I still remain somewhat confused by the consuming passion of people trying to prove the 5D is superior to the M8 (or vice versa). These are two entirely different camera systems used for diffferent purposes and for different shooting styles. Each has strengths and weaknesses. It's not unlike trying to compare a view camera to a 35mm SLR.
The price point argument (but the 5D is X thousand dollards cheaper) is silly as well. If I want a world class rangefinder I can either buy a Leica or a competitor. If I buy a new film based Leica again, it would cost me $3000 USD or so. The extra cost of the M8 corresponds roughly to 30-45 rolls of color film + processing (depending on film type and processing). This seems like a reasonable cost differential if one wants to have a digital rangefinder.
If you insist on having a full frame camera, can you rationalize the cost of the 5D over the cost of a EOS-3? Wait a minute, that's a whopping $2000. How much does 30-45 rolls of color film cost again... Hmm about $2000. That's interesting. The cost differential from film to digital for a pro-quality Leica rangefinder is the same as the cost difference for a pro-quality Canon SLR....
Pondria wrote:
When properly profiled and/or calibrated, a camera should generate identically matching values for standard color patches, for instance, like Macbeth chart. Be it 5D or DMR, it should yield the same Target RGB values for all 24 color patches. If NOT, it is producing wrong colors.
Just going to say that I'm selling my Summicron-M 35mm f/2 chrome 8-element. This is the first and collectable Summicron. It's in very good condition and comes with front and back lens caps and a Hoya filter. Glass is perfect and I recently had the lens serviced. There's some small marks on the barrel but apart from that very good condition. Just wanted you to know before selling on ebay or on commision. I prefer shipping within Europe but international is possible.
Just going to say that I'm selling my Summicron-M 35mm f/2 chrome 8-element. This is the first and collectable Summicron. It's in very good condition and comes with front and back lens caps and a Hoya filter. Glass is perfect and I recently had the lens serviced. There's some small marks on the barrel but apart from that very good condition. Just wanted you to know before selling on ebay or on commision. I prefer shipping within Europe but international is possible.
And to be pedantic, devices have metameric failures.
Adequate profiling is a starting point, but does not create a perfect camera. Camera profiles are created under carefully controlled, or at least constant, lighting conditions. They do not provide a unique solution to the multivariable lighting occuring in the real world. The accurate recreation of an image requires a match to the spectral sensitivity of the sensor to that of the image. The profile attempts to do this, but it is only going to provide the match if the lighting conditions are identical.
Edited by braindeadmac on Dec 19, 2006 at 03:30 PM GMT
Jeff, the point is not how different the systems are, but if they are competitive for a given type of job. For some jobs, a 5D and an M8 could easily both be usable, for others not. For those jobs where they are both usable, it is interesting to compare characteristics. I am selling my 5D for an M8, so for me it is particularly pertinent. I am not out to prove that the M8 is better, but I am trying to prove that the M8 is at least roughly equally good across disciplines, and I think that this is the case. More to come...
ClubShooter wrote:
But it's not the distance to the lens that determines the angle of incidence, but the position of the rear nodal point. That's where the light cones cross. The further back the rear nodal point is, the steeper the angle of incidence. How this relates to the position of the exit pupil is simply a matter of design. Your 16-35 is a retrofocus, meaning the rear nodal point is behind the exit pupil, this is how it can clear the mirror. It may still have an angle of incidence as steep as, or even steeper, than a CV 15, which has its rear nodal point either at the exit pupil or inside the lens. ...Show more → If the nodal point was between the last element and the sensor it would mean that the image inverted at some point between the rear element and the sensor - a quick test with a light source and a piece of white paper shows this not to be the case.
braindeadmac wrote:
Brainiac is correct on both points. Aperture has no affect on light collimation, and angle of view has little (negligible) affect.
We must be talking about two different things then. Clearly a light cone from further out in the field is going to arrive at a different angle than one near the center. So they are not mutually collimated (parallel). In fact, if they were collimated we wouldn't have the cyan vignette on the M8 with wide angles and 486 filters.
Wikipedia seems to support my position: http://en.wikipedia.org/wiki/Collimation
It says point sources like stars are effectively collimated, while something as wide as the sun is only approximately collimated. If the 1/2-degree sun isn't collimated except in a liberal sense, then how can say a 75-degree field be?
shirozina wrote:
If the nodal point was between the last element and the sensor it would mean that the image inverted at some point between the rear element and the sensor - a quick test with a light source and a piece of white paper shows this not to be the case.
Why do you think a nodal point behind the lens inverts an image while one inside the lens doesn't? They'll have the same effect. Just different placement.
fish_shooter wrote:
You missed the point, which is the additive effect of light attenuation. Maybe combining two ND filters might have been a better example.
You talked about #25, which attenuates by cutting a portion of the spectrum. This is very different from an ND filter which attenuates by limiting transmittance even across. It's hard to get a point which makes no sense. IR cut filters are closer to the former than the latter, and stacking them won't have nearly the effect you think it might.
shirozina wrote:
If the nodal point was between the last element and the sensor it would mean that the image inverted at some point between the rear element and the sensor - a quick test with a light source and a piece of white paper shows this not to be the case.
