1DSMK2/1DMK3 image crops (reposted)

OK first image is from 1DSMK2 second is from 1DMK3.
No post processing & opened in ACR. MK3 had sharpening turned on (low range)
100% crops.

Cheers,

Paul

An interesting comparison.

Is it safe to assume you used exactly the same lens? I can see from the EXIF data that the focal length was different between the two shots although a 70-200 was used for both.

IMO a somewhat better test would be to use the same lens "and" focal length, then resize the larger file down and the smaller file up such that both files are at the average size (in pixels) between the two original files, then sharpen the resultant files and take a look. This process is what I refer to as normalization (both original images undergo interpolation, one up and the second down).

It is entirely possible the difference between the two posted pictures is due to resolution differences of your 70-200 lens at the two chosen focal lengths. What do you think?

Regards,

Joe Kurkjian

well I think that what I am seeing is a 1DMK3 file that looks equal to a 1DS2 file.
The difference in the focal length I guess is due to changing lens, model moving, going downstairs, looking at image, coming back, reshooting. I am sure you get what I mean.
Anyhow, there should be a more obvious difference & if there isn't, then that confirms my thoughts on the remarkable quality of the 1DMK3.

Cheers,

PB

Sorry, I forgot to mention one thing.

Paul, you wrote:
"well I think that what I am seeing is a 1DMK3 file that looks equal to a 1DS2 file."

Actually, to my eyes the m3 file looks considerably better (sharper) and I'm not sure that is a real possibility; I mean it could actually be "considerably" better but I doubt it (the proof is always in the final print).

Talk about prints, that is why I suggested the "normalization" approach because it is the dual of comparing "equal sized prints" of pictures taken from two different cameras. To me equal sized prints is the bottom line and I'm sure it is for you because you actually sell prints for a living. I'm an amateur and have no more room in the house for prints but still believe in the sanctity of a print; in the Kurkjian house-hold, if prints are a measure of "keeper rate" then I'm down to about 1/2 percent (pathetic, I know).

I omitted an important item in my first reply. Because the object is three dimensional, when shooting at the same focal length the 1Ds mkII should be shot stopped down by about 1 stop to achieve the same DOF as the 1D mkIII shot. Sorry about that, I should have said something in my original post but I just slipped up.

BTW, is it possible for you to host the RAW files of the two files you have posted? I would sure like to try a few PP options on the pics but if you can't host the RAW files it isn't a big deal as I would really prefer to see the "same focal length, foot zoom" option shot to eliminate the resolution variable associated with the different focal lengths.

Regards,

Joe

Hi Joseph,

thanks for you input. You make some interesting observations. I did put the raw files on my server but when I checked the view it came up blank. Looks like I can't do it.
I am going to post some more shots & I will keep in mind what you have said.
You mentioned that the MK3 looks sharper. I actually inadvertently had the sharpening set in the camera (at a low level)
The interesting thing I am seeing is that I would have thought the 1DSMK2 image would have left the MK3 in the dust. I am not seeing that. You would think then, pixel for pixel that the MK3 is the superior camera ?

Cheers,

Paul

I agree with you ...

regarding the "1Ds mkII should be noticeably sharper than the 1D mkIII". Well, let us see what we have once you get the same sharpening, or lack of it, applied to both images among other refinements. I'm looking forward to more interesting comparisons from you. BTW, do you also have a 1DmkII or any chance of borrowing one from a friend?

I have another question but feel your answer is a "no" because I know the genre of your specialty. Anyway, here goes, do you happen to own a 300 f/2.8, 400 f/2.8, or 500 f/4 by any chance?

OBTW, thanks for the open mind; it is refreshing to find one every once in a while.

Regards,

Joe

Joe,

I am not in a position to borrow a camera. With regard to the lens, I don't have any of those but I am soon to purchase the 300/2.8 is.
I have some ideas about the direction I would like to take & the 300 is part of my plan.

Cheers,

Paul

To be honest, the images look about the same to me. Minor differences that wouldn't really be a deciding factor either way for me.

Steve that is the point. The Mk3 @10MP produces images as good as (or better) than the 1DSMK2 @16.7 MP.
At least that is what I am trying to determine. Your claim that they look the same is what I am alluding to.

Cheers,

Paul

Hi Paul,

It is difficult to tell, without the raw files, so much is lost in compression, and other effects. The first image, in some ways could be said to be 'better', the lighting is slightly different - look at the reflections in the eyes, the difference in contrast in the right ear/cheek. This outweighs any other differences that may be present. However, if your aim is the final print quality, then from what you are indicating, the camera is fine. Just think, within a few years time, you'll be repeating these test's, with a still better camera .....

