Richard,

Was that shot with the 450D and Leica 28 R: taken anywhere near to the corner of Gower and Torrington ?

wolfloid wrote:
The other problem with the crop DSLR's are their awful viewfinders - exactly the opposite with the M8.

The 450D's viewfinder craps on the M8's viewfinder in the following ways:
- you get a rough idea where the edge of your picture is
- the M8 has no way to focus wide aperture normal or wideangle lenses on off-center subjects since camera rotation (focus-recompose) changes focus
- you can preview depth of field
- patterns (e.g. railings or brickwork) can be reliably focussed
- no need to turn camera vertical to focus horizontal lines
- no parallax problems when lining things up
- adequate camera information
- adjusts focal length with lens!
- zooms
- lens obstructions (e.g. condensation) are visible
- macro
- long telephoto
- ability to select a focus point
- optional autofocus!
- manual focus confirmation with selected point
- viewfinder stabilised with kit lens and some others
- Liveview with 10x magnification

I could go on, but you get the point.


Edited by brainiac on Jul 18, 2008 at 01:50 PM GMT

jcolwell wrote:
Richard,

Was that shot with the 450D and Leica 28 R: taken anywhere near to the corner of Gower and Torrington ?


Yes indeed. Well spotted, especially for a Canadian ;-)

No, correction: it was the South end of Gower St, some distance from Torrington Place. In the UK we always say the Street/Place/Square/Mews/Crescent/Close part since Colwell Street is often next to Colwell Mews and Colwell Square and Colwell Crescent etc., and leaving it off can lead to great confusion.

I own a M8 and am very happy with it. Image quality is very good, much better than I had from my 20D and L lenses.
Look around at flickr for M8 images, so you can see for yourself.

Leica glass is expensive, but other options are Zeiss and CV which are very good or equal to Leica (for a much lesser price). You can also buy the older designs (pre-asph) which are also very good.
If you check ebay or leica forums you can find a used M8 for a reasonable price.

The camera has its problems (need for UV-IR filters being one of them) , mine is working flawless for the few months I own one.
Noise can be visible in higher iso, iso not as good as Canon

Pro
- Great images quality
- Pretty small
- Outstanding lenses from Leica, Zeiss and Cosina

Con
- Noise at higher iso
- Camera can have problems (UV-IR, shutdowns)

Rumor is Leica will show an updated M (M9 or M8-II) at Photokina. Also some new lenses.

Handy links
Leica User forum
Rangefinder Forum
flickr M8 Pool
flickr M-mount Pool

Photo I took of a friend, view the full version to see the detail.



Full version

brainiac wrote:
So, in short, my opinion on this is based on careful consideration of output from the mooted cameras. But there's no need to believe me - do your own comparison. Fairly.

Ridhard, forget the M8's built-in noise reduction and compare RAW files instead of jpegs. Your comparison is horribly flawed because of limited experience with the M8. It's not a straight-out-of-the-box P&S camera, you have to learn how to use it and get the most out of it.

brainiac wrote:
[


In that spirit, here is a crop which shows that the M8 is editing out a particular frequency of data. The umbrella spokes should have been resolved, but got smoothed. .

Richard:

I saw your original files from that test and the lens was either out of focus, or just out of spec. you may have had a dud Leica lens, as rare as they are.

Robert

> Ridhard, forget the M8's built-in noise reduction and compare RAW files instead of jpegs. Your comparison is horribly flawed because of limited experience with the M8. It's not a straight-out-of-the-box P&S camera, you have to learn how to use it and get the most out of it.

Douglas, we've been through all of this before. I base my opinions not only on what I could do with it, but on files provided and processed by experts like Jack Flesher and Guy Mancuso. Leica has stated that the M8 does noise reduction before saving the raw file. It appears that this was necessary because of the lack of an AA filter, which was left off not because it helped, but because it worsened the incident light/reflection problem of the short register.

Their files, with all their know-how, didn't show anything different to my impressions from using the camera. It really is flogging a dead horse to say that if I spend more time with the camera some kind of magic will occur. When Jack did his test he had been using the M8 for a year. His test used Canon 5D, 1Ds2, and 1Ds3 files which were processed in what many here felt was a sub-optimal way due to the choice of raw processor, and still they surpassed the M8.

