Great test Marco. Looks to be well controlled and very consistent with my own 1DS m8 comparison using the 24L and 21ZM. I bet if you sharpened the M8 a bit more it would look sharper than the 1DS but that can go back and forth and they do seem to be too close to give one or the other the edge. I preferred the files of the 1DS over the 1DS2 so the 1DS is my personal reference. Also, your observations on the DR are very consistent with my own tests. Nothing like a repeatable test to add merit to a comparison. Thanks for sharing.
Marco wrote:
Ok guys, this is the "pixelpeeping test" with one of the latest M8 (firmware 1.09).
I compared it to my old trusty Eos-1Ds (mkI 11.1 Mp).
M8 with Apo-Summicron 75/2 @ f/6.3 - 160 ISO - 4sec
1Ds with Apo-Macro Elmarit 100/2.8 @ f/8 - 100 ISO - 8sec
Tripod mounted, mirror lock-up on the 1Ds.
Natural late afternoon light from a window on the left.
I converted the raw files in C1, Eos-1Ds with the standard 1Ds daylight profile and M8 with Jamie Robert's Leica_m8_JHR_v1 profile.
There was no IR filter on the Leica.
I prefer the look of the Leica, though the colors are 50/50 with the Canon, some are correct on the 1Ds, other are correct on the Leica... anyway, giving we all know that we need a correct profile for the M8, the result is better than I expected.
Here are the 100% crops, I sharpened the 1Ds more in C1 (40/0 in soft sharpening) to compensate for the lack of AA filter.
I equally slightly sharpened both files in PS to my taste, to give the best sharpness without halos.
Pbase compression steals something of the original crispness of both files.
The 1Ds has maybe a very slight edge, but only because of more pixels, effective resolution is on par, indeed I don't seem to find words or logos on the 1Ds file that I can't recognize on the M8.
The M8 has more "crispness", but again it's not something you couldn't obtain with a bit of postprocessing on the 1Ds.
For all practical purposes, resolution wise both cameras are equal, which says something for the lower Mp M8 but also for the 4 years old 1Ds
Admittedly also for the 75 Apo, which on the M8 needs to work at 1.33x the aerial resolution.
The M8 has other advantages, like slightly higher dynamic range and particularly a great shadows recovery ability. The 1Ds has much more noise in the shadows and I need to get spot on exposures, possibly bracketing and blending in PS to get what the M8 delivers in one shot.
All in all I'm happy to have 1Ds' quality (I'd say even a bit more) in a small package with those tiny outstanding lenses.
And I can go to at least 2 speed stops down with the M8, sort of integrated IS...
But don't think that as pure resolution the M8 could outresolve this old Canon beast, at least when used with similar quality lenses.
As for the 5D and 1DsII, I can't speak of, but I guess you can make your own assumptions
I agree with Jack that shooting Jpegs was not the best choice for a comparison test especially considering the M8's early firmware. For someones personal workflow the choice to shoot Jpeg may make perfect sense but not for an objective test.
Jack Flesher wrote:
Okay, one time response only...
1) you used in-camera jpegs for your test and many pro-level bodies have minimal jpeg processing engines in them and 2) you uprezzed the M8 jpeg to match the 5D and uprezzing jpegs is known to ADD/ENHANCE artifacts.
So even assuming you did focus correctly to begin with -- and I mean perfect focus, not just "within the stated hyperfocal" -- which I feel is doubtful by the looks of it, the best claim *you* can make is:
When comparing an in-camera jpeg from a 5D to an uprezzed in-camera jpeg from an M8, it is my (Brainiac's) opinion the 5D file looks better.
And if you said that, I'd probably even agree with you
I'm still hoping to see more shots from the weekend. Well... unless it was a Brokeback Mountain thing. (in which case I suggest posting those shots in black and white on the Leica Users forum)
OK, here is what I think.
If it is so difficult to compare properly, doesn't it actually mean that the two cameras are pretty much the same in IQ in all "Practical" matters ? Can we just settle at that and move on ?
We Pedantics may want to continue but the rest of the people don't need to stay with us
Pondria wrote:
OK, here is what I think.
If it is so difficult to compare properly, doesn't it actually mean that the two cameras are pretty much the same in IQ in all "Practical" matters ?
Again, I disagree with your base asumption -- it's not difficult to compare cameras properly, so I would answer, "No!!!"
All that is required is that the tester is willing to take the time required to:
1) Properly capture, focus and expose the test images; 2) properly process raw captures to their optimal potential (this can take several days to figure out) and 3) properly convert the optimised files to high-quality jpegs for web view. 4) And even if all that is done, the best web jpeg is still not going to be as telling as a print, so I would add that remarks from an experienced printer comparing prints of those files would be relevant too.
