Gid
Well-known
Part 3 is now on reidreviews. Nothing to disappoint M8 owners. Interesting, though, that the RD-1 is better noise wise, although the difference is marginal.
Sean Reid's M8 Review
- Thread starter Gid
- Start date
- Latest activity Latest activity:
- Replies 92
- Views 11K
Grober
Well-known
I know this seems a bit backward but here's my priorities over the next two weeks:
1. hound my Leica dealer and remind him that I was the first to place an M8 order at their store many months ago;
2. if in fact this dealer gets one in Leica's first M8 allocations (rumored to be this Friday!), I'll hasten to his store and seize mine with both hands;
3. buy a Leica-approved SD card or two + a thrid-party protector for the monitor;
4. order two Leica acccessories: M8 handgrip and Neoprene case;
5. THEN subscribe and read Sean Reid's complete M8 review.
-g
1. hound my Leica dealer and remind him that I was the first to place an M8 order at their store many months ago;
2. if in fact this dealer gets one in Leica's first M8 allocations (rumored to be this Friday!), I'll hasten to his store and seize mine with both hands;
3. buy a Leica-approved SD card or two + a thrid-party protector for the monitor;
4. order two Leica acccessories: M8 handgrip and Neoprene case;
5. THEN subscribe and read Sean Reid's complete M8 review.
-g
Last edited:
ywenz
Veteran
Okay, is it me or does the RD-1 stand up really well to the M8!? The lower resolution of the RD-1 is apparent when compared to the M8 at 100% crop. The M8 image is bigger but does not show addtional details. I suspect a lot of the sharpness in the M8 crop has to do with the fact that there is no AA filter.
The RD-1 is better than the M8 at high ISO. This is very apparent. Shame on Leica for not measuring up to previous generation technology in this regard.
I continue to think Sean's review does not sell the M8 well at all. The lackluster images speak volumes. And what is the deal with the lack of wide open shoots - images with the "Leica glow"? Even if they're used simply as eyecandy?
The RD-1 is better than the M8 at high ISO. This is very apparent. Shame on Leica for not measuring up to previous generation technology in this regard.
I continue to think Sean's review does not sell the M8 well at all. The lackluster images speak volumes. And what is the deal with the lack of wide open shoots - images with the "Leica glow"? Even if they're used simply as eyecandy?
Last edited:
jlw
Rangefinder camera pedant
Interesting how well the R-D 1 holds up considering that it's now "old technology" with a lower pixel count.
I am absolutely, positively sure that Epson would never consider sticking its toe in these waters again, considering that the R-D 1 was not a barn-burning sales success, and that (justified) complaints about initial quality control, service availability, etc., along with (dubiously justified) kvetching by digi-gadget pundits about the manual shutter wind lever etc. neutralized any "halo effect" they might have hoped to achieve by introducing such an innovative niche product. Still...
...contemplate that the R-D 1 is essentially a Cosina Bessa R3a with a Nikon D70 sensor in it. Now suppose that Epson had not taken such a whupping over the R-D 1's real and imaginary flaws, and now commissioned Cosina to manufacture a camera with the Zeiss Ikon's rangefinder mechanism and a Nikon D80 10-megapixel sensor. Let's assume this hypothetical camera would retain such well-liked R-D 1 attributes as the analog-like controls, the folding LCD, the excellent black-and-white performance, etc.
Given the price difference between a Zeiss Ikon and a Bessa, and between the D70's initial price and the D80's current price, I'd guess that such a camera could sell for about $1,000 less than the M8.
Now, wouldn't that be an interesting situation? Too bad it's never going to happen, given all the brickbats tossed at the R-D 1.
(Yeah, I know, a lot of you would prefer a Nikon D80 or Canon Rebel sensor in a camera that could sell for $1,495, but I just don't see how it would be possible to add a good rangefinder mechanism and the body precision it requires for $500 over the price of the parent mass-market DSLR...)
