Great new LFI article...

[Mokkacream]

Advertisement (gone after registration)

How does one here in the U.S. obtain a copy of this magazine? Is it subscription only? If so, will they deliver here in the U.S.?

 

 

LFI International

[modschiedler]

The photography in LFI is terrible. The test shots that appear in lens articles are terribly boring. What they need are better photographers showcasing what the lenses can do.

 

Well it somehow depends upon what you want to demonstrate. If you want to do a comparison with as many aspects as possible held equal there’s often no way out opting for static and, yes, rather boring subjects. And pictures like these often just do the trick. They show what they are meant to show. Nothing else. If you are interested in what good photographers can do with their equipment, look at the portfolios.

[TimF]

The photography in LFI is terrible. The test shots that appear in lens articles are terribly boring. What they need are better photographers showcasing what the lenses can do.

But test pictures aren't supposed to be art, are they? They're there to demonstrate certain things, such as in the case of the first group how much detail is rendered clearly (compare the lettering on the white sign and the brickwork). The second group, which Ian complains about aren't meant to be great portraits - the purpose is to compare colour rendition between the three cameras, nothing more.

[plasticman]

Clarifies what Hans and others were trying to find about whether or not M9 outputs untouched files.

As I was expecting, Leica stays true with its core values and promises, giving us a tool that is most likely the most precise reproduction with highest fidelity of a certain moment...A great article and a great edition (November 2009)

 

I don't understand what you mean by this? The article says this (p47):

 

"...the higher ISO values [of the M9] benefit from moderate noise suppression which is run prior to storing the raw information on card"

 

So the M9 is running an in-camera noise-suppression algorithm.

 

Is this what you meant? I'm really confused by both your statement and the article itself, which seems to go out of its way to find every characteristic of the M9 to be 'better' than the other two cameras: more "punchy" color = better; more red (apparently for improved noise characteristics) = better, and so on, but which also on page 45 says "Raw files generated with the M9... reach the memory card untouched." Which is then directly contradicted over the page - apparently by the same author!

I don't wanna say that the article was dishonest or intentionally boring (possibly hoping that no-one would read all the way to page 47?), but it was definitely one of the weakest pieces of writing I've seen in LFI for quite a while, and I didn't even believe a lot of the guff they wrote about the M8.

 

Btw I see nothing wrong with the (very boring) M9 images in the article (other than the glaring pop-out magenta of the night-time sign on p46, which looks totally artificial and stuck-on), but I'd say the reader pretty obviously gets the feeling that the magazine wanted a certain camera to 'win'.

 

And I also agree with Ian about the awesome photos by Stefan Rohner in the magazine. That guy shows every time that the equipment doesn't matter - he even posted another photographer's iPhone image of his daughter on his blog recently that blows away most of what I've seen here the last few months.

[jacobh]

Unfortunately the article leaves several things to be desired. For starters they do not justify why they chose a Summilux 1.4 on the M9 and a Summicron 2.0 on the Canon. One may argue that the characteristics at 5.6 do not differ much but why chose different optics in the first place? Further, in order to eliminate any influence of the optic's characteristics at least some tests should have been repeated after exchanging the optics. As it stands now I find the conclusions not fully convincing.

 

Jaap Stil

[plasticman]

But test pictures aren't supposed to be art, are they? They're there to demonstrate certain things, such as in the case of the first group how much detail is rendered clearly (compare the lettering on the white sign and the brickwork).

 

Incidentally, while I'd say it's entirely fair to scale images to the same size if final print is what's being compared, the way the images are presented on page 42 (labelled as crops which seems to me to imply 100% crops of the sensor) is also at least slightly misleading. To scale images to different dpi's and then compare those as crops doesn't seem totally scientific to me - optimal dpi for printing is not necessarily the highest, and sharpening and other scaling techniques would definitely have changed the results if another result had been desired, imho.

So not great images, and not a great test, I'd say (though the girl does look cutest in the M9 image on p44, she looks like she really wants the camera in that shot, so I guess that might be worth spending the extra $$!)

[plasticman]

Advertisement (gone after registration)

I'm genuinely surprised that no-one has anything whatsoever to say about the contradictory statements about M9 in-camera noise-reduction in the LFI article. No-one want to clarify this?

[modschiedler]

I'm genuinely surprised that no-one has anything whatsoever to say about the contradictory statements about M9 in-camera noise-reduction in the LFI article. No-one want to clarify this?

