MF at 10X magnification in good light - which is pretty impressive and certainly superior to manual focussing on a microprism screen."<br>

<br /> You are incorrect. AF has more variability than you are aware of. Manual focus with a split prism is much more accurate (as proven by my test).</p>

<p>Try it yourself and share the results. Please set your lens at f 1.4, shoot 10 pictures of a detailed target indoors, and post the center crops at 100% (all of them in sequence) for comparison.</p>

<p>It is an easy test to do.</p>

<p>Most importantly: I run the test to assess my equipment and know what it can and cannot do. Sharing it is just a secondary step for those who are interested.</p>

<p> </p>

<p>Mark,<br>

If you don't shoot film with manual focus cameras, this test will be an important to you. For photographers that use film on manual focus cameras (and autofocus) this test is more relevant.</p>

<p>"AF lenses for critical tests should be focused manually...it's the AF system that may be producing a soft image, not the lens"<br>

Actually both, the performance of the AF combined with the poor performance of the Canon EF 50mm's wide open directly affect the quality of the results. <br>

Also what is your system for manual focus. Do you use film or you propose to do this with liveview?</p>

<p>Mauro, what about the autofocus system of the Elan 7e compared to that on a modern Canon dSLR? 5D, or 1D, or even their newer film cameras? I can't recall specifics right now, but I do believe the higher up cameras up have better AF sensors (cross)... what kind of sensors & how many does the Elan 7e have?</p>

<p>Furthermore, your particular body/lens combo could be back/front focusing, which would be more obvious at larger apertures. That's another source of variability; one that could be eliminated by using a 5D Mark II which allows for calibration of the body w/ specific lenses.</p>

<p>If you're just aiming to test *your* system, then, my comments are irrelevant :)</p>

<p>In all fairness, though, I <strong>am</strong> amazed at the Minolta results that show very little loss of sharpness wide open. I always see a significant loss of sharpness at f/1.8 on my Canon 50mm as well as at f/2.8 on my Canon 24-70L, tested hundreds of times on a dSLR so that I know with 99% confidence that misfocus was not an issue.</p>

<p>Cheers,<br /> Rishi</p>

<p>Mauro -- I shoot a lot of film (usually over 120 rolls/year) as well as digital, mostly using Nikon glass, often using mostly prime lenses. I also use Mamiya 645, Pentax 6x7, Hasselblad, Rollei, Holgas, Dianas, etc. I have never used a DSLR camera with Live view, and don't think I ever will. As a Nikon user, it would be all to easy for me to trash Canon glass. I see your point, and here's mine: One could exhaustively test every lens made against each other to see which ones produce the sharpest images from a test target. However... no matter how good or bad a lens may be, most are sharper than the people using them, and nobody ever asked me, when looking at a print, "was that a Minolta 45mm MD lens, or a Nikon 50mm Nikkor-H non-AI lens? One can have the sharpest lens available, but it's the sharp eye and mind that makes the photo.</p>

<p>Hi Rishi.</p>

<p>Yes, the Minolta lenses perform outstanding wide open.</p>

<p>Regarding the AF, all systems may/will be different whereas with manual focus you can be certain.</p>

<p>Since you are very scientifically oriented, can you shoot several shots with AF in with your DSLR and you 50mm at f1.8 and post all center crops at 100% (without selecting them)? It should be fairly quick and add to the value of the thread. (In my test I was indoors at some 20 feet from the target - give or take) .</p>

<p>Mark,<br>

"most are sharper than the people using them"</p>

<p>Yep. But poor focus will always hurt.</p>

<p>The reason I asked about live view is because you stated that for critical use AF lenses must be focused manually. Now how do you do that on an SLR without split prism or live view?</p>

<p> </p>

<p>It is pretty amazing that 30+ year old lenses are pin sharp wide opened, outresolving an almost 24MP scanner (with glass).</p>

<p> </p>

<blockquote>

<p>It is pretty amazing that 30+ year old lenses are pin sharp wide opened, outresolving an almost 24MP scanner (with glass).</p>

 

</blockquote>

<p>Amazing, sure. But in my book, more <em>suspicious</em> . I don't understand this regress in technology. Unless Canon specifically wants you to spend more money on a higher end 'L' series lens... e.g. the f/1.2. But as far as I've read, the f/1.4 is sharper than the f/1.2 anyway.</p>

<p>Hence, I'm sort of baffled by your results, Mauro.</p>

<p>Could you please post the full-frame of one of your shots? I want to see how much of the frame your resolution chart takes up. I'm guessing that it only takes up a small portion of the center of your frame. In which case, perhaps the Minolta only shines b/c it was optimized to be sharp at the center, whereas the Canon lenses are optimized to be sharp throughout the frame? I say this b/c I own a Minolta X-700 with the 50mm f/1.7 lens. While it is sharp in the center, results are dismal at best 30% in from either side at wider apertures (vignetting + softness).</p>

<p>So perhaps modern lenses are softer at the center in order to be sharper throughout the frame? I don't know; this is an entirely uneducated, uninformed, unscientific guess on my part!</p>

<p>Sure, Mauro, I can do some tests. Remind me what size you printed that resolution chart. Also, what settings do you render it at to not get any 'jaggies' (even with anti-aliasing, I tend to get jaggy diagonals when rendering the vector graphics original file).</p>