I am absolutely, positively sure that Epson would never consider sticking its toe in these waters again, considering that the R-D 1 was not a barn-burning sales success, and that (justified) complaints about initial quality control, service availability, etc., along with (dubiously justified) kvetching by digi-gadget pundits about the manual shutter wind lever etc. neutralized any "halo effect" they might have hoped to achieve by introducing such an innovative niche product. Still...
...contemplate that the R-D 1 is essentially a Cosina Bessa R3a with a Nikon D70 sensor in it. Now suppose that Epson had not taken such a whupping over the R-D 1's real and imaginary flaws, and now commissioned Cosina to manufacture a camera with the Zeiss Ikon's rangefinder mechanism and a Nikon D80 10-megapixel sensor. Let's assume this hypothetical camera would retain such well-liked R-D 1 attributes as the analog-like controls, the folding LCD, the excellent black-and-white performance, etc.
Given the price difference between a Zeiss Ikon and a Bessa, and between the D70's initial price and the D80's current price, I'd guess that such a camera could sell for about $1,000 less than the M8.
Now, wouldn't that be an interesting situation? Too bad it's never going to happen, given all the brickbats tossed at the R-D 1.
(Yeah, I know, a lot of you would prefer a Nikon D80 or Canon Rebel sensor in a camera that could sell for $1,495, but I just don't see how it would be possible to add a good rangefinder mechanism and the body precision it requires for $500 over the price of the parent mass-market DSLR...)
Nachkebia
Well-known
Zeiss Ikon digital with full frame non AA filter sensor (fuji or sigma) would be haven to snap around with digital 😀
MarcoS
R9/DMR . M8 . R-D1
The framings in the comparison between R-D1 and M8 (and M8 vs 5D) weren't compensated for the different cropping factor.
The crops did help in showing the better per-pixel quality of the M8, because the subjects are more similar in visual dimensions, but in real world we always compensate for the different cropping factor, moving forward or backward, even changing lens... we need a certain framing and we behave accordingly.
The R-D1 image is 14% bigger in linear dimensions than that of the M8, hence when viewing 100% crops it has a slight advantage as far as resolution, like if it was a 7.8 Mp camera.
Same for the M8 vs 5D comparison, with the M8 being advantaged.
As for the noise comparison, I always found my R-D1 quite clean at hgh ISO, much better than one could argue judging from the (D70) sensor's technology.
I think I'll be happy with the M8 noise, which is more of grain than "noise" indeed.
Great review anyway 🙂
The crops did help in showing the better per-pixel quality of the M8, because the subjects are more similar in visual dimensions, but in real world we always compensate for the different cropping factor, moving forward or backward, even changing lens... we need a certain framing and we behave accordingly.
The R-D1 image is 14% bigger in linear dimensions than that of the M8, hence when viewing 100% crops it has a slight advantage as far as resolution, like if it was a 7.8 Mp camera.
Same for the M8 vs 5D comparison, with the M8 being advantaged.
As for the noise comparison, I always found my R-D1 quite clean at hgh ISO, much better than one could argue judging from the (D70) sensor's technology.
I think I'll be happy with the M8 noise, which is more of grain than "noise" indeed.
Great review anyway 🙂
S
sreidvt
Guest
ywenz said:
The noise difference between the two is minimal. The M8 does resolve much more detail but I do agree with you that the R-D1 performed quite well. I wouldn't say that the M8 produces lackluster pictures at all but if you feel that way you'll certainly save your wallet a bit.
My reviews aren't ever designed to sell anything. I don't care who makes what, I just try to describe the thing honestly to the best of my ability.
Wide open pictures will come in the lens reviews, that's all about the lens, not the camera.
Cheers,
Sean
S
sreidvt
Guest
jlw said:
Epson cooked it's own goose with marketing and supporting the R-D1. I agree that the camera you describe could be very appealing. The R-D1 sensor, btw, is based on the D100 unit. There's always Zeiss...some day.
Cheers,
Sean
S
sreidvt
Guest
MarcoS said:
Hi Marco,
Thanks.
