brainiac wrote:
... It's interesting though that the metal grille on the ground is better resolved on the 1DsII even though everything looks less crisp. This is why I think the crispness of non-anti-aliased images is deceptive. They look crisper because contrast between consecutive pixels is much higher, but they can still be delivering less information.
You put it very concisely and precisely. The resolution is about whether you can distinguish the two black dots close-by. People are usually mix the resolution with fuzziness or crispness of the apparent look.
The fuzziness introduced by the AA filter can be easily corrected with the post-processing. And some people seem to regard this as "cheating". But the same people seem to be OK with S/W correction of Aliasing caused by lack of AA filter. Go figure
I very rarely see aliasing on my SLR/c, which doesn't have an AA filter. On the few occasions it does pop up, it normally affects only a small area of the frame. It is easily corrected in post processing, adding touch of fuzziness to the area . In comparison, an AA filter affects the entire frame (every edge in the frame anyway) and *every* frame captured. I prefer the apparent crispness of output without an AA filter. While you might be able to get make an AA filter shot look similar to a non-AA filter shot, I don't think you can get exactly the same look, especially for very intricate scenes.
Edited by pdmphoto on Nov 12, 2006 at 10:19 PM GMT
The M8 picture looks better to me but that is subjective and not scientific. I've tried the 1DS2 and prefer the images that the older 1DS produces despite it's lower megapixels and poorer high ISO performance. So someone explain that to me please. Perhaps it's the weaker AA filter.
In fairness to the M8 image the Canon one is a bit larger so composition was not exactly the same And look at that Canon glow around the letters. Perhaps it needs a bit more USM.
So all in all I'm looking forward to getting the M8 and using it with filters
brainiac wrote:
Guy - are you talking about the train display-board images that I posted above, and which come from the luminous-landscape review? If you are, I really have to question the idea that the M8 'slamdunks' the 5D. This test suggests the reverse. Remember, this is a 5D at 3200 with undisclosed (?) lens versus the M8 at 1250. hardly a fair test, you might think, and yet, if you look at the following enlargements of those samples, in pictorial information delivered, the 5D is a considerably ahead:
You can correct the softness, but you won't get the full information back. Something gets lost when you put a soft filter in front of a sensor, surprisingly.
carstenw wrote:
Who here other than Pondria still have their F717? I sold my F828 already!
I like F717, really I like 5MB files. I like no need for RAW conversion. I like, Oh Yeah, Zeiss lens. And the real killer feature is the DEEP DOF. It's so nice that you can get very decent DOF even wide open.
Dave, I would be very grateful if you could find a moment to post lowrez versions of _the whole_ frame of each of you doorway shots.
All this talk of how the pixels look is quite misleading. When it comes to making prints, or viewing the file at a resolution where the pixels are invisible, or interpolating upwards, the crisp-with-less-resolution look is gone. You are simply left with less resolution. Look at the keyhole cover in Dave's comparison: the 1DsII shows it has an oval shape. The M8 thinks it's a rectangle. I believe that preferring the M8 shot of the train display is a serious error because the 5D shot has, in real useable data terms, about twice the resolution. The lens quality and post-processing issues are not what we are trying to judge here. Those who think the non-anti-aliased pixel crispness is going to be visible in prints are, in my view, deluded. The higher resolution of the canons, on the other hand, will show up in prints. Are we taking pictures so that we can enjoy looking at small parts of them on a screen at 100%, or are we trying to resolve an image in the truest way, with credible colour, high resolution, and immersive reality. Stand back from the train display pictures and ask which one offers a more convinving reality. To my eye, the M8 at 1250 is resolving about half the reality that the 5D resolves at 3200. It would be great to be able to do the same comparison at iso 160.
Nice pixels are no substitute for ability to convey reality and resolution is an important part of that.
Why you guys compare the M8 with a SLR (5d/1dsii)?
When comparing cameras IMO one would have to compare the 5d/1dsII with the Leica-DMR.
And the only camera IMO to compare the M8 with is the Epson rd1.
