[Matsu]

But the trajectories of the two companies are very different. Panasonic is an electronics and video company adding some photographic products to complement that focus. Olympus is a photographic and optics company. IMHO, a step-up product is more important in the latter case, because they sell still image cameras that compete with both compacts and larger sensor DSLRs. Also, they not only compete with these devices from other maker's systems, they compete within their own platform (with Panasonic).

I think the sensor size is fine to OK for stills, and good to excellent for video. It trounces compact camera sensors in both cases, and has some unique advantages for video even over larger imagers.

I like both. I would get one for an affordable take anywhere compact, or as a dedicated video platform, but not as a primary still image system. I don't advocate killing it in any case, just adding a larger sensor option to the platform.

[Dorian Gray]

Any particular review of the Sigma 17-50 mm f/2.8, Matsu?

On Olympus, I don't think they misjudged sensor size, unless you argue that their choice of size roped them into using Kodak and Panasonic sensors. That's been a problem, certainly. Sony and Canon have been significantly ahead for years, and of late they have if anything increased their lead over Panasonic and Samsung (Kodak having given up a while ago).

If you make a camera with a CMOS sensor today it must have a sensor by Sony or Canon, or it will be uncompetitive.

If Olympus had world-class Four Thirds sensors they'd have 20-megapixel cameras with vastly improved signal-to-noise characteristics across the ISO range. That would change a lot of things, and the better resultant sales might even have allowed Olympus to invest more in useful lenses. You would think the Four Thirds and Micro Four Thirds systems would be chock-full of good primes to maximise the size advantage (like Pentax), but instead they have a handful of humdrum primes and dozens of mediocre zooms that nobody uses as a reason to invest in the system. Those "consumer" zooms are needed, and Canon and Nikon have loads of them too. But Canon and Nikon tempt users into their systems with brilliant lenses, many of them primes.

The telecentricity requirements were definitely too conservative for general photography. In practice, SLR mirrors took care of angles of incidence all by themselves.

Not sure where Olympus go from here.

[Dorian Gray]

Any last guesses on the price of the 50 mm f/1.8 AF-S before Nikon release this thing? I'm going to go with $349, as expensive as that sounds. Probably with a hybrid aspherical element, but I'm still hoping for a PGM element...

[Matsu]

Lenstip Review

[PB PM]

Originally Posted by Dorian Gray Any particular review of the Sigma 17-50 mm f/2.8, Matsu? On Olympus, I don't think they misjudged sensor size, unless you argue that their choice of size roped them into using Kodak and Panasonic sensors. That's been a problem, certainly. Sony and Canon have been significantly ahead for years, and of late they have if anything increased their lead over Panasonic and Samsung (Kodak having given up a while ago).

That's partly true, expect for Canon. The only camera makers that gained market share in 2010 were, Sony, Nikon and Panasonic. Canon held it's 2009 market share, and all others lost ground. Kodak is a strange company in the post film era, but they sell sensors to the likes of Leica, so volume may not be their target.

The real problem that Olympus faces is a lack of doing anything different than anyone else. Sure they started 4/3s/m4/3s, but they haven't done anything new with it. They make decent cameras, but they don't look any different from what other, better known camera makers are selling. Ask the average camera buyer about the Olympus brand, and they wont know what brand your talking about. If anything that is the real problem they face.

As for the 50mm F1.8G, the official price is $219.95 USD. Super cheap, but it wont be available till mid-early June. I'll throw in a pre-order for sure.

[Matsu]

The price is very encouraging if the lens performs as described. Totally worth a 35-40% increase over the D model. If they can apply the same magic, for a similar increase, to the remaining wide AF/AF-D primes, they're going to have a killer line-up. If I were Nikon, a 35mm AF-S f/2 would be my next lens.

[Escher]

For what it's worth, I just ordered a new Nikon D5100 and Olympus XZ-1. Should be a vast improvement over my D40x and dead Canon and Pentax P&S cameras.

[Matsu]

You have to let us know how the video and live view shooting work for you, and also how you rate the viewfinder.