 

I think it’s just not such a big story behind that. The Leica people do a very "slight" noise reduction when it comes to high ISOs, whereas with CMOS sensors you have such kind of "manipulation" as a default procedure, "by nature", so to say. Probably the LFI people simply had the feeling that they shouldn’t conceal that M9 DNGS are noise-suppressed as well, but on a totally different level compared to any CMOS-generated file.

Regarding what you say about the "crops": What’s wrong with scaling the images on p42 to a comparable cut-out, thus making resolution issues and output capacities visible?

[ho_co]

I'm genuinely surprised that no-one has anything whatsoever to say about the contradictory statements about M9 in-camera noise-reduction in the LFI article. ...

 

Mani, I think you've covered it so far. It's obviously a major discrepancy. Thanks for calling attention to it.

 

Most of the readers of the forum can't comment on it, because we haven't any more knowledge than you.

 

My copy of the magazine hasn't yet arrived. Who wrote the piece? Would Michael Hußmann likely have any inside knowledge of the situation?

 

A couple folks on the forum thought that noise reduction was applied but were unable to find solid evidence, so the amount must be small.

 

Definitely something that needs follow-up.

[Wonka]

Won--

You can subscribe through the German site already mentioned, in either German or English. I did that for several years in the US. It's reliable, but the magazine arrives about a month after it goes on sale in the kiosks (if you can find someone who sells it on this side of the Atlantic).

 

I suggest you instead contact German Language Publications instead (GLPNews.com home). Find the LFI page by typing "Leica" into the search box on that page.

 

In my experience, with a subscription through GLP, you'll get the magazine two to three weeks earlier than going through the Germany link.

 

And I think you can order back-issues from either site.

 

Thanks for the advice. I'll have to look into this.

[mjh]

Who wrote the piece?

The author is Holger Sparr. I would have to ask Holger about the details; at the moment I just have his article to go on. The relevant passage is this: “It [the M9] sports new and more responsive colour filters [read: a new, improved dye used in the red filters resulting in a slightly broader transmission curve and an increased sensitivity of the red-sensitive pixels] as well as a new analogue-to-digital converter. In addition, the higher ISO values benefit from moderate noise suppression which is run prior to storing the raw information on card; and it only affects noise which would be difficult to suppress in post production. At pixel level the extent of this step is only mildly visible ....” (Comments in angular brackets by me.)

 

This sounds a bit vague (as does the German original, in case you were wondering: “… gönnte Leica der M9 in den oberen Empfindlichkeitsstufen eine ganz dezente Rauschunterdrückung noch vor der Sicherung der Rohdaten, die sich nur um Rauschanteile kümmert, die sich nachträglich kaum noch unterdrücken ließen.”), probably deliberately so. Maybe Holger merely relayed what Leica had told him, since who else but Leica’s engineers would know what exactly is going on?

 

So the article doesn’t spell out what kind of noise reduction is applied, but one can guess. A likely candidate for the kind of noise one would wish to eliminate as early as possible would be fixed-pattern noise due to non-uniformities in the sensor. Now this is not a CMOS sensor which wouldn’t yield good results without some healthy amount of fixed-pattern noise reduction applied at all ISO values. But even in a CCD it could cause artifacts such as banding that would be more prominently visible in images taken at higher ISO settings where the increased amplification of the sensor output would also amplify any non-uniformities present. Fixed-pattern noise is fairly trivial to get rid off if you know what that fixed pattern is, but quite tricky to remove if you don’t. Naturally the camera could be aware of that pattern but a raw converter would not, so it would be logical to eliminate the fixed-pattern noise in-camera.

[diogenis]

I don't understand what you mean by this? The article says this (p47):

 

"...the higher ISO values [of the M9] benefit from moderate noise suppression which is run prior to storing the raw information on card"

 

So the M9 is running an in-camera noise-suppression algorithm.

 

Is this what you meant? I'm really confused by both your statement and the article itself, which seems to go out of its way to find every characteristic of the M9 to be 'better' than the other two cameras: more "punchy" color = better; more red (apparently for improved noise characteristics) = better, and so on, but which also on page 45 says "Raw files generated with the M9... reach the memory card untouched." Which is then directly contradicted over the page - apparently by the same author!