Lots of people enjoy M8's and that's great, but people interested in the relative quality of images from the M8 and other cameras should be allowed to speak about what they see. I regard Jack Flesher's test as being fairly conclusive that even sub-optimally processed 5D files are better, in the ways I have mentioned, than optimally processed M8 files. Specifically, 5D files have less red-fringing around highlights, more detail, and better ability to hold on to information in shadows. Go ahead and prove that wrong. I'm not saying you can't, but until you do, I base my opinion on personal use and the quality of files provided by experts in M8 use, and every example I have ever seen, including Hans' very kindly shared example above.

Don't get me wrong - I think the M8 is a very nice camera. But I think it's important that there seems to be a camera available for 1/3rd the price which provides better image quality in the specific ways I have mentioned. It's not enough to say M8 users claim their files are better. We have yet to see a single online example of an M8 outperforming a 5D in image quality in a fair test. Meanwhile, if you want to see the reverse, I refer you to Jack Flesher's 4-way test of the derelict building, if you can still find the version with the M8 files.

robsteve wrote:
I saw your original files from that test and the lens was either out of focus, or just out of spec. you may have had a dud Leica lens, as rare as they are.

Here we go again. I tested the camera twice, according to my needs, with two different lenses, one from Leica, and one from Zeiss. In each case the M8 file contained high contrast between adjacent pixels, in well illuminated parts of the image. The lenses were not faulty - they performed exactly as expected. Jack Flesher's test concurs exactly with mine: less detail, red fringing (moire), and noisy lossy shadows. These are features of the camera. Don't just argue with me for the sake of it, show me a fair comparison which contradicts the results which Jack Flesher and I both got in our tests.

Richard it may be that in your test according to your needs the M8 did not perform well, but learning to use a camera is not magic. It's experience and trying various ways of processing the files. Your needs don't necessarily translate into a blanket statement as you have made that the 450D produces better image quality. You'll have to tests a little more rigorous than your out-of-focus jpg files.

I agree, but I still don't understand why the M8's superiority is always anecdotal. Can somebody please show us a fair comparison which shows the M8's advantages?

I dug up Jack's test. Personally I think the Canon files might have been processed in DPP to better effect, and the 5D file seems to be overexposed compared to the M8 shot, but here they are anyway. I draw your attention to the red fringes around the bush branches, and just about every other high contrast edge in M8 files. I have seen these red fringes in just about every M8 raw file I have opened, and I believe they are moire related. Also note the lack of detail in many of the flat surfaces in the M8 file. It should be there because the pixels are, but something has removed it. It looks suspiciously like some fourier-transform/dust'n'scratches type of modification has been applied. Leica has stated that NR occurs in camera, even on raw files.

M8 file:
http://www.jirvana.com/raw_large/Canon_1ds3/Alviso_Session/m8_L2000077.jpg

5D file:
http://www.jirvana.com/raw_large/Canon_1ds3/Alviso_Session/5D__MG_2034.jpg

Thanks are due to Jack Flesher for kindly sharing these files and helping to enlighten us all.

It seems unlikely that my test was flawed in the light of Jack's files here, since his test shows exactly the same differences between the cameras as mine did: red-fringing and a detail advantage to the 5D. Here are two crops from my own comparison which was pilloried when I did it, not least by Jack Flesher himself. These are a 200% crop of the M8 and a 177% crop of the 5D, both resized to match each other in Photoshop using bicubic:

M8:

I hope no one will attack you for your opinion, but you are presenting some very questionable statements as fact. If there are very different experiences, I think it reasonable that people will challenge it and offer their own views.

I had the M8 and 5D together for several months and did some testing. My conclusion, based on printing, was that the image quality was basically a toss up. Both produce outstanding files.

But when you look at a test like Jacks you have to face the fact that more detail is visible at 100% on the screen with a 5D - if for no other reason than the file is bigger. This makes a 1-1 comparison a little tough. IF that is the criteria for judging then the 5D wins on detail, but if you look to produce similar size prints, it could be a different story.