[Dorian Gray]

Originally Posted by PB PM That's partly true, expect for Canon. The only camera makers that gained market share in 2010 were, Sony, Nikon and Panasonic. Canon held it's 2009 market share, and all others lost ground.

Was referring to sensors only. Canon and Sony are way ahead of the rest.

Originally Posted by PB PM As for the 50mm F1.8G, the official price is $219.95 USD.

Incredible price! Far less than I expected. I wonder if it will be any good stopped down? I guess it will have more lateral chromatic aberration than the all-spherical f/1.4G.

[PB PM]

Originally Posted by Matsu You have to let us know how the video and live view shooting work for you, and also how you rate the viewfinder.

The viewfinder of the D5100 is the same as the one found on the D3000, D5000, D3100. In other words, not that great.

Dorian: Wasn't sure if you were talking about market share or sensors. With sensors it is tough to tell. Canon has been rather dormant on that front since the release of the 18MP sensor two years ago, although one could argue it is still the best APS-C sensor on the market today. As for Sony, they make good sensors, but I have noticed that third party variants of there sensors always perform better than the ones found in their own cameras. That could be due to software/firmware in cameras, or that companies like Nikon ask Sony to make tweaks to the sensors they order. Also Nikon may not make sensors in their own fabs, since they don't have any, but they say that are designing their own sensors, such as those found in the D3s, D700, D3100, and I kind of believe it. I say that because the size/crop factor of the D3100's 14MP is slightly different than that of the Sony 14MP sensors.

[Matsu]

This all brings up an interesting question about just how much sensor is enough? With the right lighting, I can get some incredible detail out of 12MP, enough that I would have to soften a face a bit in tight head and shoulders portrait, but that doesn't mean I don't want more under other conditions, like groups and fast paced scenarios where it makes a bit more sense to center the subject loosely and then crop for the geometry/composition you want.

BTW, I've been trying to find any print copy of "The Decisive Moment" for a gift, no luck. I appreciate any leads...

[PB PM]

That is something I've wondered about myself. In all honesty I don't notice much difference between higher resolution images from the Canon 18MP or 21MP sensors and Nikon's recent 12MP sensors. I guess the same could be said of the new 14MP and 16MP sensors Nikon is using. Sure if you are cropping you'll see a difference, but actual image quality isn't much different.

Also the higher resolution sensors bring other issues, like huge file sizes. I remember when I had a sample T2i that RAW files were 25+ MB each, you could fill up a lot of hard drive space and fast! Compared that to the 12-14MB files i get from 14bit images from the D700. I could get almost twice as many images on a drive!

[Chinney]

Good points. Clearly MPs alone are not going to make a great picture even from a strictly technical perspective of how good an individual sensor is* (putting aside, it goes without saying, all the other elements that make a nice photograph). But it is a "number" that can be compared so all manufacturers refer to it prominently.

*Indeed, is it not the case that for digital compacts, fitting too many MPs on a small sensor can sometimes actually lead to poorer image quality? Correct me if I am wrong on this point.

[Dorian Gray]

Originally Posted by Chinney Indeed, is it not the case that for digital compacts, fitting too many MPs on a small sensor can sometimes actually lead to poorer image quality? Correct me if I am wrong on this point.

You're perfectly right, though your statement could be expanded to say that "too many pixels" is a number that increases with process shrinks. Too many for last year isn't necessarily too many for today.

In other words, it's not true that a 3-megapixel sensor made with 2011 technology would offer dramatic signal-to-noise improvements over today's 12-megapixel sensors, though photographers with a loose grip on the technology sometimes clamour for such a thing.

At the end of the day, a certain amount of light falls on a sensor, and it doesn't really matter how finely that fixed amount is divided up. Individual pixels will be noisier if smaller, but they'll also be accordingly smaller in the final viewed image.

[Dorian Gray]

The new 50 mm f/1.8 Nikkor will probably compare favourably with anything similar on the market, particularly at its aggressive price point. Very fast lenses like the 50 mm f/1.4 models from Nikon, Canon and Zeiss are optically impressive when well stopped down, but they deliver technically poor images at f/1.4 and even f/2. The Sigma 50 mm f/1.4 has more contrast, but that applies mostly to the centre of the frame — and this lens is not as good as the others at f/8 (lateral chromatic aberration, field curvature, etc.).

