Daan B wrote:
And the difference between 24mm and 40mm. And maybe a difference in focus. The ISO1600 crops are much less distinguishable when it comes to fine detail...
good points daan. i make no contentions that this is a "definitive" test. just my observations, and very casual ones at that.
should be noted that the iso 1600 were shot with a 10-22 on the 7d, and a 24-105 on the 5d. so again, the image is composed very similarly, but the glass and focal lengths vary significantly, though f8 was the aperture for both...
CarpeyBiggs wrote:
ok, here's a very quick comparison. sorry, i don't have the patience, discipline, or desire to set up some very controlled experiment between the two cameras.....
Honestly, for us "landscapers" this is one of the nicest "quick" comparisons I've seen and a beautiful scene to boot. Thanks for doing it.
Amazing how good the 5d files look for such an "old" camera.
Just out of curiosity - did you uprez any of the 5d files to 7d size. I'd be curious to see that.
I must be dense but if the images were framed the same, how can this be a 1:1 comparison? Surely these can't be 100% crops, so the question is if the 5D was scaled up or the 7D down? Or am I missing the boat completely?
michael49 wrote:
Honestly, for us "landscapers" this is one of the nicest "quick" comparisons I've seen and a beautiful scene to boot. Thanks for doing it.
Amazing how good the 5d files look for such an "old" camera.
Just out of curiosity - did you uprez any of the 5d files to 7d size. I'd be curious to see that.
stanj wrote:
I must be dense but if the images were framed the same, how can this be a 1:1 comparison? Surely these can't be 100% crops, so the question is if the 5D was scaled up or the 7D down? Or am I missing the boat completely?
they are 100 percent crops. 7d at 24mm is roughly the similar FOV to a 5d at 40mm.
CarpeyBiggs wrote:
they are 100 percent crops. 7d at 24mm is roughly the similar FOV to a 5d at 40mm.
Right. FOV != "pixels per duck". So the image (the row of trees) was filling the frame on the 7D just the same as on the 5D. But the 7D has 18M pixels and the 5D has 12M (or something to that sense). So each tree (or other imaged "unit") should be bigger / wider / taller when viewed at 100% on the 7D image than on the 5D image. Yet in the comparison they are sized precisely the same. That's the part I don't get.
though, like i said, this isn't a perfect comparison. i just framed them up by eye, shot a photo, and posted it. it's possible i didn't zoom in far enough on the 5d or something...
stanj wrote:
I must be dense but if the images were framed the same, how can this be a 1:1 comparison? Surely these can't be 100% crops, so the question is if the 5D was scaled up or the 7D down? Or am I missing the boat completely?
They are 100% crops, which is why the 7D images are larger (that camera has more pixels).
The results are about what I would expect, assuming the 7D's sensor is a small but significant improvement on the 50D's. You have to remember the 5D has nearly 2.6x more light to play with. The 5D's sensor is significantly less efficient than today's best sensors (e.g. 50D, 5D Mark II, D700, etc.), but brute force counts for a lot when comparing with a smaller sensor.
In terms of detail resolved, the 7D would require a far sharper lens than the 5D to get the same kind of per-pixel sharpness and thus significantly more image detail. A merely decent zoom at f/8 isn't going to cut it. You'd need a brilliant lens at f/4, and ultra-critical focus placement, to really do this sensor justice.
It might be a decade or more before a 1.6x-crop-factor camera can beat the 5D for noise across most of the brightness range, though cameras like the Nikon D300 and possibly the 7D beat it today for deep-shadow noise and hence dynamic range.
stanj wrote:
I need better glasses
Maybe not! Measuring between fixed structures in the images, I reckon CarpeyBiggs was off in his framing by about 3% in favour of the 5D in the ISO 100 comparison. This isn't enough to matter. But in the ISO 1600 images the 5D is favoured by about 10%. That's enough to cause noticeable differences in resolution. You were right to expect the 7D images to appear larger.
The 7D isn't going to overcome its noise disadvantage to the 5D by framing 10% tighter, but the differences would be slightly less pronounced (in the ISO 1600 comparison only).
Edit: CarpeyBiggs, what was the shutter speed for the 5D's ISO 1600 image? The 7D was at 1/60th according to the Exif data.
'aight, gimme some direction here on trying to make this test more controlled... i can run out in the backyard with the two cameras, slap the 35 f2 on, shut it down to f8, and fire off a couple sequences.
how do you get a proper apples to apples comparison after that? what size should the 5d file be uprezzed to so the images can be compared accurately?
