Luminous Landscape Forum
Equipment & Techniques => Cameras, Lenses and Shooting gear => Topic started by: situgrrl on June 29, 2006, 10:15:31 am
-
I bought an Olympus E1 about 18 mths ago as my first DSLR after weighing it up against the usual suspects. My decision was based upon price and the quality of the lenses on offer.
Unfortuately, it has failed to perform up to my expectations in a number of circumstances, I find it's low light (high ISO) performance shockingly poor and am disappointed that I cannot enlarge beyond 9x12 in most circumstances. It's manual focussing ability is also seriously lacking and it's AF slow.
However, I love the build quality (though not the weight!) and the quality of both the 14-54 and 50-200 (which whilst I do not own, have rented on a number of occasions)
I have been holding out for Olympus to upgrade this camera but there still seems to be no announcement and I'm stuggling with what I've got.
My priorities of decent performance to 3200 suggest Canon is the way forward and the 20D was seriously considered first time around - in the end, the cheaper Olympus lenses won out. Having used the 350D, 10D, 20D and 5D extensively, I've decided that I'm no fan of reduced format sensors purely for their pokey viewfinders and correspondingly bad MF ability. My problem though is budget - I s'pose I could stretch £1800 for a body and lens - I'm more than happy to go second hand but even then, a 5D or 1DS is useless without a lense! I simply cannot afford Canon L and instead am looking at Sigma EX as primes are impractical for me without 2 bodies to use concurrently.
I had considered the Sony R1 but it is little better at high ISO. What are the current Konica-Minolta's like? I like the idea of body mounted AS and would consider one of the new Sony DSLRs if they will perform at 3200.
I'm in the "not quite pro" catagory of shooter - I try to do it full time but also find myself doing PR, barwork and even hostessing to pay the bills. The most grueling (and for me, enjoyable!) work the camera will do - and the reason I'm stuggling with what I have, is live events, music, dance, bits of theatre/caberet etc.
Your help will be greatly appreciated!
-
Hey there - I've got an E 1 as well - and I can sympathize if you need to use ISO 3200 on a regular basis. 800 is fine and even 1600 in a pinch, but 3200 is one stop too many.
My one bit of advice if you do hang on to the E 1 - shoot raw, convert to 16 bit tiff and use a noise reduction program (like noiseware) that works in 16 bit mode. The noise cleans up much better than on 8 bit jpegs.
It doesn't sound like the kind of work you are doing suits the E1 very well - low light isn't it's strong point. The 35-100 f 2.0 lens might help with what you are doing - but it's not cheap (although cheaper than a full frame camera!) If you've got the opportunity to rent one of those to try it out.
I know a few people with the KM 7D and they are very happy with it in low light. Its got good noise handling and built in anti- shake - combine that with a good bright prime and it works well.
The rumors are that Oly will announce something in the way of a new body quite soon - probably before Photokina - so hopefully in 2 mths.
I don't have many problems getting a decent 12x16 inch print from my E 1 - but I tend to shoot at ISO 800 or lower and always in RAW.
Good luck - hope you manage to find a solution to your issues one way or another.
-
Sounds like a job for the 30D.
-
DP - I'd rather ruled the 30D out as a more expensive version of the 20D with few concrete gains - should I reconsider this?
Dave - I always shoot RAW and have been using Noise Ninja but to little affect - do you have a "before and after" using noiseware?
I've found 800 to be okay at a push, 1600 to be really quite horrible and 3200 to be pointless. There seems to be a serious reduction in dynamic range with these speeds to the point when I can only really see about 3-4 zones at 3200 and no more than 5 at 1600. In addition, the noise is so prominant it makes the pictures look poorly focussed....
RE a replacement, I checked the DPR forums and someone said he'd emailed Oly and they had confirmed a new body imminantly. I don't not believe him and accordingly have also emailed them, will post any response I get here!
-
Hey there - I've got an E 1 as well - and I can sympathize if you need to use ISO 3200 on a regular basis. 800 is fine and even 1600 in a pinch, but 3200 is one stop too many.[a href=\"index.php?act=findpost&pid=69441\"][{POST_SNAPBACK}][/a]
3200 is a mathematical fudge on most, if not all cameras that have it. If you are shooting RAW, 3200 is not worth using, as it has no advantage over 1600 pushed to 3200, because that is what it really is. The difference is, the camera's ISO 3200 probably clips a stop of RAW highlights, if it doubles the real 1600 digitization, rather than assuming a stop lower for the white or greypoints.
If you shoot RAW, usually the only benefit of 3200 is consistency in EC and FEC settings.
-
The Olympus 4/3rds system seems to be an attempt to capitalize on Olympus's strengths in producing fine lenses. Olympus has a tradition, going back to the days of 35mm film, of producing exceptionally fine lenses.
Somebody has had the idea of using this edge in lens technology to produce a smaller camera with a smaller sensor and smaller pixels that can match (and with a bit of luck) even surpass the performance of the slightly larger, heavier APS-C formats.
That's a fine idea. Unfortunately there are always 2 major components to final image quality and they are both equally important; the quality of the lens and the quality of the film or sensor. It has always been thus.
I've always been of the opinion that to get this 4/3rds system to fly requires a marriage between a sensor manufacturer that has a performance edge (such as Canon) and a lens manufacturer with a performance edge (such as Zuiko).
This hasn't happened and herein lies the problem. No matter how good the lens, if the film or sensor is not up to scratch, the performance edge of the lens is basically wasted, or at best severely compromised. The reverse is also true, but not to the same extent because the fact remains that all good lenses are capable of higher resolution than all good sensors, including Canon's. In a sense, sensors are in a position of 'catch up' with regard to lens performance. When you are in that position, it doesn't make very much sense to widen the gap by producing 'super' lenses for use with mediocre sensors (or grainy and relatively low resolution film).
My own (admittedly biased) opinion is, Zuiko should strike up an arrangement with Canon to modify their fine lenses for use with the 20D, 30D etc. That Zuiko 300/2.8 would really shine on a 20D, not to mention their fine wide angle zooms.
-
wait another month or 2 new stuff going to be anounced
-
Waiting for announcements from now until Photokina in September seems wise, unless you are in a hurry.
On one hand, Olympus has been giving strong signs of replacing the E-1 with a new top of the line model (the E-3?): E-1's are being heavily discounted lately, which looks like clearing out stocks after ending production, and Olympus has put out numerous hints of a new high end model coming soon, and at least one reputable professional E-1 user has talked of testing a prototype of a new body. The 14-35 f/2 might at last appear too.
On the other hand, other companies are likely to have new offerings at the advanced amateur to professional level. The expected 1DsMkII replacement may be out of your price range, but I hereby speculate that Canon will take the EF-S system to at least 10MP (35D?), Nikon will replace the D70s with a body using the same 10MP sensor as in the Sony A100, and the coming Pentax 10MP body might be a good price/performance option too.