Perhaps the best non-Leica 50 mm lens on the market at f/2 is the Zeiss ZF/ZF.2/ZE Makro-Planar 50 mm f/2. See its MTF charts here. (Page 3 for performance at infinity.) This lens has particularly heavy vignetting at f/2, and an unattractive outward kick in its field curvature towards the extreme corners, but it's very well corrected for the common aberrations facing fast fifties, including sagittal coma flare, despite being an entirely spherical design.

I therefore thought it might be interesting to compare Nikon's published MTF chart with that of the Zeiss. The comparison looks like this:



Notes:

1. Black curves are for the Zeiss at f/2, red and blue for the Nikkor at f/1.8 (so one-third of a stop faster).
2. Red is for 10 lp/mm, blue for 30 lp/mm. The Zeiss curves are for 10, 20 and 40 lp/mm, so only the 10 lp/mm curves are directly comparable. Visually extrapolate 30 lp/mm for the Zeiss.
3. Zeiss curves are measured on real samples. Nikon curves are calculated with optical design software, making the assumptions of perfect manufacturing quality and no diffraction (the latter is practically irrelevant at large apertures like f/1.8).

These things considered, the Nikkor is undeniably impressive at f/1.8. The Zeiss is highly corrected for lateral chromatic aberration so its stopped-down performance is extraordinarily high, and the Nikkor may not match it there. But at large apertures it looks pretty good in these charts. The 10 lpm/mm curves for the Zeiss and Nikkor are almost superimposed!

The Zeiss is $1283 at B & H Photo.

The Leica Summilux-M ASPH is better than the Zeiss above in almost every way, but it's $3695 at B & H. Worse, it's perpetually out of stock and going for $5000 on eBay because Leica can't make them quickly enough to satisfy demand. (It's a very difficult lens to manufacture.)



Edit: there are ten full-size or very large sample images from the new Nikkor on Flickr here. They don't have useful EXIF data, but some of them are also here, where the used f-stop is specified.

[PB PM]

Hmm Nikon Canada is playing it's usual game of over charging with the new 50mm f1.8G, MSRP here in $249.95. Street price seems to be $232 already though, so not too bad. I placed my pre-order today, and will do some testing of the new lens vs the old once it arrives.

If you go to the Nikon website it gives you the EXIF data on those samples. Most seem to be shot at F2.5, but even the F1.8 samples look good. Bright areas (lights) in bokeh look much rounder than those from the 50mm F1.8D.

[PB PM]

More big photography new today, PMA 2011 (now called CliQ) has been cancaled for the year, and will happen along side CES next January. Not sure what this means, maybe all the big players (Nikon, Canon, Sony) pulled out due to the events in Japan?

Maybe we wont be seeing any major new camera releases this year at all? Rumors have been saying that we wont be seeing the D700 replacement before this fall, and that the D4 might come out around CES/the new date for CliQ. I wouldn't be surprised if the 5D MkIII is pushed back similarly.

[Dorian Gray]

Yeah, I think my timing on the D700 was pretty decent after all. It now costs about 350 euros more than I paid for it, and a successor isn't on the cards until after the summer.

Looking forward to seeing a few shots with the 50 mm f/1.8G AF-S!

[PB PM]

I feel the same way, about timing with my D700 purchase. Rumors in Feburary, when I bought it, were point to a late spring or August release. I knew I made the right choice at the time and the situation now makes it even clearer. I think that it is currently impossible to buy a FX body in Canda unless you are a NPS member.

[Matsu]

You did very well indeed. Although you can easily buy an FX body in Toronto. I'm left trying to figure out if I can get what I need out of DX, for not wanting to buy a full frame at this point, but that means faster glass, which is also expensive, though portable. With a flashgun I can do almost everything I want, so I've been thinking of making a slight lateral move to a smaller cheaper Nikon body with a slightly updated DX, either of the 5100 or 7000, with the full intention of making that a back-up/second camera once new full frame cameras arrive.*

When they do, we can all determine whether it makes any sense switch to another brand, but it's not either here or there. Nikon's got the best glass from 14 to 70mm, in the shape of the 14-24 and 24-70. And all of the 24, 35, and 85 f/1.4, are up to or above a competitive standard. It's got a solid CLS system - which I'm no master of... yet.