<p>Cheers,<br>

Rishi</p>

<blockquote>

<p>You are incorrect. AF has more variability than you are aware of. Manual focus with a split prism is much more accurate (as proven by my test).<br>

Try it yourself and share the results.</p>

</blockquote>

<p>I tried this myself using all the lenses I have on my 5DmkII. There is some very slight variability, but it as nothing compared to the inaccuracy of focussing manually on an EgS screen. In fact, the single-shot AF inaccuracy was completely trivial in comparison. I was surprised at the time. The Elan 7 may be different, of course. Even if one's eyes are superb you would have to focus using 10X with Live View to match the accuracy of the AF system on the 5D. The AF system also accomplishes this in about a second or less, unlike MF which takes considerably longer.</p>

 

<blockquote>

<p>Amazing, sure. But in my book, more <em>suspicious</em>.</p>

 

</blockquote>

<p>I agree: extraordinary claims require extraordinary evidence. Amateur lens testing in my opinion is rarely convincing.</p>

<p>Robin, my suspicion is more on Canon than on Mauro; Mauro, in general, does excellent & very scientific work.</p>

<p>Regardless of the variables in Mauro's study above that *weren't* accounted for, I have to fully agree with his conclusion that the Minolta lenses don't seem to lose sharpness at wider apertures. In my experience, the Canon wide lenses (50mm f/1.8 & 24-70 f/2.8L lenses, to be exact) lose significant sharpness when shot wide open.</p>

<p>What I'm wondering is whether this apparent 'regress' in technology is actually due to the fact that the newer Canon lenses focus on maintaining sharpness throughout the frame whereas perhaps the Minolta lenses don't. Optics is physics & you can't always win.</p>

<p>Rishi</p>

<p>Rishi,</p>

<p>The test chart was placed so it fit approximately 12 times on the negative (vertically). This will allows you to see the resolution extinction within the 1-5 marks where it is easiest to measure. The size of the target doesn't matter as long as it fits around 12 time (the largest the easier on the printer).</p>

<p>If you print with HP use max dpi, if you use Epson print with max resolution+finest detail.</p>

<p>I dont have the full shot here but I will post tonight (the corners at all apertures for you).</p>

<p>Robin,</p>

<p>There is one fact you can always count on "Good results are true. You can't make a lens perform better than its best. You can make a lens test perform worse than its best though". The Minolta sharpness is unquestionable.</p>

<p> </p>

<p>Rishi,</p>

<p>do you want just the corner crops or the entire frames. I would need an ftp for the latter.</p>

<p>Here are results from my Canon 50mm f/1.8, wide open, hand held.</p><div></div>

<p>Here is the center portion.</p><div></div>

<p>Thanks Mark.<br>

Can you please post the 100% center crops of all the consecutive shots to see the variances?<br>

Also what camera did you use (so I can compare to the 24mp scan - it appears from the center crop that you used a 6mp camera) and what distance to the target?</p>

<p>If you show 5 or 6 consecutive shots it should be enough to evaluate focusing consistency.</p>

<p>Mauro I have used that lens on a minolta srt101 camera and it was about 40 years ago. It was a fantastic lens and it made minolta one of the hottest selling cameras at the time. I really enjoy your testing it is interesting and a pleasure to read as well. Keep up the good work Michael.</p>

<p>Thank you Michael.</p>

<p>Rishi, this is what you requested. It has the center crops and the upper right crops at all the apertures.</p><div></div>

<p>And Rishi here is the entire setup (originally about 22 megapixels scan) downsampled for the post.</p><div></div>

<p>I did some quick test shots of a Canon 1.8 (borrowed it) earlier this year and to be honest, your results appear to be significantly worse than what I saw. I am a little concerned that you may have an issue with the autofocus system on your camera. I didn't keep the images I took so I cannot post them. I did both manual and auto focus. I also have a split focus screen on my 5D to insure manual focus is accurate. </p>

<p>My research on the Canon 1.8 and 1.4 showed that they were not optimized for wide open performance. A test shot of with the 1.8 focused on city lights showed significant distortion of the street lights in the corners. I looked at the 1.4 at a store and wasn't impressed with the build. In the end I got the Sigma 1.4. Distortion in the corners is much less than the Canon and from what I have read the Sigma incorporates a Aspherical element in the design. The Canon 1.8 and 1.4 do no while the Canon 1.2 does (I never tested this one). It would appear to me that the Sigma 1.4 and Canon 1.2 are optimized for wide open performance while the Canon 1.8 and 1.4 are not. </p>

<p>Steven,</p>

<p>Remember it is harder to see the shortcomings on a 12 megapixel file (5D) than on a 22 megapixel (scans I posted).<br>

Can you take several shots of your Sigma at 1.4 with AF at about 20 feet from a detailed subject and post all the center crops sequentially for comparison? I would be interested in the results.</p>

 

<p>Steven, this is what the 50mm 1.8 from my test looks wide ope if I sample down to just 12 megapixels like the 5D. It looks relatively more acceptable - but in reality is very short of the Minolta manual focus that outresolve the 22 megapixel scan wide open.</p>

<p> </p><div></div>