P. S. I doubt you share Ray's view that only Canon can make good DSLR sensors; I for one expect that between Panasonic nMOS, Kodak FF CCD, the CMOS sensors that Kodak is apparently developing, and other options like 4/3 consortium sleeper Fuji, Olympus will be able to offer good sensors. And if you can afford the Zuiko f/2 zooms, you could often half the ISO speed compared to f/2.8 zooms with DX, EF-S, etc. Even the f/2.8-3.5 zooms (including the coming Panasonic/Leica with OIS) allow using 1/2 to 1 stop lower ISO speed than comparably priced lenses for "APS-C" formats.
-
I doubt you share Ray's view that only Canon can make good DSLR sensors
[a href=\"index.php?act=findpost&pid=70133\"][{POST_SNAPBACK}][/a]
It remains to be seen.
-
No matter how good the lens, if the film or sensor is not up to scratch, the performance edge of the lens is basically wasted, or at best severely compromised. The reverse is also true, but not to the same extent because the fact remains that all good lenses are capable of higher resolution than all good sensors, including Canon's. In a sense, sensors are in a position of 'catch up' with regard to lens performance. When you are in that position, it doesn't make very much sense to widen the gap by producing 'super' lenses for use with mediocre sensors (or grainy and relatively low resolution film).
There are a lot of photographers who strongly disagree with your "lenses are better than current sensors" theory. The 11MP Canon 1Ds is an unforgiving bastard of a camera that exposes shortcomings in all but a select few L primes, and the 16.7MP 1Ds-MkII is even more unforgiving. At this time, there's little to be gained by increasing sensor resolution; what Canon really needs to do is concentrate on designing and producing lenses that are worthy of the sensors already in production. The only area where digital sensors really need improvement is dynamic range.
-
At this time, there's little to be gained by increasing sensor resolution; what Canon really needs to do is concentrate on designing and producing lenses that are worthy of the sensors already in production.
[a href=\"index.php?act=findpost&pid=70166\"][{POST_SNAPBACK}][/a]
Jonathan,
I have a 5D and the equivalent of a 22mp full frame 35mm sensor cropped to 8mp (the 20D). Having recently taken many test shots of targets ranging from 300 metres to 20 metres using both bodies with my 100-400 IS zoom at 400mm with 1.4x extender, I'm finding that the 20D is producing consistently sharper results at f22 than the 5D at any f stop with this 560mm lens. As we both know, no lens at f22 is particularly sharp in the 35mm context, yet the 20D is able to extract greater resolution at that rather diffraction limited f stop.
These results would seem to indicate to me that there's a long way to go before sensors are unable to gain more resolution from existing lenses.
-
Jonathan,
I have a 5D and the equivalent of a 22mp full frame 35mm sensor cropped to 8mp (the 20D). Having recently taken many test shots of targets ranging from 300 metres to 20 metres using both bodies with my 100-400 IS zoom at 400mm with 1.4x extender, I'm finding that the 20D is producing consistently sharper results at f22 than the 5D at any f stop with this 560mm lens. As we both know, no lens at f22 is particularly sharp in the 35mm context, yet the 20D is able to extract greater resolution at that rather diffraction limited f stop.
You've just proved my point. Aperture has no effect on sensor resolution, but it does affect lens resolution. Your lens' resolution performance is crap; its inherent distortions are greater than the diffraction limit at any aperture larger than f/22. Your lens is the problem, NOT the sensor. Borrow a 135/2L or even a 70-200/2.8L and you'll discover a whole new world of what is possible resolution-wise with your cameras. As I've said before, a perfect lens is sharpest wide open, and good lenses (like the 135/2L and 70-200/2.8L) perform best at their wider apertures. "Best when stopped down all the way" is the hallmark of consumer-grade coke bottles or a symptom of element misalignment or some other similar problem.
-
the hallmark of consumer-grade coke bottles
[a href=\"index.php?act=findpost&pid=70173\"][{POST_SNAPBACK}][/a]
A felicitious phrase, IMHO
JC
-
the hallmark of consumer-grade coke bottles
[a href=\"index.php?act=findpost&pid=70173\"][{POST_SNAPBACK}][/a]
Well, I always insist on professional, L-series coke bottles!
Eric
-
You've just proved my point. Aperture has no effect on sensor resolution, but it does affect lens resolution. Your lens' resolution performance is crap...
[a href=\"index.php?act=findpost&pid=70173\"][{POST_SNAPBACK}][/a]
Very amusing nonsense, Jonathan. F22 is not crap. It's just diffraction limited rather than aberration limited. The best lens imaginable (the perfect lens) cannot produce more than 45 lp/mm on a camera such as the 1Ds. The 20D can produce around 60 lp/mm with an ordinary lens. Get your facts straight. An ordinary, current Canon lens can deliver 50 lp/mm at 50% MTF at f16, and probably 60 lp/mm at 40% MTF.
-
I'm finding that the 20D is producing consistently sharper results at f22 than the 5D at any f stop with this 560mm lens.
[a href=\"index.php?act=findpost&pid=70170\"][{POST_SNAPBACK}][/a]
Not surprising, given the distinctly closer pixel spacing of the 20D: 6.4 microns vs 8.2.
Are you assuming that because different sensors show different overall resolution, then the sensors are the weak link? If so, you have forgotten the fact that total resolution is determined by the product of the lens and sensor MTF, so that even if lens resolution is lower than either sensor's, the overall MTF can be detectably higher with the higher resolution sensor.
Example: suppose that at 50lp/mm, one sensor has 75% MTF, another has 50% MTF, the lens has 40% MTF. The combinations give 30% and 20% MTF respectively. You can probably see the difference between 20% and 30% MTF, but in this example, the sensors are both "better" than the lens. The visible difference does not show that the lens is outresolving the sensors.
The reverse is true too: just because one lens give better total resolution than another with the same sensor does not prove that the sensor is outresolving either lens: it doe however show that using the sharper lens is worthwhile to get the most out of the sensor.
-
To address the original poster's question, it sounds like there are two issues here.
1.) Poor performance at high ISO. A physically larger sensor is generally going to have better performance than a smaller one since they capture more photons. DPReview's review of the E-330 has a comparison to the EOS 350D, if you think that the improvements have been sufficient then there's every reason to expect that the E-1 replacement will meet your needs since it'll be better than the E-330 almost certainly.
2.) Poor auto focus and the inability to focus manually. For manual focusing you're not going to get much better than the E-1 in that form factor given that it has 100% coverage and 96% magnification. If this feature is really important, you'll have to go to a larger format. As to the auto focus, I'm sure it's improved, but that's a harder thing to make a guess at what it'll be like.
-
Are you assuming that because different sensors show different overall resolution, then the sensors are the weak link? If so, you have forgotten the fact that total resolution is determined by the product of the lens and sensor MTF, so that even if lens resolution is lower than either sensor's, the overall MTF can be detectably higher with the higher resolution sensor.