I am not a Leica diehard who will proclaim it better than 4x5 (it and the 5D are fairly far behind even a 6x7), I don't think the noise at hi ISO is any more film like than another DSLR, and I don't think there is any magic to the files allowing incredible shadow detail extraction. But I also know that the camera produces great images - I have no idea what you mean by red fringing moire, but I have not seen any of it - it is not a feature of the camera. But it does require the expensive filter and they will flare at times.

Edited by tennclay on Jul 19, 2008 at 05:13 AM GMT

Get a beater 5D for ~$1500 used, and a Canon 35/2 for $200.

problem solved.

tennclay wrote:
But when you look at a test like Jacks you have to face the fact that more detail is visible at 100% on the screen with a 5D - if for no other reason than the file is bigger. This makes a 1-1 comparison a little tough. IF that is the criteria for judging then the 5D wins on detail, but if you look to produce similar size prints, it could be a different story.


I was pretty much with you except for this statement. Indeed, a bigger file, more pixels, can be the source of more detail, provided the lens is up to it. That is a pretty good reason for more detail. If you can see a detail difference on the screen, then the detail difference can be seen in a print, provided the enlargement is sufficient. A 1 to 1 comparison is not really difficult at all -- either you reduce one to the size of the other, or you enlarge one. Actually I think enlarging one to the size of the other will expose the differences in resolution and detail more readily. Now, this is fine if what are interested in is which camera produces the most resolution, or detail, is is sharper, or produces less artifacts like moire. This is valid comparison in it own right. However, each user should look at purchase with respect how they will use the output (all other things being equal, or a separate part of the purchase calculus). Depending on your output, you may never see a difference. No question about it and that might certainly be the case with decision between the 5D and M8. For your uses, in your printing, they might show detail that is practically the same. That, however, doesn't mean they are the same.

I don't doubt that the 5D might provide marginally better detail and resolution than the M8 as Richard (and Jack) point out, but I also don't doubt that you don't see a difference in your prints. The 2 positions are not mutually exclusive.

Tennclay, I agree with everything you say.

I don't think image quality is a matter of pure opinion. There is factual substance to many of the IQ issues that are discussed on this forum.

As regards red fringing, you can see it easily in Jack's M8 file in the following places:
- bush branches, bottom right
- lower broken edge of fibreglass sheet, top center
- along edges of white paint on gutter, top of turret
- around roof tiles where they edge against black background

This effect is also visible above the specular highlight on the button at the top of the camera in the crop that I have posted, as well as elsewhere in that image. I can show many other examples of this, but need I? It's false image data, and it seems to happen around high contrast edges wherever they are sharply resolved. If you have an open mind, please look at your M8 archive to see if your M8 does this too. It could be that Jack, Guy Mancuso, and I have all been unlucky enough to get hold of multiple M8's from a bad batch, but that's a bit of an ad hoc conjecture when there isn't much data suggesting that M8 copies exist which don't do this.

Another plug for Olympus OM lenses for a true lightweight travel kit. They're small, unobtrusive, very light, and take beautiful pictures (at least with my 5D). I use the 24/2.8 and 28/3.5, and also some Contax Zeiss lenses as well. By far, the Olys are the lightest of the bunch. A majority of the OM lenses are based on a 49mm filter size as well, so in that respect if you do use any type of filter (such as a polarizer), you'd need just one, lightening your possible load even more. The Contax Zeiss 85/2.8 is another excellent lens...takes incredibly sharp images, and it's no bigger or heavier than most 50mm primes.

As someone else may have already mentioned, you can switch out your focusing screen to one that's more suitable to MF. With my 5D, I'm using the Canon Ee-S screen. Also, if you go this route, you'd need the proper OM-EOS adapter (or whatever brand of lens you choose). These can be picked up on ebay...however, most of them ship from Hong Kong, so you'd have to plan accordingly as far as delivery time/need time goes. You can also use an AF-confirm adapter if necessary to get the quick pics indoors...when you get proper focus, the focus point will light up in the viewfinder. Some folks like them, some don't...but if you're in a situation where you need to snap a quick pic, it can be helpful.

Lotusm50 wrote:
tennclay wrote:
But when you look at a test like Jacks you have to face the fact that more detail is visible at 100% on the screen with a 5D - if for no other reason than the file is bigger. This makes a 1-1 comparison a little tough. IF that is the criteria for judging then the 5D wins on detail, but if you look to produce similar size prints, it could be a different story.