[Moment of selfishness]Fucking tsunami.[/moment of selfishness]

[PB PM]

Yeah, I just did some looking, and it appears that the Toronto shops are the only ones with D700s in stock. Almost everyone else is listing as out of stock or back-ordered. Of course the smaller stores seem to get dibs after Henry's, which is likely the reason for that. Even then, I'm not sure that they are allowed to sell an FX body to you if you aren't an NPS member, due to the limited supplies coming out of Japan. My guess is that at the very least NPS members get first dibs on any stock that does come in.

Not a bad idea to stick with DX gear at this point, considering that FX gear is selling at or above* MSRP in many shops. The D7000 would make a good backup to a current FX or even next gen FX body due to it's overall performance. At least as long as the poor buffer performance doesn't get in your way. The D5100 seems like an odd ball, targeted at video shooters more than anything else. Not being able to auto focus with non AF-S lenses is kind of a killer, which is why I'm going to trade my D3100 in towards a D7000 at some point.


*Which is downright ridiculous

[Matsu]

In other news the GH3 from panasonic seems to have a significantly improved 16MP sensor. ISO 6400 JPEGS looks the equal or better of the APSC competition on dpreviews test shot widget. RAWs not available yet. What's weird is how bad some of the Sony/Nikon 16MP sensors look in that same widget. They test significantly better elswhere on the web. Weird. Not sure how the 4/3rds sensor will fare for dynamic range though. Makes me really interested to see what the next generation of full frame cameras will do...

The problem for m43 is still the lack of FAST small lenses. This system nees at least f/1.4 primes, maybe even f/1.

[Dorian Gray]

I'll wait for raw files, but the G3 (not GH3, which will probably arrive eventually but isn't here yet) looks decent but not revolutionary in the JPEG comparisons. I guess it will have noticeably more noise at high ISO than the Canon or Sony sensors, but fall far behind the Sony at low ISO.

The 16-megapixel Sony sensor's quality lies primarily in its phenomenally low shadow noise at low ISO. Its high-ISO performance isn't really any better than the 18-megapixel Canon that's been out for ages.

Here are three videos which vividly show the superiority of the Sony at low ISO: http://testcams.com/blog/2011/05/03/...dynamic-range/

If you don't have much time the second video is the one to watch. This guy who made the videos is knowledgeable and articulate: a rare thing when it comes to subjects like dynamic range!

[Matsu]

Yeah, it's G3, sorry I overspelt it in the post above. It looks interesting, but would a cheapish 17mm f/1.4 be too much to ask for in 4/3rd guise? Long zooms could be a good thing on this camera as well. Sometimes you need relatively more depth of field with long telephotos, maybe fastish zooms and primes between 100-400 could develop into something of a niche for 4/3.

[PB PM]

I think the issue is that Panasonic knows that it cannot compete in the pro area, so they just aren't trying. Consumers are their target, and that is why they have the lenses that they do.

Samsung is going to be releasing some fast lenses for their NX cameras, but they are APS-C sized, and are behind everyone in terms of ISO performance.

[Matsu]

I think for teles and long zooms they could compete quite well. Say for example you needed a 200-400 f/4. A 70-200 f/2.8 is substantially smaller and woudl more range and the same reach on 4/3rds. To get back all the light gathering, it would have to be f/2, but 2.8 is plenty bright and gives enough depth of field control at those distances. They could make it really small by further reducing it to 100-200. It's a trade off, but one that might well be worth it if you have to carry more than one large lens/camera combination.