BJL,
Has it not always been the case that using a better lens shows up on either film or sensor? Even the 3mp D30 performed better with a high quality lens. No matter how good or bad the sensor, the lens with the higher MTF will tend to look better for the reasons you've stated, ie. system resolution is a product of both lens and sensor/film resolution. Increase the resolving power of either one and the final result looks sharper. Increase the resolving power of both simultaneously and the increase in system resolution will be more dramatic, and of course, the reverse is true.
I'm really surprised that Jonathan seems to think that if sensor A produces a sharper result than sensor B using a soft lens, sensor B might produce a sharper result than sensor A using a sharp lens.
The concept of either a sensor or lens outresolving the other is nebulous. What does it really mean? Are we just comparing MTF response at frequencies common to both sensor and lens? The sensor or lens with the highest MTF response at say 30 lp/mm wins and can be said to 'outresolve' the other? Why 30 lp/mm? If we choose 60 lp/mm then every good lens will outresolve the 1Ds.
I suppose I shall now have to do some tests with my finest lens, the TS-E 90/2.8, to prove my point.
-
Ray, I completely agree with your most recent post, and perhaps misunderstood the earlier one: so long as sensors and lenses are even roughly close in resolution there are gains in overall resolution to be made from both sensor and lens resolution improvements. Which his why I do not understand what you said earlier:
In a sense, sensors are in a position of 'catch up' with regard to lens performance. When you are in that position, it doesn't make very much sense to widen the gap by producing 'super' lenses for use with mediocre sensors
[a href=\"index.php?act=findpost&pid=69495\"][{POST_SNAPBACK}][/a]
This is contradicted by your newer idea that lens improvements can improve overall resolution, apparently true even with some "L" lenses and the relatively large roughly 7.2 micron pixel spacing of the 1DsMkII.
I am also mystified by the phrase "mediocre sensors" in the context of the resolution of 4/3 sensors: the 4/3 sensors have rather consistently had closer photosite spacing and so probably higher resolution (in lp/mm) than other DSLR's of the same era: three years ago the E-1's 6.8 micron while the rest were at 7.5, 8 and 9; more recently 5.3 and 5.5 while everything else in the price range is at 6 or 8, or in Canon's case, 6.4.
Surely the basic goal for a format (13.5x18mm) that is 10-20% smaller (linear measure) than its best selling competition (Canon's 15x22.5mm) is to have both lens and sensor resolution slightly higher in lp/mm, so as to achieve about the same "final image resolution", meaning "lines per picture height", or lp/mm on equal sized prints.
And as I have said before, getting somewhat higher lp/mm out of a downsized lens design (slightly smaller focal length and image circle diameter) does not require a "super lens": it follows almost automatically when one downsizes all dimensions of a larger format lens design, or when one modifies the rear elements to reduce the lens's overall magnification (like adding the "built-in focal reducer" of an inverse telephoto design, or removing the "built-in teleconvertor" of a true telephoto design, or something in between).
-
This is contradicted by your newer idea that lens improvements can improve overall resolution, apparently true even with some "L" lenses and the relatively large roughly 7.2 micron pixel spacing of the 1DsMkII.
BJL,
I don't see any contradiction. Olympus started off with a bunch of lenses that had better MTF at 60 lp/mm than most Canon lenses at 40 lp/mm, yet over all image quality (including noise levels) has always appeared to be slightly worse (in the reviews I've read, anyway) than the closest Canon equivalent. I've attributed this slightly lack-lustre performance to the inadequacy of the sensor rather than the lenses which by all accounts are exceptionally fine.
-
BJL,
I don't see any contradiction. Olympus started off with a bunch of lenses that had better MTF at 60 lp/mm than most Canon lenses at 40 lp/mm, yet over all image quality (including noise levels) has always appeared to be slightly worse (in the reviews I've read, anyway) than the closest Canon equivalent. I've attributed this slightly lack-lustre performance to the inadequacy of the sensor rather than the lenses which by all accounts are exceptionally fine.
[a href=\"index.php?act=findpost&pid=70262\"][{POST_SNAPBACK}][/a]
The contradiction is this: first you said that it is pointless for Olympus to emphasis high resolution lenses, then you acknowledge that improvments on both lens and sensor resolution will improve overall resolution so that improvnig lens resolution does have a point.
Secondly, you seem to be confusing two different measures of resolution: lp/mm and lines per picture height (or "effective print resolution" or "effective pixels worth of resolution"). I have seen tests which show that E-500 JPEG's have slightly less lines per picture height of resolution than the 20D but significantly more than in its price peers with 6MP sensors.
a. That is consistent with the smaller sensor still having somewhat higher resolution in lp/mm than the Canon 8MP sensors, and certainly that the Nikon/Sony/Fuji 6MP competition, due to the different sensor heights, which would be enough to justify the goal of lenses with higher lp/mm resolution.
b. Converting E-500 RAW files with other RAW converters increases resolution: apparently the in-camera processing is conservative about resolution in relation to other processing goals.
Thirdly, Olympus likely aims at higher pixels counts in the fairly near future, so it would be stupid to design and sell lenses now that cannot keep up with the higher resolution of new models coming in the near future.
Lenses are probably going to have longer lifetimes than DSLR bodies, so should out-resolve current DSLR bodies!
-
In response to some of the original questions...
There are a couple of things that might help with autofocus. Although, in my opinion, Olympus is a little behind in that regard.
It is possible to improve autofocus, particularly in low light, by using an Olympus flash in the hotshoe. You can set the flash so it will not fire, but the flash focus lights will still light up and help with autofocus. Of course, for your style of shooting that might be a little distracing. I have not tried it myself, but it is supposed to help. I am failry certain it works with the FL-50, and probably the FL-40.
The extra battery pack is supposed to help with autofocus too, and there are probably some inexpensive ones on ebay right now.
None of which helps with the sze of the viewfinder or the high iso capability. There may be an E-3 this fall and there may not. If there is it will probably be expensive, and the viewfinder will probably not be any larger. There are a lot of rumours going around that Olympus will release an intermediate range and a high end range, like they have with their lenses, but they are just rumours. Time will tell I suppose.
Personally I have been thinking about switching to Nikon. I like the smaller format, mostly I like doing telephoto pictures, so it is a pretty good fit for me. I am somewhat frustrated by the lack of an affordable telephoto longer than 200 mm, or a long macro lens though. I know Sigma..., but that announcement was months ago, and no release dates have been announced yet.
I just bought a 50-500, and a 1.4 convertor though. I figure if Olympus does drop the ball, Panasonic could still pick it up an run with it. If not, my investment is still pretty low.
-
first you said that it is pointless for Olympus to emphasis high resolution lenses
BJL,
My wording was a little strong, but not that strong. I was trying to convey the idea that it is not efficient to have too much disparity between the performance of the lens and the performance of the sensor and that the 5mp E-1 seems to have started out that way. It has always seemed to me a pity that Olympus was not able to take the market by storm by developing a sensor that was as state-of-art as its lenses.