Sorry guys, but you both are somewhat wrong. The nodal point is indeed located between the last element and the sensor. Actually, the distance from it to the sensor is exactly the focal length. But the nodal point is more a mathematical concept then the actual point concentrating all ray forming an image. Depending on the lens design there might be no single ray coming through it to the sensor edge. A good illustration of the incidence angles for retrofocus and symmetrical lenses is shown in http://www.pinnipedia.org/optics/vignetting.html . Sure thing, the retrofocus lenses produce less steep angles (and less vignetting because of this).
ClubShooter wrote:
Why do you think a nodal point behind the lens inverts an image while one inside the lens doesn't? They'll have the same effect. Just different placement. Obviously the nodal point inverts the image - I'm saying that if it's inverting it outside the lens group you would be able to physicaly observe this happening by looking at the projected image on some paper as you moved it between the rear element and the distance of the focal plane - I've done it and it does not happen therefore the nodal point is not outside the lensgroup.
My use of the term nodal point is obviously wrong - I mean a point which is one end of the cone of projection or where the light cones intersect ( whatever that is called).
I just read through the posts since I last logged in (several pages!). Thanks to those who had a go at the sensor filter issue, I still don't know whether my interpretation that sensor filter thickness affects focus differentially towards the edges is right. Although the discussion of Canon wides was interesting one reason for pursuing the issue was the thought that given the steep angles of incidence with the Leica WA lenses if focus was affected by a thick sensor filter it could explain why they went with a thin filter.
shirozina wrote:
My use of the term nodal point is obviously wrong - I mean a point which is one end of the cone of projection or where the light cones intersect ( whatever that is called).
You probably mean exit pupil, which is the apparent position of the aperture viewed from the rear of the lens.
Tom
ClubShooter wrote:
You talked about #25, which attenuates by cutting a portion of the spectrum. This is very different from an ND filter which attenuates by limiting transmittance even across. It's hard to get a point which makes no sense. IR cut filters are closer to the former than the latter, and stacking them won't have nearly the effect you think it might.
How do you explain a thinner one transmitting more IR? (we are speaking absorption filters here, not interference filters)
Tom
edit: Also see Alex's post on the next page of this thread
Edited by fish_shooter on Dec 20, 2006 at 09:50 AM GMT
I maintain the 5D is mushy in the corners with extreme wide angles because of the design of the sensor.
I will retract my statement. My revised statement is: "The 5D is musy in the corners with extreme wide angles." :)
I think this is incorrect. The examples of mushy corners from my beach shots that you posted are at f2.8. Yes they are mushy at f2.8. By f5.6 or f8 they are starting to behave better. The brighter sunlight shot illustrates this. The lens is far from perfect. But the f8 performance shows that the mushiness is the lens rather than the sensor. The 5D sensor corners are capable of showing as much sharpness as a lens can supply. Look at the bottom right corner here: http://cyberphotographer.com/p/20061209pc/20061128_153749_highrez.jpg
> ...I suspect Richard has had more optical physics than either of us judging from his college degree.
Don't pay any attention to that. I am learning a lot from all of you guys, and it is a pleasure. I enjoy what everyone has to say, and I read it carefully. The laws of physics as learned in a college degree often don't predict the counterintuitive outcomes that actual results show. Any camera is a complex system. A year and a half ago I was still doubtful that digital would surpass film. Now I understand why my prejudices were wrong.
brainiac wrote:
I think this is incorrect. The examples of mushy corners from my beach shots that you posted are at f2.8. Yes they are mushy at f2.8. By f5.6 or f8 they are starting to behave better. The brighter sunlight shot illustrates this. The lens is far from perfect. But the f8 performance shows that the mushiness is the lens rather than the sensor. The 5D sensor corners are capable of showing as much sharpness as a lens can supply. Look at the bottom right corner here: http://cyberphotographer.com/p/20061209pc/20061128_153749_highrez.jpg
That one is better the sample I used had f1024 or something like that in the exif.
If you go right into the bottom right corner of the shot you linked above, past even the little lump of sand, it softens up quite a bit. The sample from the M8 showd sharp edges on frost crystals, right into the corner of the frame. It might be just the limitations of the lens, not the sensor.
The sample from the M8 showd sharp edges on frost crystals, right into the corner of the frame. It might be just the limitations of the lens, not the sensor.
Yes - judging by 5D corner performance with longer lenses, the 5D sensor behaves no differently at the corners than it does in the center. It is certainly hard to find a very wide lens that satisfies the 5D's corners. I try with a Zeiss 21 and Nikkor 17-35. Results are great, but if you want every pixel to count in the corners you will need to stop both of these lenses down to f8 or f11. It is worth noting that this is a fairly marginal issue, since the resolution throughout the rest of the frame is excellent - probably greater than the M8 can support, according to my tests. In other words, if (if) the M8 is sharper in the 2% of the frame that is the extreme corners, it probably resolves less in the 98% of the frame that is not in those extreme corners.