I was pretty much with you except for this statement. Indeed, a bigger file, more pixels, can be the source of more detail, provided the lens is up to it. That is a pretty good reason for more detail. If you can see a detail difference on the screen, then the detail difference can be seen in a print, provided the enlargement is sufficient. A 1 to 1 comparison is not really difficult at all -- either you reduce one to the size of the other, or you enlarge one. Actually I think enlarging one to the size of the other will expose the differences in resolution and detail more readily. Now, this is fine if what are interested in is which camera produces the most resolution, or detail, is is sharper, or produces less artifacts like moire. This is valid comparison in it own right. However, each user should look at purchase with respect how they will use the output (all other things being equal, or a separate part of the purchase calculus). Depending on your output, you may never see a difference. No question about it and that might certainly be the case with decision between the 5D and M8. For your uses, in your printing, they might show detail that is practically the same. That, however, doesn't mean they are the same.

I don't doubt that the 5D might provide marginally better detail and resolution than the M8 as Richard (and Jack) point out, but I also don't doubt that you don't see a difference in your prints. The 2 positions are not mutually exclusive.



I would agree 100% with your reply - and that is why used the word "could" in my last sentence. .

Yes, it all depends on the print size, and if one image starts with more detail you will be able to see that difference at some size print. It is only logical that my comment about viewing at 100% would translate to viewing a print at some threshold size. Personally, I only print up to 24" wide, and at that size, usually, neither camera produces prints acceptable to me - if there is fine detail involved. I prefer medium format film (I have no MF Digital, and probably never will!) for the larger ranges.

There is a famous thread in which a printing expert declared that his 30x40 prints from the M8 looked like a print from a LF 4x5 camera. That was taken to mean that the M8 files were the equivalent of 4x5 scans, and endlessly referenced across the web as proof of the 'magic' of the M8. Of course not much mention was made of the post processing techniques used (like adding noise to the file to mask the resizing artifacts) to produce that print, or the fact that similar techniques could be used to produce even larger 4x5 prints.

I did a well controlled test of the 5D vs a Mamiya 7II here a long while back. It put a dent in some "conventional wisdom" that the 5D had surpassed medium format and many arguments were put forth to prove that either the differences would never matter or that my technique was biased toward the film camera. Some of those arguments were - "could never see the difference in print" - of course you would at some print size, or "your interpolation was inadequate" - I posted the raw file and no one could produce a better up-res with any of their tools, or "I would do a pano stitch with the 5D" - of course that would never keep up with a pano stitch of the MF camera!

It all points out that personal preferences and goals should be the deciding factor. My intent was not to declare that the M8 produces files equal to or better in every way than the 5D, but more to address the statements presented as fact that the M8 has poor image quality - which is simply not the case.

tennclay wrote:
I did a well controlled test of the 5D vs a Mamiya 7II here a long while back. It put a dent in some "conventional wisdom" that the 5D had surpassed medium format and many arguments were put forth to prove that either the differences would never matter or that my technique was biased toward the film camera. Some of those arguments were - "could never see the difference in print" - of course you would at some print size, or "your interpolation was inadequate" - I posted the raw file and no one could produce a better up-res with any of their tools, or "I would do a pano stitch with the 5D" - of course that would never keep up with a pano stitch of the MF camera!


Yes. I've looked at my Mamiya 7II slide and my 5D output, and I could never understand what people were talking about when they would say that the 5D had surpassed medium format. There is no comparison -- the Mamiya 7II is significantly better -- AND I noticed the difference in prints from Epson 4800 and 4000 printers, not just on the monitor.

You can see the XSi + 24mm f/2.8 zuiko makes a small package.

I would consider the E420 + 25mm pancake. Great little kit.

"probably because of the camera's in-built noise reduction"

Are you talking about shooting JPEGs?

wolfloid wrote:
"probably because of the camera's in-built noise reduction"

Are you talking about shooting JPEGs?

No. During an interview a Leica employee stated that the camera does some noise reduction on all files before they are saved, even raw files. It is likely that this was necessary because the camera could not easily have an AA filter. Corner fall-off with wide lenses was bad enough as it was. The short register means some lenses have acute angles of incidence, and that increases reflection/fall-off problems in the corners.