Of course, in the fullness of time that might change .
-
I just bought a 50-500, and a 1.4 convertor though. [a href=\"index.php?act=findpost&pid=70296\"][{POST_SNAPBACK}][/a]
I'd be interested in seeing some test shots, Scott, at 500mm with and without extender.
-
Meh, sorry Ray, I meant 50-200. I'm thinking about a 50-500 (if it ever becomes available) but I think autofocus is going to be a struggle without a convertor.
The other thing that can help with autofocus is to make sure that you have the latest version of firmware (1.4 I think). That I know makes a difference.
-
I really need a new camera now after my Olympus was stolen this weekend. I'm not yet sure whether insurance will cover due to circumstances of the theft. The police are obviously involved.
I'll keep my fingers crossed! In the meantime, lens suggestions for a 20D? Yeah, I know, what a dull, predictable question! I've not really had time to look at the moment - too busy with police, insurance, crying etc etc. I need a 28mm equiv to around 135 ish. The 17-40 L seems ubiqitous, is their a faster Sigma or Tokina alternative? I'm thinking a couple of cheap second hand primes might sort the top end.
-
I believe there is a tamron 17-50 f2.8. There is also a canon 17-55 f2.8 IS for oodles more coin.
The 17-85 isn't bad but it is VERY dark and suffers optically (for a variety of reasons) from 17 to about 24.
The 85 f1.8 is well liked.
-
Very amusing nonsense, Jonathan. F22 is not crap. It's just diffraction limited rather than aberration limited. The best lens imaginable (the perfect lens) cannot produce more than 45 lp/mm on a camera such as the 1Ds. The 20D can produce around 60 lp/mm with an ordinary lens. Get your facts straight. An ordinary, current Canon lens can deliver 50 lp/mm at 50% MTF at f16, and probably 60 lp/mm at 40% MTF.
I never said f/22 was crap. What I said was a lens that performs best resolution-wise at f/22 is crap. A good lens (say one that delivers highest resolution at f/4 instead of f/22) can deliver 50% MTF at significantly higher LP/mm than one performing best at f/22. As to getting my facts straight, try shooting with a 135/2L prime lens at a variety of different apertures, and you'll discover just how ridiculous your premises are. My 135/2L, 100/2, and 70-200/2.8L IS (when stopped down just a bit) are the only lenses I own that are capable of matching the resolution of the sensor in my 1Ds. With all of the others (35-350L, 17-40/4L, 50/1.8, and to a much lesser extent, the 24-70/2.8L), the lens distortions are the primary resolution limiting factor. If you have not made comparison tests with a good sharp prime lens, you have no real understanding of what kind of resolution your sensor is capable of, and I suggest you quit making such unfounded and ignorant statements. I've shot over 110,000 frames with the 1Ds, the 1D-MkII, and the lenses I've mentioned. I have my facts straight, you do not. And until you shoot some frames with a 135/2L or similarly sharp prime lens, you have no basis on which to support any of your arguments or the confusions you've derived from them.
-
I'll write the 17-85 off on that since I can't live without a good wide angle!
Tamron lens mentioned hasn't yet been released (well, Warehouse Express say price TBC) but there is a sigma one for £279.99. Does anyone have an opinion on this lens?
Primes I'm considering are the 50mm 1.4, 85 1.8 and 100 or 135 2.8. The alternative to these is the Sigma 70-200 2.8 but it leaves me a fairly huge hole in my range.
Since the E1 has been discontinued, the insurance have told me to go to Jessops and have them advise me to an equivilent; do you think that they will agree a Nikon D200 is similar?
Thanks people!
-
Bribe them.
-
The MSRP of the E-1 was about the same as the D200 when it first came out, maybe a little more. You might be able to make an argument there. You could pick up an E-1 for a song right now, but it might be a good opportunity to make a break.
The only question is whether you might want full frame at some point. If the E-1 viewfinder seems small, an APS one still won't be as big as full frame. I'm not sure if Nikon intends to go full frame in the future or not. If you buy a 30D now, you could use the lenses later on a full frame Canon body.
-
My 135/2L, 100/2, and 70-200/2.8L IS (when stopped down just a bit) are the only lenses I own that are capable of matching the resolution of the sensor in my 1Ds.
And the resolution of your 1Ds is...? 45 lp/mm? 48 lp/mm? Maybe 50 lp/mm? Wow!
C'mon, Jonathan. I wasn't born yesterday. We all know that expensive prime lenses with a Photodo rating above 4 produce images with greater accutance that 'appear' sharper than images from average zooms, but resolution is a different matter. It's measured in line pairs per mm and the expensive prime lens generally improves resolution only marginally, that is by just a few lp/mm. As regards absolute resolution, it's generally the sensor that is the limiting factor, not the lens. When I tested many of my lenses with the D60 a few years ago, comparing an average zoom like the 28-135 IS with clearly better lenses such as the despised 50/1.8 and the highly regarded TS-E 90mm, I was struck by how little the absolute resolution varied. As I recall, the only lens that produced significantly less resolution than the others was my el cheapo Sigma 20/1.8 at full aperture. Even that lens when stopped down to f8 produced almost identical resolution to other lenses.
The TS-E 90 by the way would be at least as good as your 70-200/2.8 at 90mm.
-
My thinking is that the D200 costs similar to what I have on the sales reciept, I'll never blag them for a 5D. Almost certainly I'll get Canon gear because of future viewfinder upgrades (I don't need full frame for the res but for the composition and focussing!) and better noise characteristics at high ISO....oh, and the 3 month wait for delivery on Nikon at the moment - it's the Edinburgh festival in a few weeks and I need some gear!
-
You should be able to get a 30d and a lense for the price of a D200 body. I am not sure if availability is any better for Canon over Nikon though. The D200 is great to work with, though & I get great results in low light with a 50/1.4. But if you need something now, go Canon.
My thinking is that the D200 costs similar to what I have on the sales reciept, I'll never blag them for a 5D. Almost certainly I'll get Canon gear because of future viewfinder upgrades (I don't need full frame for the res but for the composition and focussing!) and better noise characteristics at high ISO....oh, and the 3 month wait for delivery on Nikon at the moment - it's the Edinburgh festival in a few weeks and I need some gear!
[a href=\"index.php?act=findpost&pid=70441\"][{POST_SNAPBACK}][/a]
-
I was trying to convey the idea that it is not efficient to have too much disparity between the performance of the lens and the performance of the sensor and that the 5mp E-1 seems to have started out that way.
[a href=\"index.php?act=findpost&pid=70307\"][{POST_SNAPBACK}][/a]
OK then, that makes more sense. But again, given that Olympus and Kodak always planned to go well beyond 5MP (Kodak talked about 10MP+ on its early 4/3 web site), having extra lens resolution beyond the needs of 5MP seems like reasonable advanced planning.