Now we have a digital rangefinder (M8) which seems to deliever IQ on the same level as a DMR or a 5D (at least at low/medium ISO).
Is it really important if the M8 does a tiny little better or worse than DSLR counterparts?
bathman wrote:
side question. I read that even digital sensors suffer from sample variation and resolution might be affected by that. How about that Unless sample variation includes some sensors having more photosites than others I would have thought this unlikely. However sample variation is likely to effect high ISO performance re noise.
Resolution - why don't we RESOLVE not to sqabble over resolution until we have seen a proper side by side tests done under something approaching controled conditions? I know there will never be 100% aggreement on what this should be but handheld shots through a window being posted as a comparison are just a waste of bandwidth and server space. It does not take much imagination or understanding to accept that a 10mp sensor without an AA filter is not going to be a million miles away from a 12.Xmp sensor with an AA ( weakish though on the 5D) filter and is going to struggle to make up a more than 50% advantage in pixels of a 16mp 1Ds2.
That's the problem. Lots of people are reporting how good their M8's are, but they are too busy enjoying them to post fair tests against the 5D. Understandable really: a beautiful new camera is always an inspiration to go out and enjoy photography afresh.
Why you guys compare the M8 with a SLR (5d...
The suggestion that these two cameras are not in competition is not true in my case. I shoot parties and weddings. Both cameras are portable 35mm cameras. One is 300 grams heavier, one is $2000 dearer. I would leica lighter camera, but I am not prepared to take a reduction in quality at the cost of $15000 (I have no M lenses and I must have 2 bodies in case of failure). So far, everything I have seen suggests to me that there may be a shortfall in image quality. The absense of fair tests is beginning to look like a reluctance to face this uncomfortable possibility.
It is almost certain that the M8's image quality will be a long way from competitive in 2 years time. Will the M9 be available then? That's why actually achieved image quality is very important to me. I don't want to be stuck producing uncompetitive image quality in a competitive market.
I will have to find a friendly Leica dealer with an M8 in stock and do the test myself. If I do, I promise to share it here ;-)
And if a really much respected member here claims that the M8 has better resolution than other professional systems this needs some investigation of course.
"It is almost certain that the M8's image quality will be a long way from competitive in 2 years time"
There is no way M8 images are not going to be competative in 2 years time - if they are good enough now they are good enough in 2 or 20 years time IMO. If your images are not competative in the market place now then the last thing you can blame at the present state of technology is your equipment. Please don't take this the wrong way but wedding and party photography does not make massive demands on resolution in the same way that architectural or landscape photography would. Or to put it another way if you are shooting landscape and architecture then a 10mp DSLR (or RF) is going to put you at a competative disadvantage in a lot of areas with someone with a 39mp digital system or even 5x4.
BTW - Dave, I don't want to bite your head off at all, and I am grateful to you for sharing your images, but the reason I would like to see the full frames of the doorway shots is that I was curious about the method you used to uprez the M8 shot to cover the same area as the 1Ds2 shot. You seem to have preserved pixel sharpness remarkably well for a 1.63x magnification. If these are actual 100% crops then don't forget that the 1DsII frame will be covering a much larger area. Consequently you are only showing us 60% of the 1Ds2's resolution. In other words, you are effectively showing us a dramatically down-rezzed version of the shot in the Canon example, because the Canon picture has a much wider angle of view.
The best way for you to clarify that you did not make this error, and that you did interpolate the M8 shot so that it covered the same area as the 1Ds2 shot, would be to post the full frames of each shot. Then we can all be clear about what we are looking at.
Let us know, because it's an easy mistake to make, but your post could be very, very misleading, and we wouldn't want someone to spend $5000 unnecessarily.
Weddings move very fast, and every picture is a compromise. My clients sometimes want to make a cropped print of two people in a picture of five people. High resolution seldom counts against the photographer. It gives clients more options. It's the opposite of salt. You can always reduce it, but you can never add more.
And look at that Canon glow 