I suggested that noise reduction software was active after my first test of the camera when it was released. The suggestion was ridiculed in this forum by Leica enthusiasts. Months later Leica announced that some noise/moire suppression takes place in the camera processor. In my opinion that is why the M8's 10 megapixel files seem to have the detail of 8 megapixel files. Some people call this look 'clean'. If you look carefully at Jack's M8 file and the 5D one which I linked to above, you should be able to see many areas where texture is less than the pixels should be able to resolve. Examine the lack of texture on the rusty upright in front of the turret, for instance. A certain frequency of detail seems to get zapped, as if by a dust and scratches filter.

Edited by brainiac on Jul 20, 2008 at 11:36 PM GMT

"- you get a rough idea where the edge of your picture is
- the M8 has no way to focus wide aperture normal or wideangle lenses on off-center subjects since camera rotation (focus-recompose) changes focus
- you can preview depth of field
- patterns (e.g. railings or brickwork) can be reliably focussed
- no need to turn camera vertical to focus horizontal lines
- no parallax problems when lining things up
- adequate camera information
- adjusts focal length with lens!
- zooms
- lens obstructions (e.g. condensation) are visible
- macro
- long telephoto
- ability to select a focus point
- optional autofocus!
- manual focus confirmation with selected point
- viewfinder stabilised with kit lens and some others
- Liveview with 10x magnification

I could go on, but you get the point. "

So many of your points sound like a checklist for photohobyists - macro, adequate camera information, patterns on brick walls etc. and all of it at the end of a dim tunnel. OK it meets your criteria, and everyone who wants to use long lenses or shoot macro, but what I want is to see my subject clearly, and I don't need the other stuff. In fact it's totally unimportant.

I want to see a momentary expression, a fleeting gesture, a particular relationship of the elements in the photograph as they come together, for that a bright clear view is essential - I don't want some vague indication of them. It also helps that it is easy to see beyond the frame, even to use the right eye and have an unimpeded view with both eyes open and the frame 'hanging in space'. All these are things I feel I'm sacrificing when I go back to the best FF viewfinders. The squinty little APS viewfinders are just the pits if you want to SEE what you are photographing, which, after all is the most fundamental thing. All the rest is just technical help, which, if you don't need it, is all pretty redundant.

As for framing, well this is really very quick and easy and pretty accurate once you get used to using a rangefinder.

As for off centre focusing, there are also quick and pretty reliable workarounds, as I mentioned elsewhere on the film thread. Wideangles on SLRs are notoriously innacurate in focusing wide open, particularly in dim light, whether the subject is in the centre or at the edge. Your objections sound more like a firm preference for SLRs and nothing more.

wolfloid wrote:
So many of your points sound like a checklist for photohobyists - macro

Macro is not a preserve of hobbyists.

...adequate camera information...

Pro bodies tend to have more information around the view, not less.

...patterns on brick walls

...patterns are everywhere - it's quite easy in a rush to overlap two eyes. Grass, railings, trees, fabrics, they can all catch you out and slow you down.

...and all of it at the end of a dim tunnel

Have you tried a 450D? With an f2 or f1.4 lens?

I want to see a momentary expression, a fleeting gesture, a particular relationship of the elements in the photograph as they come together, for that a bright clear view is essential - I don't want some vague indication of them. It also helps that it is easy to see beyond the frame, even to use the right eye and have an unimpeded view with both eyes open and the frame 'hanging in space'. All these are things I feel I'm sacrificing when I go back to the best FF viewfinders. The squinty little APS viewfinders are just the pits if you want to SEE what you are photographing, which, after all is the most fundamental thing. All the rest is just technical help, which, if you don't need it, is all pretty redundant.

I don't have a problem seeing what I'm photographing. Light is live. I often keep both eyes open so that I can view with one and frame with the other.

> Wideangles on SLRs are notoriously innacurate in focusing wide open, particularly in dim light, whether the subject is in the centre or at the edge.

Not on mine. You must have had some badly calibrated SLRs. The solution is to get the camera fixed. Rangefinders need to be calibrated sometimes too. Miscalibration of focussing mechanisms affects most kinds of cameras. Only liveview is immune to that problem.

> Your objections sound more like a firm preference for SLRs and nothing more.