-
But again, given that Olympus and Kodak always planned to go well beyond 5MP (Kodak talked about 10MP+ on its early 4/3 web site), having extra lens resolution beyond the needs of 5MP seems like reasonable advanced planning.
[a href=\"index.php?act=findpost&pid=70570\"][{POST_SNAPBACK}][/a]
Yes, but only reasonable. There's no 'killer' application. The lenses are no good by themselves. By the time Olympus develops a sensor that can extract the most from their lenses, Canon will have moved on to sensors with even greater pixel density and possibly (hopefully) improved lenses. Now that Sinar and Rodenstock have demonstrated that it's possible to produce a lens for an even larger format than FF 35mm, with the specs of a small format Zuiko lens, the future is not looking rosy for the 4/3rds system as a viable competitor to 35mm.
-
I don't think Olympus is going to develop a sensor. I think they will probably buy one from someone else, and I think with Panasonic participating they will be able to that.
-
I don't think Olympus is going to develop a sensor. I think they will probably buy one from someone else, and I think with Panasonic participating they will be able to that.
[a href=\"index.php?act=findpost&pid=70653\"][{POST_SNAPBACK}][/a]
You might be right, but when I say 'develop', I don't mean 'manufacture', but have input in the design process and collaborate with whoever does the manufacturing. The 4/3rds sensor is not a standard size, is it?
-
And the resolution of your 1Ds is...? 45 lp/mm? 48 lp/mm? Maybe 50 lp/mm? Wow!
C'mon, Jonathan. I wasn't born yesterday. We all know that expensive prime lenses with a Photodo rating above 4 produce images with greater accutance that 'appear' sharper than images from average zooms, but resolution is a different matter. It's measured in line pairs per mm and the expensive prime lens generally improves resolution only marginally, that is by just a few lp/mm. As regards absolute resolution, it's generally the sensor that is the limiting factor, not the lens.
I wasn't born yesterday, either. The sensor resolution of the D60 is about 45 lp/mm, which in conjunction with the 1.6x crop factor, isn't going to tax the capabilities of cheap glass all that hard (since it's only using the "sweet spot" of the lens). So your D60 tests aren't particularly relevant.
OTOH, the sensor resolution of the 1Ds is just a bit over 56 lp/mm, And you're forgetting that it's not just the lp/mm, it's the MTF at that lp/mm that makes the difference. Unfortunately, the MTF data for lens performance at 56 lp/mm isn't available, but I damn well guarantee that the 135/2L is going to have a significantly higher MTF at 56 lp/mm than a 100-400L + 1.4xTC.
I strongly suggest you redo your lens comparison tests with your 5D, which has a sensor resolution of 60 lp/mm. Make sure you include some decent primes. If you're basing your perception of sensor resolution off the results you're getting with the 100-400L + 1.4X TC, I guarantee you'll be pleasantly surprised at how much better results you can get with better glass.
-
By the time Olympus develops a sensor that can extract the most from their lenses, Canon will have moved on to sensors with even greater pixel density and possibly (hopefully) improved lenses.
[a href=\"index.php?act=findpost&pid=70639\"][{POST_SNAPBACK}][/a]
Oh no, Ray's crystal ball is back, and yet again it predicts Olympus failing to adequately improve its products, while Canon succeeds in overcoming its current limitations. In case you haven't noticed, 4/3 and EF-S are currently at about the same effective resolution.
My crystal ball is currently down, awaiting input from Photokina-related announcements.
-
I damn well guarantee that the 135/2L is going to have a significantly higher MTF at 56 lp/mm than a 100-400L + 1.4xTC.
Of course it is. But I damn well guarantee that my 100-400 with or without extender will capture more detail of any target, from the same distance, than your 135/2L.
By the way, according to Photodo tests, the 135/2 is sharpest at f8, same as the 50/1.4 which has identical ratings at f8 and which is the lens used in Dpreview resolution tests.
The sensor resolution of the D60 is about 45 lp/mm
Jonathan, life in the army seems to be having an effect on your grasp of photographic facts. The sensor resolution of the D60 is similar to that of the 1Ds2. I think you are confusing total picture resolution with absolute resolution. The sensor with the higher pixel count, whatever its size, almost always delivers a higher total picture resolution. The Nikon D2X is on a par with the 1Ds because it has a similar pixel count, even though the sensor sizes are different. There are 8mp P&S cameras that produce picture resolutions equal to the latest 8mp Olympus 4/3rds systems.
If you're basing your perception of sensor resolution off the results you're getting with the 100-400L + 1.4X TC, I guarantee you'll be pleasantly surprised at how much better results you can get with better glass.
There's no argument that better glass does not produce better results. I merely thought it was noteworthy that at an aperture of f22 even, the 20D can produce a higher 'absolute' resolution (line pairs per mm) than the 5D with the same lens.
I think you've clearly missed my point, haven't you?
-
Oh no, Ray's crystal ball is back, and yet again it predicts Olympus failing to adequately improve its products, while Canon succeeds in overcoming its current limitations. [a href=\"index.php?act=findpost&pid=70717\"][{POST_SNAPBACK}][/a]
Yes. Predictions are often possible given sufficient facts. Your partisan approach (as opposed to my completely unbiased approach) would lead me to believe you actually own an Olypus E1 and are supporting it as one might support a football team. Is this correct?
-
Yes. Predictions are often possible given sufficient facts.
[a href=\"index.php?act=findpost&pid=70732\"][{POST_SNAPBACK}][/a]
Facts that were noticeably missing from your prediction!
-
Of course it is. But I damn well guarantee that my 100-400 with or without extender will capture more detail of any target, from the same distance, than your 135/2L.
Only if the 100-400 is set to a focal length > 135mm, and you crop pixels from the 135/2L frame to get the same coverage. But if you set the focal length such that you capture an equal pixel count and equal FOV with both lenses, the 135/2L will beat the 100-400L hands-down, and the higher the camera resolution, the more pronounced the difference will be.
Jonathan, life in the army seems to be having an effect on your grasp of photographic facts. The sensor resolution of the D60 is similar to that of the 1Ds2.
The lp/mm figures I cited were derived from the horizontal pixel count (4064 for the 1Ds) divided by sensor width (36mm for the 1Ds and all other "full-frame" cameras), then dividing that number by 2, since you need at least two pixels to record a line pair: (4064 / 36) / 2 = 56.44 lp/mm. But I transposed some digits when calculating the D60; the correct figure is 67.67 lp/mm.
There's no argument that better glass does not produce better results. I merely thought it was noteworthy that at an aperture of f22 even, the 20D can produce a higher 'absolute' resolution (line pairs per mm) than the 5D with the same lens.