There are pros and cons to the various ways one can do that kind of comparison. My current feeling is that same lens and same vantage point provides the best comparison because the perspective is identical and, importantly, so is the DOF (until one allows for CoC of different sensor sizes, etc.)
With the first 5D/M8 comparison I worked by approximately matching FOV. There, the 5D lost DOF because of its longer lens and that difference was problematic. With the second comparison I kept focal length the same and just let the M8 crop in. Again, I'm coming to prefer the latter approach and that's the way I plan to go from now on, I think. I think it provides the purest information.
Cheers,
Sean
jlw
Rangefinder camera pedant
This whole mystique of the absence of an anti-aliasing filter is something that leaves me a bit dubious -- I'm sure there are situations in which it gives a better result, and I suspect that there also will be a lot of situations in which it gives a worse result. Many of the M8 sample pix in Sean's review look to me as if they're simply over-sharpened, leaving me to wonder how much of the effect is due to the absence of an AA filter and how much of it is the product either of in-camera processing or of the default settings of the Capture One software he used for all his test shots.
Since it seems to be one feature (or absence-of-feature) that everyone is talking about in connection with the M8, let's think a little harder about this whole anti-aliasing business. First, why is it necessary? Well, unlike film (which, for all practical purposes, is continuously sensitive across its whole area) a digital sensor is sensitive to light only in small "pits" distributed in a regular pattern. The locations of these pits correspond with the pixels in the final image, but it's important to remember that each sensitive pit is much smaller than the area of the final pixel -- there's a big void between each pair of pits that isn't sensitive to light at all. You can think of these voids as forming a large, fuzzy, dark grid that's superimposed over the image formed by the lens.
When you're photographing objects with strong patterns (uncommon in some types of photography, common in others) this "dead zone" can cause unexpected image effects. If a darker area in the pattern happens to line up with a dark area of the "dead zone," the final image will show it as much darker than a similar area that happens to fall into a photosite's sensitive area.
When the subject's pattern is regular, this interaction can cause moiré effects, in which a larger, ghostly pattern is distractingly visible.
Although it's seldom discussed and is less obvious, it's also clear that the interaction can cause effects on non-regular patterns. This phenomenon is now unpleasantly familiar to film scanner users as "grain aliasing" -- if you scan a coarse-grained black-and-white negative, the resulting image will look much grainier than a conventional "wet print" of the same negative. The interaction of the film's random grain pattern and the scanner's regular "dead zone" pattern causes a larger, irregular pattern to appear.
I wonder if some of the M8's vaunted microcontrast might not be a subject-level version of the same phenomenon -- "texture aliasing," you might call it? Admirers of this effect might say that as long as the texture looks good, who cares whether it's real or artificial?
To get back to the suppression of aliasing, this generally is done in two ways -- but it's important to realize that they both do exactly the same thing. Fundamentally, the strategy is to reduce aliasing by "contaminating" some of the signal from each pit with part of the signal from the surrounding pits. This "contamination" averages the values a bit, so that false patterns are suppressed.
One way to accomplish this averaging is to cover the sensor with a filter that has no effect on "low spatial frequencies" (coarse details) but applies a slight blurring to "high spatial frequencies" (fine details.) This property is why an anti-aliasing filter is also sometimes called a "low-pass filter" -- low-frequency detail passes right through it, while high-frequency detail is scattered just enough to average itself over several adjacent pits.
The other option is to do exactly the same thing mathematically, using a software algorithm. The effect on the image should be exactly the same. The advantage, of course, is that you can turn off anti-aliasing software when it's not needed, or tweak its parameters to give the best match to a particular subject. The disadvantages are that it's slower, and that the results may be worse if the user (or the algorithm) makes the wrong choices.
[Non-subscribers to "Reid Reviews" may want to skip the following section:]
It interested me to note that while Sean's comparison of the M8 and the R-D 1 showed an affinity for the software-based approach, he gets a bit equivocal depending on whether the comparisons favor the M8 or not.