[Matsu]

Hey guys, I'm hitting a camera show this weekend, and I may dump my D300 in favour of a D7000 or D5100 and a stabilized 17-50 f/2.8. Unless someone is selling a D700 at a really good used price (not likely)

Here's a question. From Dorian's videos it seems quite possible to shoot the 16MP sensor for highlight tones and pull the shadows back up pretty cleanly. Is it possible to work the other way? Expose the subject (always shooting RAW) and pull the highlights down a bit? I know at least one photographer who swears by shooting the D700 (a different beast of course) this way. Though somewhat against conventional (digital) wisdom, he says there's two stops of highlight info in the RAW, and provided the subject is properly exposed, you can just pull the highlights back to good effect. His rationale is that the shadows stay a lot cleaner when you don't have to pull them up as much.

Maybe it depends on sensors? Some you shoot for highlights, others for shadows? Thoughts...

[PB PM]

In 14bit mode the D700 does have a lot of room to work with, at least at ISO200-800. Personally I wouldn't shoot like that guy very often, but I do from time to time. I try to shoot somewhere in the middle, although I feel more comfortable pulling in highlights than shadows. I haven't pushed the D700 too much in that area, maybe because I'm so used to shooting with the D300, which wasn't great in that regard.

[PB PM]

So this dynamic range issue hit a note with me. I did a test of the D3100 vs the D700 and wrote up a quick blog post.

I was surprised by my findings, first of all the D3100 actually seems to maintain more shadow detail than the D700. That came as a surprise since the D3100 only shoots 12bit RAW, vs the 14bit RAW files that I shoot with the D700. Of course, there is a flip side, the D700 seems to retain more highlight detail. The solution would be to underexpose the D3100 and overexpose with the D700, and you'd end up with similar dynamic range. Of course there is another issue here, the D700 has superior noise performance, and thus maintains more detail in the image itself, so it is still the winner in the end.

[Dorian Gray]

When someone says "there's two stops of highlight info in the RAW", they're necessarily comparing the raw file with the JPEG. Since the D700 creates contrasty JPEGs by default, and since like other "professional" Nikons it meters to preserve highlights — often leading to dark-looking photos — it's possible there are sometimes two stops of highlight information to be gained by appropriate development of the raw file, compared to the in-camera JPEG (and histogram).

This doesn't change the fact that all digital cameras have an abrupt point at which the pixels saturate, and that all decent cameras like the D700 have RGB pixels that saturate at approximately the same exposure, when shooting a neutral subject in neutral light. Go over that point, and any recovery is mere extrapolation from surrounding, non-saturated pixels.

When shooting non-grey subjects such as blue skies, red flowers and human faces, one channel usually clips (saturates) before the others, which sometimes allows the raw developer to make reasonable estimates of the clipped channel from the non-clipped channel(s). The results can be good.

Point-and-shoot cameras have small, noisy sensors, and their owners are more concerned with getting bright, clear pictures than protecting every last highlight detail. So these cameras tend to meter hotter, resulting in less highlight headroom. This isn't the fault of the sensor, but rather the meter.

Low-cost DSLRs also tend to give more exposure than a D700, D300, D3, etc. This may contribute to the idea that a D700 has more highlight recovery available than a D5100, for example.

Ultimately though, the real difference in sensors isn't how they blow highlights, which is really a metering issue, but how good their signal-to-noise ratio is in the shadows. This determines the dynamic range of the sensor. It's in this area that the new Sony sensors are absolutely superior to previous sensors, but only at low ISO. By ISO 400 and certainly ISO 800, Canon cameras are capable of extracting most of their native sensor performance. It's only at ISO 100 and 200 that they throw away large chunks of their sensor's dynamic range by injecting copious amounts of read noise into the signal path.

Cameras like the D700 and D3S also suffer from high read noise at low ISO, but not quite as badly as the Canons, which in addition to being noisy in the shadows, also have non-random pattern noise. I'm sure the next full-frame Canon and Nikon sensors will target read-noise at low ISO as the main development goal. The D3X is the only full-frame camera with good read-noise at low ISO, and it only does that with multi-read tricks (and therefore a very low frame rate in 14-bit mode).

Having owned a D300S and D700, I'd be reluctant to buy a D7000 or D5100. These cameras have great sensors, but they don't have the build/materials quality and precision of the D300S/D700.