nugat Posted February 13, 2009 Share #1 Posted February 13, 2009 Advertisement (gone after registration) Sean Reid posted his full review of G1. To me the biggest surprise is the behaviour of Leica lens on The G1. They hold the resolution in the center but there is a significant fall off in the corners of the 4/3 sensor AS COMPARED to M8. Sean puts forward a hypothesis that these lenses are not really optimized to the 4/3 sensor (as they are to film M's and M8). It is surprising to me as the 4/3 sensor is smaller (25% of FF compared to 56%) than M8 and uses the "sweetest spot". If this fall-off was a fact, the use of Leica (M -mount) lens on G1 would be questionable. Sean, are you sure it is not due to the smaller DOF on the G1 sensor? How big was the original set up and from what distance you took the pictures? Did you try focussing on the corners of the set up? Link to post Share on other sites More sharing options...
Advertisement Posted February 13, 2009 Posted February 13, 2009 Hi nugat, Take a look here Reidreviews.com on G1. I'm sure you'll find what you were looking for!
Mauribix Posted February 13, 2009 Share #2 Posted February 13, 2009 My guess is that this behaviour is somehow related to the micro spheres that are present on the M8's sensor, optimized for the M lenses to be used on a reflective surface such as that of the CCD. Then, the micro 4/3 lenses are telecentric as well, while M lenses are not... this, together with the absence of the micro-sphere, should be enough to justify the fall off behaviour of the M-lenses in the corner of such a smaller sensor too. my2cents Link to post Share on other sites More sharing options...
innerimager Posted February 13, 2009 Share #3 Posted February 13, 2009 Interesting article and interesting timing for me as I just sent my G1 off to a new owner. I was interested in trying both Leica and Nikon lenses with it, and definitely was considering it a "2XTC" for my longer lenses. But I couldn't get past what Sean rightly gives as the 1st question- do you like working with an EVF? I don't. Not only do I get that "looking through a tunnel at a TV", I also get a sense that I'm watching myself take take the image. The experience derealizes my connection with the subject. Added to that, I really can't abide the blackout. So I gave up before getting the adapters. Now we see there is a degradation of RF lenses in the corners, so I am less frustrated by not getting the chance to try it myself. My lesson from a few days with the G1 is that I want a more traditional experience, a natural view of the subject, with no interruptions. The fact that I took the money from the G1 and put it toward a second M7 with a 0.85 finder sort of says it all for me. best....Peter Link to post Share on other sites More sharing options...
sean_reid Posted February 13, 2009 Share #4 Posted February 13, 2009 Sean Reid posted his full review of G1.To me the biggest surprise is the behaviour of Leica lens on The G1. They hold the resolution in the center but there is a significant fall off in the corners of the 4/3 sensor AS COMPARED to M8. Sean puts forward a hypothesis that these lenses are not really optimized to the 4/3 sensor (as they are to film M's and M8). It is surprising to me as the 4/3 sensor is smaller (25% of FF compared to 56%) than M8 and uses the "sweetest spot". If this fall-off was a fact, the use of Leica (M -mount) lens on G1 would be questionable. Sean, are you sure it is not due to the smaller DOF on the G1 sensor? How big was the original set up and from what distance you took the pictures? Did you try focussing on the corners of the set up? Hi, No, my hypothesis is that the lens and the camera do not interact well together in the outer zones. There's nothing wrong with the lens design or the camera design, per se. They just don't mesh. I'd caution people not to generalize the results to all RF lenses. We'll have to look at each one on a case by case basis. As for the method questions...I expect a whole bunch of them coming from people who will initially be surprised by these results. I'm about to be be travelling and photographing intensively so I won't be able to answer them all. The answers about camera distance, aperture, etc. are all in the article. Whenever I publish a result that disappoints some people there are inevitably huge debates over methods, etc. When such arise, I suggest concerned folks replicate the testing as best they can (keeping the camera perfectly aligned with their test board and using minute focus bracketing as described in the methods article) and see what they find. Bottom line...Leica was not exaggerating when they talked about the challenges of designing a camera that would work well with existing RF lenses. Even with a 2X crop, those challenges still exist - at least with one lens and likely with some others. I encourage G1 owners to test various RF lenses on the camera and report back on how they do in the outer zones. I think some will do well - depending on the lens design. Alas, despite this being a large review that looks at many of the G1's strengths I predict that some will center on this Summicron 28 resolution result and conclude that the camera did poorly in the review. That, of course, is far from the truth. BYW, the G1 doesn't just hold the resolution on center with that lens, it visually matches the M8 on axis. Cheers, Sean Link to post Share on other sites More sharing options...
