Let me be clear what I mean by hyperfocal here. There are sharply focussed single pixel objects both behind, before, and on the point where I placed focus, namely on the sticker. You are suggesting that those sharply focussed objects before or behind the focussed point show that focus is off. I am saying that they are to be expected due to ample depth of field.
Corroborating my point is the fact that despite varying the focus setting on the lens systematically by small amounts, there appears to be little difference in the resolution of the sticker. Again, that is an expected result of being within the actual hyperfocussed range of a 21mm lens on this camera (yes, each camera has its own hyperfocal range since it depends on device resolution). As before, look for yourself. The examples are still here: http://cyberphotographer.com/m8v5d/m8_zeiss21
Edited by brainiac on Dec 12, 2006 at 01:18 PM GMT
If you vary the focus and see no change in resolution then within the limits that you varied the focus you are seeing the resolution limits of sensor + lens and this is what is happening in Richard's test. So even if the 21mm was slightly miss focused it will have no effect on the test. With a higher res sensor, wider aperture, longer lens or more of a focus error you would see it but not it seems with the 21mm at 5.6 at that distance. However the JPEG's still do not 'look' right compared to RAW files I have downloaded and examined from the M8 and apart from the bickering and sniping over the test methods there is something definatley amiss.
That is a good point Shirozina, you are correct about lesser focal variation occurring in off focus places as well. However, it does not explain how single pixel objects appear both behind and before the focus target. Should we be seeing half-pixel objects perhaps? ;-)
Any object or detail that is projected by the lens onto the sensor that is smaller than a pixel will be either rendered up to pixel size or not rendered at all - obviously the more pixels per area of sensor ( higher pixel count) the less this happens and more detail gets resolved and the more you can observe a definate focus plane with a gradual lessening of resolution away from it each side.
Since this is a 'hands on M8" and not a detailed thread, I want to raise a fundamental point; why are we discussing details that in many cases don't even play in the final image as used?
In the audio realm, the group at the Absolute Sound listen to music with a refernce system and then commented on how the tested system differed (the reference system is NOT necessarily the 'best')
In images, we hardly ever print or view at 100%. What matters is the images as viewed, and blow ups certainly help but it is the overall image that matters.
It would be good to pick a refernce, say a 5D with a 21, 35 and 75 to check against a M8 with 28, 50 90 M lens.
Then we cansee where the Leica M and the lenses create the quality of the images. If I had a 5D I would do it that way, with a variety of scene from real life.
The vocabulary would be subjective; 'deeper' (maybe meaning 3D effect), 'more dynamic' (DR?) 'more vibrant' (richer colors, tone etc)
other terms that I see are 'engaging' that is, M8 images to me are more interesting, less flat and analytical than my MF stuff, and yet in the test image above it is too cluttered to see anything (and 3D by selective focus for interiors can be jarring, as it is here. Either have a VERY sharp focus on ONE object, or get it all in focus)
The bottom line is to judge the total image effect; to repeat my own metaphor - view images without getting nose grease on your screen!
brainiac wrote:
...now we know why the DMR is better than the 1Ds2.
Yes, M8 and DMR will win this contest all the time.
If you crop out exact same pixel count, the camera with the Larger Pixel Pitch and the Weaker AA wins. That is all about 1:1 pixel peeing - "Pixel" Peeping.
I am not saying this just sitting on Couch. I went out did the test
Here is the test I posted @ Canon board.
135L lens @f/5.6 was used.
The Diff between the 5D and M8 will be much less dramatic than this. But my point again is you are just comparing the pixel pitch and AA strength. There is no other magic or myth to it.
I have my M8 since about 1 month and shot thousands of pictures with it.
The conclusion for me:
- Colors need some treatment until better profiles arrive
- WB is nearly unusable with firmware 1.06
- doesn't like every SD-card
Banding and IR-sensitivity getting solved with filters and hardware-fix.
Besides using the wrong SD-card (not listed) and the known banding, I had not a single malfunction or noticed any bug!
The 5D is faster, has less noise >800ASA but everything else like build quality, ergonomics, viewfinder, lenses... didn't meet my expectations - so I went with the M8 - no red dot, no emotions, no retro-feeling was necessary to make this decission!
File-quality in combination with new Leica-lenses is simply great!
Another issue (as with the DMR) is the bad JPG-file-quality. But it is a professional camera and for professional quality you use RAW anway. Or has somebody of you thrown away his negatives when he got his prints?
The effects shown in this "comparison" just look like the images from the M8-screen, the same information-loss, artifacts...
Make a real comparison, under real-world-conditions: Don't just test the sensor, test the whole system (+lenses). Use the whole frame (also image height >10mm), make some shots at open apertures, make them RAW and make the best out of them in Photoshop - then print them. And then compare those prints regarding out-of-center resolution, colors, DR, contrast, flare - just the WHOLE IQ... !
Guy Mancuso wrote:
On an other note i still think the best Canon made was the 1ds the orginal one and had the best balance of AA filter and image quaity. The files were very good just the noise in the shadows was a issue otherwise i would have never bought the 1dsMKII .