They are not objections, but descriptions of limitations of rangefinder focussing. No preferences involved. Whether you like to focus and frame with a dim tunnel which gives you a preview of the picture, or a variably small rectangle with inaccurate edges in the middle of a large viewfinder is a matter of preference. That's why I didn't mention the peculiar annoyances of using the longer lenses on certain rangefinders (Contax G system excepted).

Please show me which of the rangefinder shortcomings I pointed out is a matter of preference. It seems to me that they are all just facts, viz.:
- you get a rough idea where the edge of your picture is
- the M8 has no way to focus wide aperture normal or wideangle lenses on off-center subjects since camera rotation (focus-recompose) changes focus
- you can preview depth of field
- patterns (e.g. railings or brickwork) can be reliably focussed
- no need to turn camera vertical to focus horizontal lines
- no parallax problems when lining things up
- adequate camera information
- adjusts focal length with lens!
- zooms
- lens obstructions (e.g. condensation) are visible
- macro
- long telephoto
- ability to select a focus point
- optional autofocus!
- manual focus confirmation with selected point
- viewfinder stabilised with kit lens and some others
- Liveview with 10x magnification

Rangefinder enthusiasts seem to think that you can't pre-focus an SLR, or shoot it with both eyes open, or without looking through the viewfinder. The truth is that you can do these things just as easily with an SLR as with a rangefinder. The techniques of spontaneity work just as well with both systems. Actually, autofocus often allows one to shoot surreptitiously or more quickly or with less need to use the viewfinder at all.

Wideangles on any SLR, whether accurately calibrated or not, are far less accurate in focusing than the same on rangefinders. This is due to the inherent focusing base.

In a rangefinder camera, focusing accuracy is dependent upon two factors, (1) the rangefinder base and (2) the magnification of the viewfinder. The rangefinder base in a Leica M6 is 68.5mm, multiply this by the viewfinder magnification which is .72 and you get an effective focus base length of 49.32 mm. This level of accuracy is constant for all lenses.

In an SLR, focusing "accuracy" is dependent upon the same two factors (1) the rangefinder base and (2) the viewfinder magnification. In an SLR, the rangefinder base is the focal length of the lens divided by the maximum aperture. For example, a 50mm F/2.0 lens has a rangefinder base of 25mm, 50mm/2 = 25mm. And a 35mm F/2.0 has a rangefinder base of 17.5, 35mm/2=17.5mm. A 24/2.8 used on an APS Canon (equivalent of a 38/2.8 on 35mm) has a base of 8.5. Multiply that by the viewfinder magnification, which is .87 and you have an effective base of 7.4mm. In other words almost SEVEN TIMES smaller than the Leica, or SEVEN TIMES less accurate.

Since our eyes don't see so well under low light - focusing an SLR camera with a wide angle lens under poor lighting conditions is neither rapid nor accurate. Even auto focusing can do nothing to compensate the tiny effective base length.
So, as you see this has nothing to do with calibration and everything to do with physics.

The rest of your remarks, I've already answered. As a photographer I don't want a camera that can fulfill all my hobbyist dreams. I know what I want to photograph, which excludes macro, sport, long-shot nature stc. etc. and I prefer the ergonomics, size, accuracy and uncluttered viewfinder of a rangefinder, which frames accurately enough for my use.

If you want the versatility that an SLR undoubtedly gives, and all the techincal aides that can help you, then fine. What I want is simplicity, the speed of working that I have developed and to see as clearly as possible what I'm shooting.

The Canon 5D with a 21/2.8 has an effective focusing base of 7.5mm x viewfinder magnification, which is .71, and = 5.325 mm. Or over nine times less than the EFB for the equivalent lens on a Leica. In other words the Leica is over nine times more accurate in focusing such a lens.

Even on the great 24/1.4 the EFB is only 12 mm. 4x less accurate than a 24 on a leica.

The 35/1.4 has an EFB of 17.5 mm, just more than a third of the Leica with the same lens, meaning the Leica is three times as accurate when focusing.

So unless these expensive Zeiss/Canon lenses are focused at infinity, it is highly likely that there is going to be focus error, which will only be compensated by depth of field.

Maybe this is why wide angle leicas can 'pop' the focus when shot close up and wide open.