sean_reid Posted February 13, 2009 Share #5 Posted February 13, 2009 My guess is that this behaviour is somehow related to the micro spheres that are present on the M8's sensor, optimized for the M lenses to be used on a reflective surface such as that of the CCD.Then, the micro 4/3 lenses are telecentric as well, while M lenses are not... this, together with the absence of the micro-sphere, should be enough to justify the fall off behaviour of the M-lenses in the corner of such a smaller sensor too. my2cents I think you're instincts on this are good. Cheers, Sean Link to post Share on other sites More sharing options...
sean_reid Posted February 13, 2009 Share #6 Posted February 13, 2009 Interesting article and interesting timing for me as I just sent my G1 off to a new owner. I was interested in trying both Leica and Nikon lenses with it, and definitely was considering it a "2XTC" for my longer lenses. But I couldn't get past what Sean rightly gives as the 1st question- do you like working with an EVF? I don't. Not only do I get that "looking through a tunnel at a TV", I also get a sense that I'm watching myself take take the image. The experience derealizes my connection with the subject. Added to that, I really can't abide the blackout. So I gave up before getting the adapters. Now we see there is a degradation of RF lenses in the corners, so I am less frustrated by not getting the chance to try it myself. My lesson from a few days with the G1 is that I want a more traditional experience, a natural view of the subject, with no interruptions. The fact that I took the money from the G1 and put it toward a second M7 with a 0.85 finder sort of says it all for me. best....Peter Hi Peter, Yes, it is an important question. I just want clarify, again, that the result under discussion is from one RF lens on the G1. It's important that we not generalize without data and since I'm headed off for an extensive shoot that data (with other lenses) can't yet come from me. But it could come from other photographers if they test carefully. Cheers, Sean Link to post Share on other sites More sharing options...
Bo_Lorentzen Posted February 13, 2009 Share #7 Posted February 13, 2009 Advertisement (gone after registration) Seems to me that there is a ovious difference in technology here... after all, the 4:3 system have always been centered around lenses projecting the light more "straight" at the chip, and also the chip is made for this light arriving at a right angle... this have always been part of the 4:3 claim to fame. The Leica lenses works at a much steeper angle, the result of great center image rendering and poor edges seems very consistent with the theory expectations to the match of G1 and M lenses. . Link to post Share on other sites More sharing options...
sean_reid Posted February 13, 2009 Share #8 Posted February 13, 2009 Yes, that seems to be the case with at least one RF lens and probably some others. It's interesting that this is still a factor with a 2X crop. It surprised me, in fact. BTW Bo, your coding template works well and I'll try to write it up soon. Cheers, Sean Link to post Share on other sites More sharing options...
sandymc Posted February 13, 2009 Share #9 Posted February 13, 2009 Sean, Any chance you could repeat the test with one of Leica's more telecentric lenss - e.g., a 75mm? Likewise maybe another wide angle - a 21 mm or something? The results from the 28mm are ....surprising. Also, when you did the focus bracketing, did you check the corners for any sign of focus shift? Aka the corner resolution getting better as the center moved out of focus? The corners just look plain out of focus to me. Umm, and there's nothing in the adapter that could be in the light path? Regards, Sandy Link to post Share on other sites More sharing options...
Bo_Lorentzen Posted February 13, 2009 Share #10 Posted February 13, 2009 Sandy, For what its worth, the 75lux seem to show less corner softness. Though I have only been playing with it so far, no actual test targets. the big difference is that the exit lens on the 75mm is much further removed from the sensor causing the light to hit the sensor at a more 90deg angle. or coming straight in so to speak.. Im guessing the difference between straight in and spreading out from the wide-angle M lenses is also the key to the corner softness. Thank you Sean, glad you like it. . Link to post Share on other sites More sharing options...