Very much agree with Guy here and as my test suggested the "original" 1Ds, once slightly sharpened, compares quite well with the M8 at pixel-level.
1Ds pixel count is only slightly higher (less than 150 pixel more in width), so the difference between the cameras are negligible, resolution wise.
However the combination between high resolution, wide DR, deep colors, lenses with great sharpness and 3D look, portability and relative "stealth" of the combo, make the M8 unique in the digital photography world.
I still prefer SLR vision, but admittedly this camera is a joy to use.
Well said both Gogopix and Pondria. I couldn't agree more with both of these posts. My test would be much better if I had the opportunity to use the M8 more extensively in real situations alongside a competing camera. But it is not worthless, and it has certainly highlighted the fact that the M8 with 21mm is a great way to get extra depth of field in a medium wideangle. If only the IR/cyan corner issue wasn't there, it would be a killer lightweight moderate resolution wideangle platform with affordable lenses like the Voigtlanders.
It is very important, as Pondria has demonstrated, that we don't look at simple 100% crops, but that instead we shoot the same angle of view and then uprez the files to match each other's pixel rates.
Pondria, yours is an excellent demonstration of the point that I have been raising since early in this thread, and a point that still seems to be ignored sometimes, even in reviews by people like Erwin Puts.
I would like to see my resolution comparison method, as described in my test, applied to a test between the DMR and 1Ds2.
Resolution does matter to photography. It always has. Years ago people would pay dearly for a particular lens because it was 2% sharper than some other lens. There should be no shame in comparing the real absolute resolution of digital cameras, just as we do with lenses. If it doesn't matter to you, that's fine, but please don't criticise those of us are interested in optimal resolution. Just ignore me! ;-)
Here is one scene taken just at sunset and the light truly was this wonderful blend of gold and red.The winter sun in these northern climes can be awesome.
Anyway, not a prized composition but I think it shows the wonderful depth of the M8/35 lux. I think this was shot at 5.6 or 8 and you can see the sharpness throuughout the image. (Yes those are sheep in the background.
Woody
Edited by woodyspedden on Dec 12, 2006 at 08:47 AM GMT
Guy,
the funny thing is that I feel in case of my lenses the 21 seems as least as sharp as my 24 at fast f-stops.
I think I just have a very good sample of the 21
Guy, I found the same re 21 vs 24.
24 is a bit crisper but I prefer the 21mm focal lenght and its overall slightly less contrasty look. Somehow it has also a nicer bokeh...
Right now I have too many lenses (12, 15, 21, 24, 28/2.8, 50/1, 75 Apo, 90 Apo, 135 Apo - waiting also for the Summilux 35 asph) and I need to narrow the setup.
So far I found I like the 28, 50 and 90 perspective.
28 and 90mm have also clean framelines.
The 135 IMO is too long for rangefinder use and the 75mm is maybe a bit too close to the 50mm. The Noctilux sometimes is wonderful, but I found I can focus the 90 Apo faster and more accurately (!) I think I'll trade it for the 50 asph.
Below 28mm I like the 21 asph but again it's rather close, usually I skip directly to the 15 CV.
CV 12 is probably too wide for most uses.
What I'd like would be a 17 or 18 Elmarit asph
Anyway Guy, how do you like the Summilux 35 asph ?
Do you prefer it to the Summicron 28 ?
brainiac wrote:
It is very important, as Pondria has demonstrated, that we don't look at simple 100% crops, but that instead we shoot the same angle of view and then uprez the files to match each other's pixel rates.
You're right Richard,
indeed this is what I did when I compared my R-D1 with the first M8 I got (defective unit btw).
I shot the R-D1 with the 21 asph and the M8 with the 24 asph, same effective FOV, both at f/8, then I upsampled the R-D1 file to the M8 dimensions and I made a gif animation of a central 100% crop.
The difference already visible at the original size became much more evident once I upresized the R-D1.
indeed this is what I did when I compared my R-D1 with the first M8 I got (defective unit btw). I shot the R-D1 with the 21 asph and the M8 with the 24 asph, same effective FOV, both at f/8, then I upsampled the R-D1 file to the M8 dimensions and I made a gif animation of a central 100% crop.
The difference already visible at the original size became much more evident once I upresized the R-D1.
P.S[/url]. Obviously I worked with Raw files, not jpegs...
Another way to compensate for the crop factor diff is to use the same lens but changing the distance so that the frame will be the same. That's what I did.
Theoretically one can argue that the lens perform differently from different distance but it may be less different than using different lenses. Just my two cents ... sitting on the couch
That's interesting Marco. .gif would seem to be a bad format for such a comparison though, because, as I understand it, unlike jpeg it has a VERY limited colour pallet. Correct me if I'm wrong. It would be great if you could supply me with two .png files with no compression, or even some other lossless file format, as I am pretty fluent in javascript and I can put up a rollover for you which would serve the same purpose as the .gif