Best wishes,

Ray

Paul, regarding your sample pictures.

I've been doing a lot of thinking (I know, sounds dangerous) about your two comparison shots and the ONLY way I can envision the m3 keeping up with a 1DsII from the standpoint of sharpness is "if" the AA filter on the m3 is "relatively" weaker than the AA design used in the m2 (N or non-N). I use the term "relatively" because there are more pixels in the m3 compared to the m2.

Anyway, I was wondering if you could possibly shoot some feathers to see if the m3 has any tendency towards the creation of aliasing artifacts. For example, this is a shot feathers to give you an idea of the spatial frequency that I'd be interested in seeing at 100 percent. There are actually three 100 percent crops in the picture below because I was comparing three RAW converters (RSP, LR, and Bibble).



Maybe you have a feather duster in your studio that could serve as an object? Or, you could visit a local stripper club as there is a high probably that feathers will be used as a prop (no pun intended) by one of the ladies. Anyway, thanks in advance for giving this request some thought, I appreciate it.

Regards,

Joe Kurkjian

thanks for the mission Joe. I will get on to it.

Cheers,

Paul

Paul Bestwick said:

thanks for the mission Joe. I will get on to it.

Cheers,

Paul

Click to expand...

Thanks Paul, I appreciate your taking it on.

Joe

Joseph A. Kurkjian said:

Maybe you have a feather duster in your studio that could serve as an object?

Click to expand...

If you want to totally avoid any bias in an aliasing comparison between different birds of a feather, or rather feathers of different birds then the following may be a useful test object to print and photograph:
For 600 PPI printers (3.8MB), and
for 720 PPI printers (5.3MB).

When printed on glossy paper at the indicated PPI setting without adjustments (you can assign a printer profile if that's available), the final target will measure 100x100mm and it allows to judge the limiting resolution of the entire imaging system, and the aliasing artifacts near that limit. Just shoot it from a distance of 25 - 50x the focal length, although the exact distance is not critical. That is one of the nice features, the limiting resolution will always produce the same central blur diameter regardless of shooting distance.

Bart

Hi Bart,

OK, lets get this right. I am the most unscientific tester on the planet........ OK
Unfortunately I have no gloss paper & these shots are straight off the screen.
I was definitely surprised at what I saw on the screen, the MK3 easily won. & the MK2 had a magenta cast.
1DMK3 is first image

Deleted image

Paul Bestwick said:

Hi Bart,

OK, lets get this right. I am the most unscientific tester on the planet........ OK
Unfortunately I have no gloss paper & these shots are straight off the screen.
I was definitely surprised at what I saw on the screen, the MK3 easily won. & the MK2 had a magenta cast.
1DMK3 is first image

Click to expand...

Paul, thanks for taking time to take the shots; I just have a few comments. First, the 1DsII shot is slightly OOF; this would minimize aliasing errors. Second, the shutter speed has to be much lower than the frame rate of the monitor. Third, I don't know how close to the monitor screen was to the sensor plane but it is possible the LCD or if a CRT was used the screen resolution was a factor (I don't know for sure).

If you want to use this as a target I'd suggest getting the heaviest studio tripod you have, stop down to f/11, shoot at ISO-100, and stand back far enough so that the monitor screen, with the target displayed at 100 percent magnification, fills about half the camera's FOV.

Joe

Images have been updated. Printed on fine art photo rag (Matt)
Left my tripod at had so just had camera propped on chair.

PB

Paul Bestwick said:

Images have been updated. Printed on fine art photo rag (Matt)
Left my tripod at had so just had camera propped on chair.

PB

Click to expand...

Thanks Paul, nice of you to take the additional time.

The new (latest) images look "very" equal with a slight edge in sharpness going to the 1DmkIII. I don't see any problems with the m3 image, thanks again for all the work. Having seen your test shots I have to say your m3 is an outstanding camera, compared to anything including the 1DsII (I'm impressed).

What are your opinions regarding the printed images? Do they look fairly identical?

Regards,

Joe Kurkjian

Paul Bestwick said:

Hi Bart,

OK, lets get this right. I am the most unscientific tester on the planet........ OK
Unfortunately I have no gloss paper & these shots are straight off the screen.

Click to expand...

Paul, you should have shot it from a larger distance. When printed at the indicated PPI setting, the resolution of the target can reach something like 10 cycles/mm (roughly 10 line pairs/mm) or a bit better on glossy paper. The shooting distance of 25x focal length (or more) will produce a projection reduction of another 25 times so you're effectively shooting a 250 cycles/mm target, enough to out-challenge your lens as well.