Early in the review, he rhapsodizes about the sharp details of an M8 picture compared to an R-D 1 picture, speculating that the lack of an AA filter has a lot to do with the Leica shot's crisp appearance. He demonstrates that much of the Epson picture's crispness can be restored by applying a small amount of unsharp masking -- but dismisses this as being not as good as "real" sharpness.
Later on, though, when comparing the M8's high-ISO noise levels to the Epson's (the Leica's are visibly higher at ISO 1600) he notes that the Leica produces a higher pixel count, and that downsampling its images to the same dimensions as the Epson's has the effect of smoothing out its noise. His samples demonstrate this. BUT -- the reason it works is that downsampling averages -- that is, anti-aliases -- adjacent pixels!
In other words, Sean seems to feel that "real" sharpness (Leica) is much better than software-generated "fake" sharpness (Epson)... but that software-induced "fake" low noise (Leica) is just as good as "real" low noise (Epson)! Huh?
Incidentally, another potential source of aberrations in Sean's writeup (you non-"Reid-Reviews" subscribers may as well go get a sandwich or something) seems to be his use of Capture One for raw conversion of all his comparison shots. If you've read Michael Reichmann's M8 review on "Luminous Landscape," you may have been as surprised as I was to see how much difference can be made by using different raw-conversion applications on the same original file; apparently, per Reichmann, this is caused by differences in how the different software manufacturers build their calibration files for each camera.
So... Sean's comparison photos seem to show that the Leica produces more vivid, detailed, and subtle color than the Epson. Then again, Leica has a relationship with Capture One (a limited version of their software is supplied with the M8 as its raw-conversion program) so it stands to reason they might have better calibration data for it than they do for Epson, correct? The only way to find out for sure would be to pass the Epson files through Epson's own converter software (if they don't have a good calibration for their own camera, they have only themselves to blame!) and see if there are any differences.
I'm not saying Sean should have done this -- he's done a lot of useful work already -- but I am saying you shouldn't be too quick to assume that your results would duplicate his, if your raw-conversion workflow happens to involve different software choices.
Oh, well, if nothing else, this newfound wealth of detailed reviews provides something to discuss for those of us who are interested in the M8 but can't currently afford one!
Since it seems to be one feature (or absence-of-feature) that everyone is talking about in connection with the M8, let's think a little harder about this whole anti-aliasing business. First, why is it necessary? Well, unlike film (which, for all practical purposes, is continuously sensitive across its whole area) a digital sensor is sensitive to light only in small "pits" distributed in a regular pattern. The locations of these pits correspond with the pixels in the final image, but it's important to remember that each sensitive pit is much smaller than the area of the final pixel -- there's a big void between each pair of pits that isn't sensitive to light at all. You can think of these voids as forming a large, fuzzy, dark grid that's superimposed over the image formed by the lens.
When you're photographing objects with strong patterns (uncommon in some types of photography, common in others) this "dead zone" can cause unexpected image effects. If a darker area in the pattern happens to line up with a dark area of the "dead zone," the final image will show it as much darker than a similar area that happens to fall into a photosite's sensitive area.
When the subject's pattern is regular, this interaction can cause moiré effects, in which a larger, ghostly pattern is distractingly visible.
Although it's seldom discussed and is less obvious, it's also clear that the interaction can cause effects on non-regular patterns. This phenomenon is now unpleasantly familiar to film scanner users as "grain aliasing" -- if you scan a coarse-grained black-and-white negative, the resulting image will look much grainier than a conventional "wet print" of the same negative. The interaction of the film's random grain pattern and the scanner's regular "dead zone" pattern causes a larger, irregular pattern to appear.
I wonder if some of the M8's vaunted microcontrast might not be a subject-level version of the same phenomenon -- "texture aliasing," you might call it? Admirers of this effect might say that as long as the texture looks good, who cares whether it's real or artificial?