sean_reid Posted February 13, 2009 Share #11 Posted February 13, 2009 Hi Sandy, No more formal studio camera or lens testing for me until mid-April. The G1 is packed to ship out tomorrow and I'm prepping for an extended shoot. It's not field curvature in the lens itself - note the M8 results. The adapter is just a fancy spacer. I know you're methodical Sandy. If you have the lens and a G1 you'll be able to replicate this and you'll be surprised again. <G> My prediction is that people will debate this and then someone else will do some very careful testing, replicate the results and gradually people will accept that this is what happens with certain RF lens X and the G1. Both kids sick and letting my wife sleep right now so I'm on nurse duty. I'll be up for awhile. Cheers, Sean Link to post Share on other sites More sharing options...
sean_reid Posted February 13, 2009 Share #12 Posted February 13, 2009 Sandy,For what its worth, the 75lux seem to show less corner softness. Though I have only been playing with it so far, no actual test targets. the big difference is that the exit lens on the 75mm is much further removed from the sensor causing the light to hit the sensor at a more 90deg angle. or coming straight in so to speak.. Im guessing the difference between straight in and spreading out from the wide-angle M lenses is also the key to the corner softness. Thank you Sean, glad you like it. . For reference, that test was done at F/4.0 at about nine feet focus distance. If you do similar tests make sure you're vertically and horizontally square to the board (wall, etc.) and bracket the focus to be sure you're hitting peak res. on center. Tiny changes in the focus ring will vary the resolution. If the focus bracketing process is sending you over the edge in boredom it is probably the right amount. Cheers, Sean Link to post Share on other sites More sharing options...
barjohn Posted February 13, 2009 Share #13 Posted February 13, 2009 Sean, Interesting article and I am still digesting it. A couple of notes. One, I think on at least one of the image pairs you have the G1 and RD-1 images reversed from the proper labeling as the RD-1 is showing as the larger (more magnified image) rather than the G1 with its 12mp. Also, for the noise testing it would be fairer and easier to tell if both images were the same size. When one image is significantly more magnified than the other it is much harder to visually compare them. Either down rez the G1 or up rez the M* and RD-1. Finally, DXO showed the M8 ISO to be spot on and the G1 to be under rated but this doesn't match your finding. It makes it hard to tell what to believe. DXO for M8 shows 151 (160), 304 (320), 619 (640), 1256 (1250), & 2535 (2500). For the G1 they show 129 (100), 266 (200), 540 (400), 1117 (800), 2083 (1600), 4055 (3200). This is considerably different than your readings. Thus when you were comparing the M8 at an actual 2535 you were comparing to the G1 at 4055 and not a fair comparison. Even at 320 yo were comparing 304 to 540? Just wondering. If there is anything I would expect DXO to get right it is the sensor ISO. Link to post Share on other sites More sharing options...
sean_reid Posted February 13, 2009 Share #14 Posted February 13, 2009 Sean, Interesting article and I am still digesting it. A couple of notes. One, I think on at least one of the image pairs you have the G1 and RD-1 images reversed from the proper labeling as the RD-1 is showing as the larger (more magnified image) rather than the G1 with its 12mp. Also, for the noise testing it would be fairer and easier to tell if both images were the same size. When one image is significantly more magnified than the other it is much harder to visually compare them. Either down rez the G1 or up rez the M* and RD-1. Finally, DXO showed the M8 ISO to be spot on and the G1 to be under rated but this doesn't match your finding. It makes it hard to tell what to believe. DXO for M8 shows 151 (160), 304 (320), 619 (640), 1256 (1250), & 2535 (2500). For the G1 they show 129 (100), 266 (200), 540 (400), 1117 (800), 2083 (1600), 4055 (3200). This is considerably different than your readings. Thus when you were comparing the M8 at an actual 2535 you were comparing to the G1 at 4055 and not a fair comparison. Even at 320 yo were comparing 304 to 540? Just wondering. If there is anything I would expect DXO to get right it is the sensor ISO. Hi John, The tables are labelled properly. Which ones are you thinking of where you thought the magnifications did not match? They all look right to me (I just doubled-checked). Any chance you reversed them in your memory? There are pros and cons to all sorts of versions of noise test comparisons. Up and down res'ing changes the appearance of noise. Sometimes I do it as an additional section. The M8 and G1 are close enough in pixel res. (MP) that its not worth resizing. One could down-res the G1 files for comparison with the R-D1 files. It should help them but it won't be a silver bullet. I might add a section like that when time allows. DXO's results do not match my own in a number of areas and, frankly, I know which data I believe. Reread the section on ISO accuracy and draw your own conclusions. The suggestions you're making may match what DXO tells you but they don't match reality. DXO claims to be measuring RAW files before conversion. I'm measuring converted files (the form we actually use to make prints). But, of course, trust whatever sources you prefer. DPreview and my site tend to see the same results in ISO testing while DXO sees quite a different result. You may want to reread the ISO section and consider all the different variables that can affect measured ISO performance. I'm quite confident in the methods I'm using. Remember that if two cameras are set to the same nominal ISO and same shutter speed (while using the same lens at the same aperture) and render pictures of about the same overall brightness, they're delivering about the same effective ISO. Again, look at the samples in the ISO accuracy section. If you're really wondering about these ISO assessments, get a calibrated meter and start making some camera tests to see what's what. I think you may be surprised at what you discover. Cheers, Sean Link to post Share on other sites More sharing options...
