How to conduct a megapixel test???

[iCe]

I want to see if there is any noticeable and usable difference in images taken with (3) different cameras. Basically it's going to be a test to see if megapixels matter. Any suggestions for how to proceed?

In my mind lighting is going to be the most important factor with camera position and lens being a close second. I was thinking of setting up some kind of studio shot with strobes, and an inanimate object for the subject. The lens would be attached to the tripod (probably use the 70-200 f2.8L IS) and I would swap bodies out and shoot once everything was set and locked down. The camera's would be in M. The file type would be RAW (won't shoot anything else). Am I missing anything in the setup?

Once setup is nailed down what would be a good way to analyze the images? Printing is an option but I'm going to try to avoid that since there will likely have to be some post processing to make an acceptable print. If nothing else a conversion to JPG and an unsharp mask.

Three bodies. 12mp, 16.7mp, 21mp

Thoughts?

[jeffkohn]

I'm not so sure what's key about lighting, aside from making it consistent from one exposure to the next.

If the cameras have different sized sensors (are you comparing the 5D, or some other 12mp camera?), then keep the FOV constant, not the focal length.

The best comparison (in fact I'd say the only one that truly matters), is comparing prints. After all, the real reason to get more MP is to make bigger and more detailed prints. I would suggest inkjet prints from a quality printer, the bigger the better.

Trying to compare images with different resolutions on-screen is an exercise in futility, you either have to up-res or down-rez some of the images which adds uncertainty to the results.

For 21mp I would use the best, sharpest lens you have if you really want to compare sensors without worrying about the lens being the limiting factor.

[iCe]

I'm not so sure what's key about lighting, aside from making it consistent from one exposure to the next.

I'm not so sure what's not key about lighting when you're testing a camera. If I tried a test on wildlife, for instance, there is no way I could get consistent lighting. Lighting affects contrast, image depth, all kinds of things (but I know you know that).

Cameras- 1DMKIII, 1DsMKII, 1DsMKIII

If we didn't live in a computer society I would agree that the final test is a print. It would be interesting to know how many digital images are ever really printed. The ratio of printed to not printed is very small imho. I'm not sure what I'm trying to figure out. I got the idea to do the test after shooting a football game (night) with my MKIII and MKII. I much preferred the MKII images over the MKIII images. It made me wonder if the difference in mp's was what I was seeing. The MKII images seemed to have more depth. Anyway... it may not even be worth doing...

Good point on the lens.

Thanks

[Jim Victory]

Originally Posted by iCe

I got the idea to do the test after shooting a football game (night) with my MKIII and MKII. I much preferred the MKII images over the MKIII images. It made me wonder if the difference in mp's was what I was seeing. The MKII images seemed to have more depth. Anyway... it may not even be worth doing...

Good point on the lens.

Thanks

Were you referring to the 1DsMKII and MKIII? If so your talking a difference in FOV that may be giving that impression as well as more MP.

As pointed out the only way to compare properly is the difference in resolution your will see in large prints. Trying to compare a cropped camera with a FF camera is always difficult and also a problem even looking at 100% crops.

If you have a certain size print, you usually get, then print them all up in that size and see if there is any difference. My guess you would have a hard time telling the difference between those three cameras up to 13x19 print.

[jeffkohn]

If we didn't live in a computer society I would agree that the final test is a print. It would be interesting to know how many digital images are ever really printed. The ratio of printed to not printed is very small imho.

That's probably true for some. I certainly don't print all of my images, but I do print a lot of my "keepers". In fact I have boxes of prints I've made of images just because I wanted to see what they look like as prints (others are on display at home or at my office). I get much more satisfaction from looking at a good print than I do an image on a computer screen.

But if a computer display is the end destination, even 12mp should be overkill. I can't think of any reason to want more MP except to print larger. I suppose an argument could be made for cropping headroom, but if 12mp doesn't yield enough resolution to crop for the web I think the $8K would be better spent on longer glass rather than a 1DsMk3.

BTW since one of the cameras you want to compare has a cropped sensor, keep the FOV constant, not the focal length.

[johnastovall]

You might want to look at the methodology used in "The Great Luminous Landscape
2006 State-of-The-Art Shootout."

[venchka]

20x24 real photographic prints from a high end print place like West Coast Imaging.

Look at the prints from 3' to 5' away. If there's a difference, pick the body you like best.

You could also make reference prints from the Hasselblad and Linhof. However, you might find out that high end digital really can hold up against 6x6 or 4x5 film. Or not. It would be interesting to see that kind of side by side comparison.

[WLPelzmann]

Michael, if you decide to make some prints, I'd sure be interested in seeing them.

If you want to compare large print sizes, but don't want the cost, you could upscale the images to a standard test size - let's say 30"x40", and then just print 8"x10" crops from the center &/or corners.

I agree with Jeff. If I didn't print big and just viewed images on a computer screen, a 12 MP sensor would be overkill. I've still got some 12x18 prints on my wall cropped from a 6MP Canon10D that look great.

Originally Posted by iCe

...... Printing is an option but I'm going to try to avoid that since there will likely have to be some post processing to make an acceptable print.....

Well, I would hope there would be differences in the post processing, even if you don't print. Each of those cameras need/accept different amounts of sharpening. Comparing unprocessed, or equally processed files would not represent their relative capabilities.

[dsjsws]

Michael,

Before you go to all the time and expense of printing. Take the same photo with each camera. Download each and view each at actual pictures, 200% and 400%.
You should be able to figure it out quickly.
It is just a quick, simple, painless and totally inexpensive way before going to great lengths.
Donna

[venchka]

Paraphrasing Tom Clancey: If you don't print it it didn't happen.

Color, tonality, and all the other subjective aspects of a real photograph won't come through on a monitor or 8x10 chunks of the full frame. You have to see the entire photograph to get the whole picture. Every pun intended.

[iCe]

Thanks to all for responding

First- I always have intent to print (but not write poetry or rhyme) so even though I said that I wanted to avoid printing, I meant for the test not life.

John- The Luminous Landscape link was very helpful and pretty much summed it up for me.

Jim / Wayne... You too both rang a bell. West Coast Imaging has a Megapixel Print Size Chart that's pretty handy for determining a recommended print size based on DPI. I included the link. Rather than get into a debate about what is best or acceptable I'll just say that my preference would be to print at 200+ DPI but one doesn't always get their preference BTW- if you use "Better" (100 - 149 DPI) as the criteria the MKIII is "good" up to a 24" x 36" print. I may be oversimplifying this but from my perspective being able to crop and print it large is the primary benefit to having more pixels available. I've read that increased pixel count (for lack of a better term) is also = to increased tonal range. It makes sense but who knows. I thought that may have had something to do with my preference for the 1DsMKII shots over the 1DMKIII shots. The problem with that is, of course, I was comparing shots from different locations on the field, shot at different settings, and shot through different lenses. Not much room for a real comparison there.

Wayne- Pick the body I like the best? Is that possible? In my next life I want to come back as an eight eyeballed octopus-like (and handsome of course) creature from the planet Photographia so I can shoot up to 8 different cameras at the same time. Even Nnn... Nuhhhh.... Niiiii.... the other camera that is starting to catch up... slowly...

Bill... You and everyone else are making me feel bad... I need to print more but the point is taken and I agree, not printing is not a good idea.

Hey Donna... How have you been? I've been doing that... But every time I shot something and looked at it I would start to wonder if the light was the same or if I skewed the result in CS3. I think the majority is correct in that you have to print to be able to see the result. Web images, and therefore images on a monitor, just don't cut it for that.