To get back to the suppression of aliasing, this generally is done in two ways -- but it's important to realize that they both do exactly the same thing. Fundamentally, the strategy is to reduce aliasing by "contaminating" some of the signal from each pit with part of the signal from the surrounding pits. This "contamination" averages the values a bit, so that false patterns are suppressed.
One way to accomplish this averaging is to cover the sensor with a filter that has no effect on "low spatial frequencies" (coarse details) but applies a slight blurring to "high spatial frequencies" (fine details.) This property is why an anti-aliasing filter is also sometimes called a "low-pass filter" -- low-frequency detail passes right through it, while high-frequency detail is scattered just enough to average itself over several adjacent pits.
The other option is to do exactly the same thing mathematically, using a software algorithm. The effect on the image should be exactly the same. The advantage, of course, is that you can turn off anti-aliasing software when it's not needed, or tweak its parameters to give the best match to a particular subject. The disadvantages are that it's slower, and that the results may be worse if the user (or the algorithm) makes the wrong choices.
[Non-subscribers to "Reid Reviews" may want to skip the following section:]
It interested me to note that while Sean's comparison of the M8 and the R-D 1 showed an affinity for the software-based approach, he gets a bit equivocal depending on whether the comparisons favor the M8 or not.
Early in the review, he rhapsodizes about the sharp details of an M8 picture compared to an R-D 1 picture, speculating that the lack of an AA filter has a lot to do with the Leica shot's crisp appearance. He demonstrates that much of the Epson picture's crispness can be restored by applying a small amount of unsharp masking -- but dismisses this as being not as good as "real" sharpness.
Later on, though, when comparing the M8's high-ISO noise levels to the Epson's (the Leica's are visibly higher at ISO 1600) he notes that the Leica produces a higher pixel count, and that downsampling its images to the same dimensions as the Epson's has the effect of smoothing out its noise. His samples demonstrate this. BUT -- the reason it works is that downsampling averages -- that is, anti-aliases -- adjacent pixels!
In other words, Sean seems to feel that "real" sharpness (Leica) is much better than software-generated "fake" sharpness (Epson)... but that software-induced "fake" low noise (Leica) is just as good as "real" low noise (Epson)! Huh?
[Sean, I'm not accusing you of bias here -- only noting that this whole realm of software interaction is one that perhaps we all need to think about more closely than we do. I only just thought of the anti-aliasing/downsampling symmetry myself...]
Incidentally, another potential source of aberrations in Sean's writeup (you non-"Reid-Reviews" subscribers may as well go get a sandwich or something) seems to be his use of Capture One for raw conversion of all his comparison shots. If you've read Michael Reichmann's M8 review on "Luminous Landscape," you may have been as surprised as I was to see how much difference can be made by using different raw-conversion applications on the same original file; apparently, per Reichmann, this is caused by differences in how the different software manufacturers build their calibration files for each camera.
So... Sean's comparison photos seem to show that the Leica produces more vivid, detailed, and subtle color than the Epson. Then again, Leica has a relationship with Capture One (a limited version of their software is supplied with the M8 as its raw-conversion program) so it stands to reason they might have better calibration data for it than they do for Epson, correct? The only way to find out for sure would be to pass the Epson files through Epson's own converter software (if they don't have a good calibration for their own camera, they have only themselves to blame!) and see if there are any differences.
I'm not saying Sean should have done this -- he's done a lot of useful work already -- but I am saying you shouldn't be too quick to assume that your results would duplicate his, if your raw-conversion workflow happens to involve different software choices.
Oh, well, if nothing else, this newfound wealth of detailed reviews provides something to discuss for those of us who are interested in the M8 but can't currently afford one!
Last edited:
iml
Well-known
jlw: really interesting post. I've learned something today.