I don't wanna say that the article was dishonest or intentionally boring (possibly hoping that no-one would read all the way to page 47?), but it was definitely one of the weakest pieces of writing I've seen in LFI for quite a while, and I didn't even believe a lot of the guff they wrote about the M8.

 

Btw I see nothing wrong with the (very boring) M9 images in the article (other than the glaring pop-out magenta of the night-time sign on p46, which looks totally artificial and stuck-on), but I'd say the reader pretty obviously gets the feeling that the magazine wanted a certain camera to 'win'.

 

And I also agree with Ian about the awesome photos by Stefan Rohner in the magazine. That guy shows every time that the equipment doesn't matter - he even posted another photographer's iPhone image of his daughter on his blog recently that blows away most of what I've seen here the last few months.

Well from what I understand, the bottom line of this article is that with leicas you get files with no de-noise processing and the most natural, accurate colours. Micro resolution/contrast were better or showed better on the M9 photos -which incidentally were simple shots to illustrate and showcase what the guys wants to prove, and not artistic photos like Rohner's which are stunning- and that when you compare with a camera with 20% more pixels.

I don't understand what this guy there made wrong ...

[ennjott]

As far as I know, with vague terms like "on-chip noise reduction", manufacturers typically refer to CDS (correlated double sampling) which is a standard technique for such applications and essentially simple analog NR similar to dark frame subtraction.

[carstenw]

While I agree with the article and the conclusions in general, in other words, I don't think the author let himself be unfairly biased in order to show the Leica in a better light in the conclusion, some of the steps along the way were a little questionable.

 

The Summilux-M vs. Summicron-M vs. Summicon-R was one such problem. Both the former lenses are significantly sharper (I have owned all three), and he should have used the latest Summilux-R to rule out this potential factor.

 

Secondly, the "sign in front of the bridge" shot was not properly done. The focus was on the sign, but in the discussion, he examined the sharpness of the bricks behind. All cameras were shot at f/5,6 which gives an advantage to the M8, with a shorter effective focal length (47 vs. 52 or so) and greater depth of field, depending on how the framing was done and so on. There might not be an issue here, but it does leave a question hanging... If the shots were taken from the same spot, the M8 would have been slightly favoured, DoF-wise, unless I messed up my math.

 

The portrait was also slightly odd, in that none of the coats were really grey, although they supposedly took the white balance from it. The skin in the M9 shot was a little too red, and in the 5D2 shot it was a little too green. The M8 looked best there, but in reality, it is probably the worst of the three for skin tones.

 

And so on.

 

But the conclusion was fair enough in the end, I thought.

[mjh]

Secondly, the "sign in front of the bridge" shot was not properly done. The focus was on the sign, but in the discussion, he examined the sharpness of the bricks behind. All cameras were shot at f/5,6 which gives an advantage to the M8, with a shorter effective focal length (47 vs. 52 or so) and greater depth of field, depending on how the framing was done and so on. There might not be an issue here, but it does leave a question hanging... If the shots were taken from the same spot, the M8 would have been slightly favoured, DoF-wise, unless I messed up my math.

Yes, the M8 was slightly favoured – which is immaterial when you compare the writing on the sign –, but even then both the M9 and EOS 5D Mark II did show more detail in the background. In these comparison shots, the factors limiting resolution are the number of pixels and the antialiasing filter (or lack of one), not the depth of field.

 

The skin in the M9 shot was a little too red, and in the 5D2 shot it was a little too green. The M8 looked best there, but in reality, it is probably the worst of the three for skin tones.

I happen to know the lady in question and I would say the M9 was the best in capturing the skin tones, with the M8 close behind.

[TimF]

The Summilux-M vs. Summicron-M vs. Summicon-R was one such problem. Both the former lenses are significantly sharper (I have owned all three), and he should have used the latest Summilux-R to rule out this potential factor.

Agreed. Without further explanation I suppose we have to presume the tester was unable to source one in time, which won't satisfy the conspiracy theorists of course!

 

Having said that, I just looked at the Canon-Leica database on Pebble Place. It suggests there that the last version Summilux needs to have the rear shroud removed to fit the 5D (and presumably 5D Mk II), so maybe that's another reason.

[woodyspedden]

How does one here in the U.S. obtain a copy of this magazine? Is it subscription only? If so, will they deliver here in the U.S.?

 

It is subscription and I have been receiving mine here in Colorado with no problems for almost four years

 

Great place to stay up to date on all things Leica

 

Woody