On screen the highest target resolution is just in the order of 2 line pairs, which at 25x FL amounts to 50 cycles/mm, well in range of the sensor+AA filter, so not a big challenge (even less when shot from closer distances). Shooting from screen could work if the shooting distance becomes 125x the focal length distance, which would require a large studio for the longer lenses.

Nevertheless ...

I was definitely surprised at what I saw on the screen, the MK3 easily won. & the MK2 had a magenta cast.

Click to expand...

Yes, while at the lower end of the target becoming really meaningful, it already was revealed that the image (processing) pipeline makes a significant difference (assuming equally well focused images) in the end result. If anything, there seems little wrong with your MK3's focus on stationary subjects ...

Now, whether it is due to the AA-filter being less strict, or a different Raw conversion protocol, or just a focus issue, remains to be seen. For that the target al least needs to be shot from further away, because that will challenge the optical path to its limits, and aliasing becomes clear (the only question is how much clearer). When the aliasing filter's influence (and that of larger microlenses on the sensels) can be factored out of the equation, then the remainder must be due to post-processing.

By the way, I probably should have added a warning for people with epileptic tendencies, don't stare at the full screen image!

Bart

Paul Bestwick said:

Images have been updated. Printed on fine art photo rag (Matt)
Left my tripod at had so just had camera propped on chair.

Click to expand...

Paul, did you use the indicated PPI for the printer/paper? The output should then measure 100x100 millimetres (3.94 inches on each side). From what distance did you shoot it?

The target will also reveal camera shake or mis-aligned optics, so any symmetrical result will indicate a stable shot.

Bart

Hi Bart,

I will reshoot when I am in the studio on Monday.
I can use 100mm 2.8 macro , 24-70 2.8L, 16-35 2.8 ll L, or 70-200 2.8L is.
Take your pick. What distance & F stop for each sensor.
I will be shooting the printed target.
Yes, I printed the target at 720 dpi but 12"x12"
Cheers,

Paul

Paul Bestwick said:

Hi Bart,

I will reshoot when I am in the studio on Monday.

Click to expand...

Appreciated, especially because it's not a subject that you commonly shoot. The results may also be educational for others, and serve as a reference for others doing their own test with their own camera/lens combo.

I can use 100mm 2.8 macro , 24-70 2.8L, 16-35 2.8 ll L, or 70-200 2.8L is.
Take your pick. What distance & F stop for each sensor.

Click to expand...

You can test with whichever lens you want (and 'digital film' is free to use), but the 100mm f/2.8 will certainly not be the lowest common denominator in the imaging chain and it might show AA-filter/sensor trade-offs best. The fixed focal length will also avoid minute FL differences when switching between camera bodies, and that lens is part of many people's lens line-up. When you use the 100mm lens, shooting the target from between 2.5 metres and 5 metres should do the trick with a 100x100mm target. When you print on non-glossy paper, I'd go for the longer distance if possible. Tests with my copy of that lens on my 7.2 micron sensel pitch camera (1Ds Mark II) shows peak performance at f/9.0, so I'd expect the same on the 1D Mark III (also 7.2 micron pitch). Using the same f/9.0 for both cameras seems a good choice, although f/8.0 is not much different and might be a more prudent choice for the yet unknown AA-filter strength of the Mark III.

I will be shooting the printed target.
Yes, I printed the target at 720 dpi but 12"x12"

Click to expand...

I figured something like that might have happened. It means the target was interpolated by the printer driver, and that interpolation quality became an added variable of unknown magnitude in the imaging chain. When the original file is printed without interpolation at the native printer driver resolution (720PPI with most Epsons), the printed result will measure 100x100mm. It will also result in a smaller crop of the full target in the resulting file to post, and it's less sensitive to accidental non-perpendicular shooting (more likely to be within the DOF range).

Bart

Paul Bestwick said:

The interesting thing I am seeing is that I would have thought the 1DSMK2 image would have left the MK3 in the dust. I am not seeing that. You would think then, pixel for pixel that the MK3 is the superior camera ?

Click to expand...

Paul,
First, I'm new to the Group, and I'm a friend of Joe... Please don't hold that against me

Now... How dare you say such things against the most Sacred of Cows in the Canon world!!

I dared to challenge the Full Frame cameras on another Forum and I was immediately flogged and challenged myself on how could i be a real Professional if I didn't shoot with a Full Frame camera!!

I can see exactly what Joe saw, the Image coming from the MKIII is indeed better than the DsMKII.

I will be getting a DsMKII on loan from CPS in September, I'm taking it to the Handgun National Championships, I want to see how it does on some of the Commercial Shots I'll be doing there this year. But I have a feeling that the MKIII should be a better bet, I sincerely doubt that CPS is going to have loaners anytime soon though

Nice images though!!

Y