Ian
Ian
jlw
Rangefinder camera pedant
sreidvt said:
Right, I forgot. Peripherally (but not quite incidentally) I just replaced my own D100 with a D80. I'd have to say that except for pixel count (which really only comes into play with large enlargements) I can't see one dratted bit of difference in image quality between the two. I guess maybe the sensor manufacturers have hit some kind of plateau. Or maybe it's just, to paraphrase what you say in your review, pixels ain't everything...
MarcoS
R9/DMR . M8 . R-D1
sreidvt said:
Sean,
thanks for your answer, I respect your point of view.
Maybe you could add a separate paragraph "same framing disregarding dof", photographing distant objects at medium apertures (like distant trees at f/8 or f/11) ? 😀
Ok, better not to make the hard work even harder for this nice guy.. 🙂
Benjamin Marks
Veteran
jlw - fantastic stuff. What a good read. Sean, if you're reading this thread, I thought that Part III of your review was the best yet. Keep up the great work!
Ben Marks
Ben Marks
S
Simon Larbalestier
Guest
Another good review Sean and well reasoned. From what i've read i think the M8 will certainly work for me even at 3200. What a pity it's not properly weather sealed and produced with the slippy MP covering.....
More often than not i'm shooting in the rain or at the very least hot humid situations - a strong reason to keep the film M's alongside.
More often than not i'm shooting in the rain or at the very least hot humid situations - a strong reason to keep the film M's alongside.
R
rjcruz
Guest
I just came over here from the FM forum. Looks like they banned Sean for promoting his pay to read reviews. I was upset that they banned him. I think he was contributing good information to the forum in the threads. On the other hand I can understand why they banned him as well. It can seem like he was only publishing with plugs to his reviews. I don't think Fred Miranda has any paying advertisers. But if you are going to promote your commercial site you might as well take out an ad. In all honesty I rather read the LL review which is free and less objective.
J. Borger
Well-known
And what have these remarks about website policy to do with the discussion we are having here about the M8?rjcruz said:
John Camp
Well-known
JLW,
Nice post. There is one advantage to software that you didn't mention, and it's to the Leica's benefit. (I'm not biased in this; I have an R-D1 which I like very much, with a Leica to arrive Saturday.) Sean in his Leica 2 review shows an aliasing effect on a piece of loosely woven cloth (the "scarecrow's hat.") But instead of running software that corrected the entire file (and thus blurred it a bit) he selected only the part that showed aliasing, and cleaned that up. The rest of the picture, which was most of it, retained its sharpness. Since most photos show no aliasing because they don't have the tight patterns, they don't need to be cleaned up at all, and will seem unnaturally sharp compared to cameras with an AA filter. (Not sharpened, but sharp.) I suspect most people who choose to shoot Leicas would prefer it this way -- to selectively clean up their own files, rather than have the camera do it for them.
JC
Nice post. There is one advantage to software that you didn't mention, and it's to the Leica's benefit. (I'm not biased in this; I have an R-D1 which I like very much, with a Leica to arrive Saturday.) Sean in his Leica 2 review shows an aliasing effect on a piece of loosely woven cloth (the "scarecrow's hat.") But instead of running software that corrected the entire file (and thus blurred it a bit) he selected only the part that showed aliasing, and cleaned that up. The rest of the picture, which was most of it, retained its sharpness. Since most photos show no aliasing because they don't have the tight patterns, they don't need to be cleaned up at all, and will seem unnaturally sharp compared to cameras with an AA filter. (Not sharpened, but sharp.) I suspect most people who choose to shoot Leicas would prefer it this way -- to selectively clean up their own files, rather than have the camera do it for them.
JC
S
Simon Larbalestier
Guest
rjcruz said:
It was Gid not Sean who opened this thread. I don't see all this as plugging his own site, Sean's contributions, as on the Leica Customer Forum, are very constructive and make the threads much more dimensional and interactive.
That aside $26.50 for a years subscription is less than night out drinking so i think it's a small price to pay for a lot of useful information.
R
rjcruz
Guest
It just seemed like an appropriate topic to address the subject since it is titled Sean Reid's M8 review. Nothing more than that.
Similar threads
- Article