Well obviously, because 20D sensor can record up to 77.87 lp/mm, and the 5D sensor can only record up to 60.67 lp/mm. But that's not what we're disagreeing about; the premise I take issue is that sensors are the weakest link when it comes to resolution performance. If you compare the MTF of any current DSLR sensor at its maximum spatial frequency (56.44 for the 1Ds, 60.67 for the 5D, etc.) to the MTF of any currently available lens at the same lp/mm, the sensor MTF will be higher than the lens MTF in all but a very few cases.
-
Only if the 100-400 is set to a focal length > 135mm
Of course! That's why we use zooms. The finest prime ever made is not much use if it's not the right focal length for the job.
(4064 / 36) / 2 = 56.44 lp/mm. But I transposed some digits when calculating the D60; the correct figure is 67.67 lp/mm.
Have you already forgotten those threads discussing the merits of the Foveon sensor? Only the Sigma SD9 & 10 can reach the Nyquist limit of one pixel per line width, which is why a 3.3mp Foveon sensor has almost the same resolution as a 6mp Bayer type.
If you compare the MTF of any current DSLR sensor at its maximum spatial frequency (56.44 for the 1Ds, 60.67 for the 5D, etc.) to the MTF of any currently available lens at the same lp/mm, the sensor MTF will be higher than the lens MTF in all but a very few cases.
That's not the impression I get. Norman Koren has tested his Canon 10D and produced a graph of the 'system' MTF (sensor plus lens) across all frequencies the sensor can record. Maximum resolution, as I recall, is around 54 lp/mm at 10% MTF. Any perusal of the Photodo MTF charts at 40 lp/mm will give you an indication of what the MTF response at a few lines greater resolution is likely to be. If it's mostly well above 50% at 40 lp/mm (and sometimes above 60%), it's likely to be not much below 50% at 54 lp/mm, let's say 45%.
I don't know how we get a combined MTF response of 10% (at 54 lp/mm) from a lens delivering around 45% (at 54 lp/mm) if the sensor is not the weaker link.
-
Post a link to Koren's tests; I'd like to see how he came up with his results. 10% MTF sounds suspiciously low to me.
The only difference between Bayer and Foveon sensors is that Bayer sensors' MTF starts falling off sooner than Foveon when approaching Nyquist because of the anti-aliasing filter. Both sensors can record detail at the Nyquist limit, the Bayer sensor will have an MTF of 50-70% (depending on the AA filter, this was discussed in another recent thread regarding one of the new MF backs), while the Foveon will be closer to 100%. IMO the resolution advantage of a Foveon is more like 33% than the 100% that the Foveon enthusiasts would like people to believe.
I don't know why so many people bother to use Zeiss and other third-party lenses on Canons if the sensor was the weakest link. My experience (and that of most other photographers) is that with cameras 8MP on up, the lens (even when using L glass) is the greatest degrader of resolution unless one is using primes.
-
Situgrrl
Interesting name - do you have a dog that you are training to sit down on command?
However, photography: I resisted digital for a long time, eventually deciding that I might as well give it a bash. Being a Nikon camera devotee since the F, and already having manual Nikkors (avoided autofocus even when I had an F4s), I settled for the D200.
It opened my eyes. It also hunched my back, as my time at the computer has gone from normal (whatever that is) to absurd.
I don't think you might need to shoot at high ratings as much as you think you do - I use the lowest rating available of 100, always RAW/NEF; the quality from these primes is quite remarkable at apertures much wider than I would previously have though to use on film, mainly because the altered perceptions of depth of field with smaller capture areas changes everything.
Yes, I was already in the Nikon system, but speaking from a practical point of view, if the results work for me, then that's all there is to it. There is so much internet chat about different marques, primes v. zooms etc. that confusion must reign supreme in the mind of anyone starting out in photography.
My advice is this: decide on a format; buy the most expensive body you can afford and a single lens at the limit of your budget. Make that lens the equivalent to a 35mm on full-frame 35mm film cameras (in the case of the D200 that means a 24mm lens). Experiment like hell and don't rush into buying lots of glass that will often end up gathering dust. I speak as a pro and I've done just that, to my shame, as has every other pro I've known!
Once you know what the single camera/lens can really do, which takes time, only then think about more equipment. Incidently, I would suggest buying new; used might be good too, but you can't know before you buy and then it's too late. A poor camera/lens can seriously impair your learning and, importantly, expectations.
Ciao - Rob C
-
I don't know why so many people bother to use Zeiss and other third-party lenses on Canons if the sensor was the weakest link. My experience (and that of most other photographers) is that with cameras 8MP on up, the lens (even when using L glass) is the greatest degrader of resolution unless one is using primes.
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=70768\")
I've never seen any MTF curves for a sensor as we used to get for film types, but it was well known that increasing the performance of a lens with film did not produce a proportional increase in image quality. There's always some increase in quality, however. Since it's not possible to change sensors as one used to be able to change film quality, the only possibility for improved performance is to use a higher grade lens or buy a new body.
I've found the Norman Koren page that shows the 10D test results.
[attachment=821:attachment]
Address [a href=\"http://www.normankoren.com/EOS-10D_3.html#Res_vs_35mm]here[/url]
-
Facts that were noticeably missing from your prediction!
[a href=\"index.php?act=findpost&pid=70744\"][{POST_SNAPBACK}][/a]
BJL,
If I don't provide facts it's because I can't find them. All I can do is make a few guesses and deductions from the fragments of information available. Since the 8mp Olympus 4/3rds cameras have no greater resolving power than the 8mp Canon DSLRs (and apparently slightly less, or at least some problems with moire), I think we can assume that the 4/3rds sensor is in the same situation as the the Canon Bayer type, ie. it takes approx 2.5 pixels or more for each line pair. Looking at the specs of the E-330, I deduce an absolute resolution of 72 lp/mm.
I believe dpreview use the Zuiko ED 50/2 to test the Olympus cameras. The MTF response for this lens appears to be an impressive 60-65% at 60 lp/mm. No figures available for 72 lp/mm. Would 50% be a good guesstimate?
Lens delivers 50% contrast at 72 lp/mm. Sensor delivers 10%.
-
Since the 8mp Olympus 4/3rds cameras have no greater resolving power than the 8mp Canon DSLRs
[a href=\"index.php?act=findpost&pid=70777\"][{POST_SNAPBACK}][/a]
Nonsense. I see absolutely no basis for your claim of 10% sensor resolution, working at 72lp/mm where you have no data for either lens or lens+sensor resolution, for which you substitute guesses.
Can you explain how you get from your guess of 50lp/mm for the lens to 10lp/mm for the sensor?
Your whole argument seems nicely summarized by your first line
... I don't provide facts ... because I can't find them
The DPReview tests give some rough resolution measures, which can be converted from "lines per picture height" back to line pairs per mm as follows: take the l/ph value, divide by the frame height listed in the review, and divide by two to get back from "lines" to line pairs.