For starters you only need to shoot the 7d at F5.6 or so as it has a greater depth of field. Also shooting the lens at 24mm vs 40mm probably isn't fair as lenses are generally less sharp at their min and max focal lengths.
CarpeyBiggs wrote:
'aight, gimme some direction here on trying to make this test more controlled...
You don't even need to go outside. Set up a tripod inside with a few household objects or scene and fire off a few images using the same settings (or matching exposures) with constant lighting. Start at ISO 100 and move on up to 3200. You can shoot multiple apertures as well if you wish to remove all doubt. If you have a cable release it may be wise to use it otherwise the timer will do. MLU is advised as well. When it comes to framing I move the tripod itself to match AOV rather than using two different FL's or comparing different framing. As for equalizing resolution just set the 5D files to the same resolution as the 7D's in PS. Voilą...controlled test.
CarpeyBiggs wrote:
'aight, gimme some direction here on trying to make this test more controlled... i can run out in the backyard with the two cameras, slap the 35 f2 on, shut it down to f8, and fire off a couple sequences.
how do you get a proper apples to apples comparison after that? what size should the 5d file be uprezzed to so the images can be compared accurately?
I think your test was pretty decent as far as these things go. Thanks for posting it.
It makes more sense to compare similar fields of view than similar (or identical) focal lengths, so you did that right (albeit a few percent off, I think!).
If you get the framing exactly the same between the two cameras (which would require about ~1.6x longer focal length on the 5D), then you could resample the 5D image to the size of the 7D's image (i.e. 5184 pixels wide). Then post 100% crops of those. But if you do that people will complain that the resizing algorithm isn't perfect. The performance of the lenses will also differ.
Another possibility would be using the same focal length (i.e. lens) on both cameras, and changing the distance to the subject by about 1.6x. That would necessarily imply a relatively close subject, which would open up more problems with focus accuracy. And lenses usually deteriorate in performance as they focus closer, so again the test wouldn't be 100% "fair".
One thing to note is that you'll not get 5D-like per-pixel sharpness out of the 7D at f/8. You'd need a larger aperture: f/5.6 at least and ideally f/4. But few lenses are sharper at f/4 than f8, even in the centre. Your 35 mm f/2 prime might be though (but only at the centre), so it would be a good candidate for this kind of test.
All of this runs into the real problem with APS-C or DX cameras: nearly all the good lenses are designed for 135 format. If Canon or Nikon made $2k primes with an image circle limited to APS-C format, they'd be stupidly sharp. But they don't, so we're left cropping the centre out of full-frame images, which always militates against sharpness. With 18 megapixels to play with, the limits of that cropped image become very clear.
M Vers wrote:
You don't even need to go outside. Set up a tripod inside with a few household objects or scene....
.
Oh, but I so prefer to see comparisons of real images - things I might really want to photograph - this comparison feels so much more meaningful and enjoyable. I much prefer the comparison presented here by the OP than than images of bookshelves!!!!
michael49 wrote:
Oh, but I so prefer to see comparisons of real images - things I might really want to photograph - this comparison feels so much more meaningful and enjoyable. I much prefer the comparison presented here by the OP than than images of bookshelves!!!!
For all intents and purposes an unchanging scene would offer more solid/reliable results not to mention the fact that it could be done day or night, rain or shine. We don't need him to post beautiful landscape scenes--this is for comparison purposes only
stanj wrote:
I must be dense but if the images were framed the same, how can this be a 1:1 comparison? Surely these can't be 100% crops, so the question is if the 5D was scaled up or the 7D down? Or am I missing the boat completely?
The APS-C format is always magnified by 1.6x more to achieve the same final print size with the same frame content, so lens resolution matters, and APS-C is inherently handicapped, Size matters in photography. So providing comparative images at same subject size from the two cameras is totally valid!
Well for landscape work I'd use the 5D any day from these results. I would like to have see the 7D raw file and see if the sharpening can be improved, as it is much softer. I realize the 100% crop will look worse on the 7D since it's a much bigger enlargement though. Maybe a 7D downrezzed to 5D size or 5D uprezzed to 7D size would be fairer.
Stan, the 7D subjects should look around sqrt(18.1/12.8) = 19% larger, which seems about right from the 100% crops.
The performance of the lenses will also differ.