What is difficult is finding somebody who can do all that well and more often than not, it requires a group of people with specific skill sets working together to get it done right... The problem is many single "testers" are simply not willing to put in that kind of prep time, so their results then become nothing more than a set of quick and dirty comparison snap shots -- which is fine if you want to compare snap-shot output, just doesn't work for a quality comparison.
Then add in the few "armchair" testers, who make claims based on their own brand of "logic" without ever having even handled the camera or looked at a raw file! These are the same folks who scream for proof whenever somebody actually using the camera says the result did not agree with their assumption; this last group are the ones that really drive me nuts...
OK, then let's continue testing. For the record, I didn't ask for the continuation
Here is the dead-lock situation I am seeing.
1. There are fools. Whenever a fool bothers to spend his time to post the test result, He is criticised for faulty method. The fool doesn't know what he is talking about.
2. There are saints. What they say shall be believed in. The Saints know the truth. But they won't show the proof because they don't want to be challenged or argued by fools.
KJbruin wrote:
Great test Marco. Looks to be well controlled and very consistent with my own 1DS m8 comparison using the 24L and 21ZM. I bet if you sharpened the M8 a bit more it would look sharper than the 1DS but that can go back and forth and they do seem to be too close to give one or the other the edge. I preferred the files of the 1DS over the 1DS2 so the 1DS is my personal reference. Also, your observations on the DR are very consistent with my own tests. Nothing like a repeatable test to add merit to a comparison. Thanks for sharing.
the 1Ds is indeed a great camera.
I never felt the need to get the markII, as I usually don't shoot above 400 ISO and considering the cost of "upgrading" I would have preferred to get the 5D as both a backup and high ISO camera.
But then this Leica madness came in
My DMR should arrive this week, so the 1Ds will probably go before the end of the year... oh well... I'm still curious to see the next 1DsIII... maybe I could go back to Canon (bodies) again
Well I'm someone who is interested in buying the M8. I use film M and have some lenses. So I'm very interested in what Guy has to say and especially the pictures (which look great). Also though I'm interested in seeing pictures from the DMR and 5D (and DS1 for that matter) since getting user output from these cameras and reading what those using them have to say about quality and what they do to get it gives me an idea of how these cameras might work for me. I can't get much 'hands-on' experience where I live so the vicarious learning is important to me and I would guess many others. I hope that others like Jack Flesher will continue to post therefore. I agree the quick 'tests' aren't tests - they're user reports under particular circumstances meeting the needs of the user - interesting pieces of information when the user is clear about their purpose and the parameters surrounding them so I can weight them accordingly. So come on guys you're doing great work for those like me still to decide on the camera.
Guy, I should have known earlier so I would not have to buy the 24 as a new lens.
BUt since 24 is the widest which works with the internal viewfinder I would not sell it.
> not so sure uprezing is even really need just blow them to 100 percent and crop out a 1000 pixel wide section of course the advantage would go to the 5D
...now we know why the DMR is better than the 1Ds2.
> you uprezzed the M8 jpeg to match the 5D and uprezzing jpegs is known to ADD/ENHANCE artifacts.
This is false. As explained in the test description, the M8 file was precisely doubled in linear size using nearest neighbour so that the file is logically identical to the original. If you don't believe that this method produces an identical file, please try it and look at pixel values. Each 4 pixel block is identical in value to the one pixel it came from therefore the picture is identical and nothing has been added or enhanced.
> ...and I mean perfect focus, not just "within the stated hyperfocal"
the hyperfocal range is defined as the range within which perfect focus and non-perfect focus are perfectly indistinguishable due to the limits of film/camera resolution, so the two are perfectly equivalent.
> the best web jpeg is still not going to be as telling as a print
that depends on...
the quality of the printer
the size of the print
the quality of the viewer's monitor (including calibration)
the ability of the viewer to analyse numerical data, for instance RGB levels in photoshop's info pallet
> When comparing an in-camera jpeg from a 5D to an uprezzed in-camera jpeg from an M8, it is my (Brainiac's) opinion the 5D file looks better.
...should have read:
when comparing an uprezzed with bicubic JPEG from a 5D to a quadruple size but identical to original JPEG from an M8, Brainiac's files look like this ( http://cyberphotographer.com/m8v5d ) because he has kindly shared them because he is such a nice guy ;-)
Edited by brainiac on Dec 12, 2006 at 01:03 PM GMT
... sell all but CV 15, 28 Summicron, 50 Summilux asph, then buy the 90 Apo-Summicron.
You'll end with a nice, light setup which covers almost anything.
I don't know about you, but I don't like to have hundreds of choices, the only exception being landscapes, where a zoom with DLSR vision would be more useful anyway... ask Robert
Cheers
P.S. I enjoyed so much your Yosemite post over at LUF, too much fun, thanks
> How about a sagittal slice of the brain using a really expensive imaging device? No need to worry about skin tones there, just gray-white differentiation.
You may find more black than gray and white. Maybe in a week or two - I am still recovering from a recent full bottle infrontofme.