The result is clear and unsurprising: the E-300, E-500 and E-330 all outresolve any of Canon's sensors, falling just slightly short of the D2X; all more or less as you would guess from pixel spacing. For example, 64-69lp/mm for the E-500, 55-63 lp/mm for the 20D.
Likewise, the E-1 outresolved any other DSLR on the market at that time.
Ironically, Canon's DSLR sensors now seem to trail those of Sony, Kodak and Panasonic for lp/mm resolution, though I expect a partial comeback soon with something like a 10MP+ EF-S format model. So how do you come to your conclusion that Canon (rather than Sony even) is the place for Olympus to turn for higher DSLR sensor resolution, as in these words:
... to get this 4/3rds system to fly requires a marriage between a sensor manufacturer that has a performance edge (such as Canon) and a lens manufacturer with a performance edge (such as Zuiko).
I am happy to accept your verdict on lenses though, as someone what has repeatedly chosen SLR systems (Canon film then Olympus digital) largely on the basis of the quality of the lens offerings in my price range.
-
Nonsense. I see absolutely no basis for your claim of 10% sensor resolution, working at 72lp/mm where you have no data for either lens or lens+sensor resolution, for which you substitute guesses.
The figure of maximum sensor resolution of 72 lp/mm was derived by dividing the total number of pixels on the E-330's sensor (long dimension) by 2.5. But I see I've been over-generous because the true resolution is 64-69 lp/mm. Of course I now realise I should not have used the term 'resolving power' when I wrote, 'The Olympus 4/3rds cameras have no greater resolving power than the 8mp Canon DSLRs.' I meant of course, total picture resolving power. I'm surprised you did not gather this from the context .
Can you explain how you get from your guess of 50lp/mm for the lens to 10lp/mm for the sensor?
BJL,
Now you're confusing lp/mm with contrast percentages. The MTF chart on the Olympus site shows the Zuiko 50/2 as having an MTF response ranging between 65% and 60% at 60 lp/mm. I therefore think it a reasonable guess that MTF at 72 lp/mm would be around 50%. But now we know the true resolution is no more than 69 lp/mm, I would think 50% would be a conservative estimate.
10% MTF (not 10 lp/mm) is the contrast at the cut-off point of 69 lp/mm. Lens delivers 69 lp/mm at 50% MTF, at least. Sensor delivers 69 lp/mm at 10% MTF, at most. Ergo, the sensor is very much the weak link.
Ironically, Canon's DSLR sensors now seem to trail those of Sony, Kodak and Panasonic for lp/mm resolution
Maybe. I'm not terribly concerned about a few lp/mm here or there. For me, low noise at high ISO is more significant. If the 20D was better than the D60 only in respect of having 8mp instead of 6mp, I wouldn't have bought it. If the 20D was only a 6mp camera but with the stirling noise performance it has at ISO 1600, I would still have bought it.
-
10% MTF (not 10 lp/mm) is the contrast at the cut-off point of 69 lp/mm. Lens delivers 69 lp/mm at 50% MTF, at least. Sensor delivers 69 lp/mm at 10% MTF, at most.
[a href=\"index.php?act=findpost&pid=70936\"][{POST_SNAPBACK}][/a]
Firstly, I see no source for this 10% figure (yes, %, not lp/mm!).
Secondly, I do not see why I should care about MTF as this particular number of 69lp/mm, just below the Nyquist-like limit from pixel size. At the very useful 60lp/mm level, there is no sign of such extremely low sensor MTF. Let alone the possibly more relevant 20lp/mm.
-
Firstly, I see no source for this 10% figure (yes, %, not lp/mm!).
Again I'm making some deductions which I think are reasonable in light of the few facts on such matters that are available. Refer to the Norman Koren tests of the 10D here (http://www.normankoren.com/EOS-10D_3.html#Res_vs_35mm)
First, 10% seems to be a general (practical) cut-off point for visible resolution. Anything less seems to be irrelevant except for the most demanding scientific purposes. The Rayleigh's limit for lens resolution at a diffraction limited aperture is a case in point. Secondly, the 4/3rds sensors are Bayer type like the Canon sensors, aren't they? We might therefore expect the combined lens/sensor MTF curve to be very similar to that of the 10D's sensor, as illustrated on the NK site. One would expect only the lp/mm figures to change. 10D's resolution limit is 55 lp/mm at 10% MTF; E-330's resolution limit is 69 lp/mm at 10%.
Secondly, I do not see why I should care about MTF as this particular number of 69lp/mm, just below the Nyquist-like limit from pixel size. At the very useful 60lp/mm level, there is no sign of such extremely low sensor MTF. Let alone the possibly more relevant 20lp/mm.
If you make prints no larger than A4, are not in the habit of cropping your images and are not short-sighted, there's probably not much reason to care about sensor MTF at 69 lp/mm. On the other hand, given a certain amount of altruism, you might consider the concerns of other viewers who happen to have keener eyesight than yours . Neither the E-330 nor 20D can meet that rigorous standard of 6 lp/mm on an A4 print.
If you have a large printer such as the 24" wide Epson 7600, as I have, then maximum print size increases by a factor of 3 in each dimension. I would therefore think (at least it's a logical proposition if not supported by empirical evidence) that the role played by 10 lp/mm on an A4 size print (regarding SQF), becomes 30 lp/mm and the role played by 20 lp/mm becomes 60 lp/mm on a 24x36" print, for close inspection purposes.
Of course, if you bring in the factor that large prints have a 'normal' viewing distance which is also greater, then it's an entirely different ball game. It can be impossible to tell the difference between a 35mm blow-up and an MF blow-up from across the room. But close inspection reveals all.
-
Wanky ain't you dead yet? The Army is dangerous man. W. will have your head on a platter.
-
... First, 10% seems to be a general (practical) cut-off point for visible resolution. Anything less seems to be irrelevant except for the most demanding scientific purposes. The Rayleigh's limit for lens resolution at a diffraction limited aperture is a case in point. Secondly, the 4/3rds sensors are Bayer type like the Canon sensors, aren't they? We might therefore expect the combined lens/sensor MTF curve to be very similar to that of the 10D's sensor, as illustrated on the NK site.
[a href=\"index.php?act=findpost&pid=70978\"][{POST_SNAPBACK}][/a]
So the bottom line is that the sensor in the E-500 and E-300 has slightly lower pixel count/resolution than that in the 20D and 350D, so that by choosing a resolution level that is a bit beyond the E-500 sensor but within the reach of the 20D sensor, the 20D is likely to give higher MTF. And from that you deduce that the FourThirds system is hopeless behind in sensor technology, solvable only by changing to Canon's sensor (and perhaps to Canon's format choices).
That is certainly a lot to read into the differences in the very small differences in the resolution numbers I have seen! DPReview gives the 20D and 350D 1850 horizontal LPH vs 1800 or 2.7% less for the E-500, while both are measured at 1650 vertical LPH. The piel counts also differ by equally small amounts: 5% more vertical for the E-500/E-300 than the 350D, 6% more horizontally for the 350D. Noting that DPReview's LPH numbers are always multiples of 50, reflecting the precision of the measurements, you base you entire condemnation of the FourThirds sensors on a single step (of less than 3%) in a single resolution measurement!