vanhulsenbeek Posted February 13, 2009 Share #15 Posted February 13, 2009 Sean Reid posted his full review of G1.To me the biggest surprise is the behaviour of Leica lens on The G1. They hold the resolution in the center but there is a significant fall off in the corners of the 4/3 sensor AS COMPARED to M8........ That indeed is my experience as well with the WATE . More at 16mm than at 21mm. At 16 mm weird things happen in the corners. Discovered this two days ago, so no biased mind here. But apart from the fall-off, I must say the WATE pictures look rich and snappy on the G1. No regrets there. Link to post Share on other sites More sharing options...
nugat Posted February 13, 2009 Author Share #16 Posted February 13, 2009 That indeed is my experience as well with the WATE . More at 16mm than at 21mm. At 16 mm weird things happen in the corners. Discovered this two days ago, so no biased mind here. But apart from the fall-off, I must say the WATE pictures look rich and snappy on the G1. No regrets there. Did you shoot JPEG or RAW? What software in post? Link to post Share on other sites More sharing options...
vanhulsenbeek Posted February 13, 2009 Share #17 Posted February 13, 2009 Did you shoot JPEG or RAW?What software in post? Always in RAW. LR and Silkypix: no difference. Mind: this was just what I discovered, so no scientific test. What I see in the corners is some fall-off in detail and some distortion, but I would have to find a brick wall and do it again, before posting anything here. No chance today: pouring with rain! Link to post Share on other sites More sharing options...
nugat Posted February 13, 2009 Author Share #18 Posted February 13, 2009 This is really interesting. Almost as if the Panasonic 4/3 sensor was optimized for Panasonic M4/3 lens and not so good for others. There already is a lot of processing going on both between the panasonic lens( eg. outstanding kit lens) and body for JPEG and in post for RAW (LR2.2, Silkypix) that we don't know about. Maybe Panasonic's route to great pictures on 4/3 is more processing of the so and so raw data? Human eye is not that great a lens, but works miracles together with the brain. The down side could be worse performance of otherwise excellent traditional lenses. Indeed more research an experiments are required, Link to post Share on other sites More sharing options...
sandymc Posted February 13, 2009 Share #19 Posted February 13, 2009 It's not field curvature in the lens itself - note the M8 results. The adapter is just a fancy spacer. Sean, Sure looks like field curvature! I'm wondering is there's not something that's giving that effect on the sensor. The other possibility is that the AA filter just can't handle light with a hight angle of incidence But the adapter isn't just a spacer, especially for a Leica M lens; there is the potential (actually, high probability) for light hitting the sides of the adapter and scattering - could give all sorts of strange effects. I'm afraid the chances of me getting a G1 just fell dramatically!!! But anyway, enjoy the break. Sandy Link to post Share on other sites More sharing options...
sean_reid Posted February 13, 2009 Share #20 Posted February 13, 2009 I can't seem to quote messages after this latest forum update. Sandy - It does look a lot like field curvature but, of course, we know that it isn't - at least not in the traditional sense of the lens itself having field curvature. Nugat - I suspect you'll find that many of the SLR lenses do well on the G1. So too, I'm predicting, will many of the more "telecentric" RF lenses. Panasonic never claimed that this camera was optimized for RF lenses - that's a "feature" we photographers have created. I'll bet we see excellent corner to corner performance from the mFT 20/1.7. Sander - Good - the replication information flow begins. If people test carefully for this, they're going to see it with certain RF lenses. Till then, I imagine I'll be villified for awhile in some discussions. <G> Cheers, Sean Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.