-
Wanky ain't you dead yet?
The rumors of my demise are greatly exaggerated.
-
So the bottom line is that the sensor in the E-500 and E-300 has slightly lower pixel count/resolution than that in the 20D and 350D, so that by choosing a resolution level that is a bit beyond the E-500 sensor but within the reach of the 20D sensor, the 20D is likely to give higher MTF. And from that you deduce that the FourThirds system is hopeless behind in sensor technology, solvable only by changing to Canon's sensor (and perhaps to Canon's format choices).
BJL,
You're putting words in my mouth. I don't see the 4/3rds system as hopeless. It has the advantages of being slightly cheaper and lighter than 'the next size up' Canon. It has almost the same resolution, as close as matters for non-pixel-peepers except perhaps for a tendency to produce moire effects, but as a total package it is lacking in certain desirable features such as Image Stabilisation and low noise at high ISO's. In other words, performance is hampered by the precise factors that give the camera its only advantages of slightly lower cost and weight.
As I wrote many posts ago, for the 4/3rds system to really fly, it needs a ground-breaking sensor design to match the quality of its lenses, otherwise it will always trail its slightly bigger brother in total image quality. It might occasionally catch up resolution-wise, might even marginally exceed its Canon competitor briefly by being first to cram more tiny pixels on its tiny sensor, but ultimately, unless it can match Canon's low noise and over all image quality, it will remain in a niche market.
But there's nothing wrong with that. Niche markets should be catered for. If money was no problem, I'd probably buy an Olympus E-500, 300/2.8 lens plus 1.4x extender, provided I could first see some comparison shots between that combination and an equivalent Canon combination of similar weight. In other words, without cost being a considerations, does (say) 4Kg of Olympus gear get me a sharper (and better quality image noise-wise) of that tiny bird 20 metres away, than 4Kg of appropriately chosen Canon gear, without being too concerned about a few grams either way?
That's probably another question that cannot be answered by any readily available facts, leaving me to just speculate and make reasonable deductions, as I've been doing all along. The Olympus 4/3rds format was supposed to be 'taking on' the 35mm format. I don't see much in the way of image comparisons on the net.
-
The Army is dangerous man. W. will have your head on a platter.
[a href=\"index.php?act=findpost&pid=70981\"][{POST_SNAPBACK}][/a]
Doug,
You really should get back on your medication.
-
BJL,
You're putting words in my mouth. I don't see the 4/3rds system as hopeless.
[a href=\"index.php?act=findpost&pid=71078\"][{POST_SNAPBACK}][/a]
My comment was only about your criticisms of FourThirds sensor technology, not FourThirds as a whole, and on that point you again seem to insist that FourThirds needs a major improvement, or else will become a niche product. (How do you define "niche product"? Less than 3% share of the DSLR market, as is the case with 35mm format?)
As to high ISO, you have heard my answer many times: how high an ISO speed is needed is related to the minimum usable f-stop, and smaller formats with their shorter focal lengths can typically use lower f-stops, and thus lower ISO speeds to get the same shutter speed in given lighting conditions. Comparisons at equal ISO are rarely relevant, because they assume that the larger format can use a lens that is likely to be bigger and heavier (longer focal length, larger aperture diameter, larger front elements), if such a lens exists at all. Such comparisons also either ignore the lower DOF in the larger format at equal f-stop, or dubiously assume that the extremely shallow DOF of a fast lens used wide open is never a disadvantage to image quality.
For all but the users of the extremes of large aperture lenses (f/1.4 or f/1.2 primes etc.), Four Thirds can match the noise levels of a larger format like EF-S at a given high shutter speed/low light level combination by using lenses of the same FOV and same aperture diameter. In comparison of FourThirds to EF-S format, this means focal length about 10-20% less and aperture ratio about 1/2 stop less. That allows using ISO speed about 1/2 stop lower, which with comparable sensor technologies should give about the same noise levels.
Since FourThirds is capable of reaching f/1.4 in primes and already has one constant f/2 zoom, EF-S only gains a true high speed advantage over the FourThirds lens system at apertures below about f/1.7 in primes and f/2.4 in zooms. Or if most FourThirds lenses stay at f/2 and smaller, as seems likely, the threshold is f/2.4 across the board. (And about f/2.5 compared to DX format.)
As far as I know, since DSLR's arrived, the four main SLR makers other than Olympus have introduced only one new lens faster than f/2.8, the Nikon 200/2, so it seems to me that most of the APS-C format DSLR world is working within that f/2.4 limit. (The 85/1.2 II is an update of a rather old design, not a new design.)
-
Four Thirds can match the noise levels of a larger format like EF-S at a given high shutter speed/low light level combination by using lenses of the same FOV and same aperture diameter. In comparison of FourThirds to EF-S format, this means focal length about 10-20% less and aperture ratio about 1/2 stop less. That allows using ISO speed about 1/2 stop lower, which with comparable sensor technologies should give about the same noise levels.
You're exaggerating, BJL. The 4/3rds sensor is 18x13.5 as opposed to the 20D's 22.5x15. Comparing heights, the 4/3rds system has less than 1/4 stop DoF advantage. Comparing widths it has a full 1/4 stop DoF advantage. That puts it well behind the 20D noise levels which, according to dpreview are equivalent at ISO 800 to the 20D at ISO 2000. I'm comparing here both the E-500 and E-300 with the 20D. Image quality and noise at ISO 800 are on a par with the image quality and noise levels of the 20D at ISO 1600, which is actually ISO 2000. That's a 1 and a 1/4 stop difference. Regarding noise at the same Dof and at high ISO's the 20D has a one stop advantage. Add IS to that and in certain circumstances that advantage becomes 3 stops.
-
BJL,
Actually, on second thoughts, considering the 'DoF equivalency' varies, depending on which dimension of the sensors is compared, height, width or diagonal, I'll be generous and give you the 1/2 stop advantage of the 4/3rds system with respect to aperture for a given DoF. That leaves the 350D with only a 2/3rds stop advantage as a result of lower noise at high ISO's and less than that at lower ISOs.
I didn't realise the E-500 was so close. Well done Olympus! . However, pixel sharpness and detail in the E-500 images fall off at high ISOs, so it's really a bit difficult to put the E-500's performance on quite the same level as that of the 350D.
Perhaps a more significant disadvantage of the 4/3rds system is a lack of IS. The image stabilisation system in Canon lenses was the major reason for my switching from Minolta to Canon several years ago. Whilst IS is not of great use in all shots, for example those that require a fast shutter speed because the subject is moving, it is nevertheless a very desirable feature of ground-breaking proportions.
Olympus really needs to introduce an anti-shake sensor to remain competitive.