8.0mp vs. new 10.1mp

i thought the general consensus was that the 350D has better IQ than the 400D due to some exposure issues on the 400D? Or was that resolved?

donboyfisher wrote in post #2351693
But most folk dont pixel peep posters . . . most folk will stand 3 or 4 ft away and admire the image as a whole.

Most folks do pixel peep posters. Every photo exhibition I've ever been to has been like that. Unless you're talking mural size prints, people very seldom stand more than 2 or 3 feet away. I just went to a photo exhibition at the Smithsonian, in which a number of prints were enlarged to 24x36 inches, and most people were looking at it, again, from 2-3 feet (and many were closer).

That said, I don't think you'll see the difference between 8 and 10 megapixels at all until you get to an enlargement so big that both are pretty much at their technical limits.

DrPablo wrote in post #2351680
If you have a sensor that can record 100 lpm and a different one that records 80 lpm, the 100 lpm will only record better if your lens transmits greater than 80 lpm of detail. In other words, if the better sensor is lens-limited and the inferior sensor is sensor-limited, then there will be more detail recorded by the better sensor (assuming an lpm target or a finely detailed subject).

Actually, the result is not 0 or 1. That is, even though your lens is 80 lpm, the 100 lpm sensor will still record more resolution than the 80 lpm sensor. The reason is: the recorded data is a combination of the lens and the sensor. The recorded quality does not change in 0/1 fasion.

See the dpreview test on Canon XTi, and the comparison with 8MP sensor results.

donboyfisher wrote in post #2351693
I think unless you are pixel peeping at poster sized print outs, then you are unlikely to notice differences between 8 and 10MP in the real world

But most folk dont pixel peep posters . . . most folk will stand 3 or 4 ft away and admire the image as a whole.

I essentially agree with this part.

Using the same arguement as above, you will conclude you do not need a very good lens (a good lens is good enough).

pieq314 wrote in post #2352170
Actually, the result is not 0 or 1. That is, even though your lens is 80 lpm, the 100 lpm sensor will still record more resolution than the 80 lpm sensor. The reason is: the recorded data is a combination of the lens and the sensor. The recorded quality does not change in 0/1 fasion.

My point was more or less that the sensor, no matter how high or low its intrinsic resolution, is always beholden to what is transmitted by the lens. If you have a 20 MP sensor vs a 5 MP sensor, and you have a crappy lens that by virtue of low lpm resolution only projects a large CoC, the 20 MP won't resolve more real world detail -- it will just record the transmitted detail with more precision. The same is true if you take a picture that is intentionally completely out of focus (which simply increases the size of the circles of confusion) -- a super high res sensor isn't going to pull more detail through the lens or make those out of focus areas sharper.

I think of high sensor resolution as mainly relevant to print size -- because you need a certain pixel density for an image to look like a photograph and not a Seurat painting. And having more pixels will give you larger print sizes in which you have that critical density of pixels (whether that's 200 or 300 dpi or whatever). But if your sensor is a fixed physical size, like 15x24 or 24x36, even if you have enough pixels to make an 8x10 foot mural at 300 dpi, you're still going to have to magnify again and again and again the circles of confusion, which are the smallest details projected onto the sensor, and the larger you go the softer the picture will look.

For this reason you can make the case that every camera, film, sensor is always lens limited. But whether that translates to a decrement in image quality depends on your print size.

DrPablo wrote in post #2352362
My point was more or less that the sensor, no matter how high or low its intrinsic resolution, is always beholden to what is transmitted by the lens. If you have a 20 MP sensor vs a 5 MP sensor, and you have a crappy lens that by virtue of low lpm resolution only projects a large CoC, the 20 MP won't resolve more real world detail -- it will just record the transmitted detail with more precision. The same is true if you take a picture that is intentionally completely out of focus (which simply increases the size of the circles of confusion) -- a super high res sensor isn't going to pull more detail through the lens or make those out of focus areas sharper.

I think of high sensor resolution as mainly relevant to print size -- because you need a certain pixel density for an image to look like a photograph and not a Seurat painting. And having more pixels will give you larger print sizes in which you have that critical density of pixels (whether that's 200 or 300 dpi or whatever). But if your sensor is a fixed physical size, like 15x24 or 24x36, even if you have enough pixels to make an 8x10 foot mural at 300 dpi, you're still going to have to magnify again and again and again the circles of confusion, which are the smallest details projected onto the sensor, and the larger you go the softer the picture will look.

For this reason you can make the case that every camera, film, sensor is always lens limited. But whether that translates to a decrement in image quality depends on your print size.

I was in a hurry earlier. Here is more details:

If you have a lens at 80 lpm, and a sensor also at 80 lpm. The recorded resolution is not 80 lpm, but something a little less (let's say 70 lpm for the sake of discussion).

But if you have a sensor that is 100 lpm, then, the recorded resolution will be higher but still less than 80 lpm (let's say it is now 75 lpm).

If you have a sensor that is 1000 lpm, then, the recorded resolution will be 79.9 lpm (for the sake of discussion).

If you have a sensor that is infinite lpm, then, the recorded resolution will be the lens resolution of 80 lpm.

In other words, you still see improvements in recorded resolution, but there is a limit in sensor lpm where the return is diminishing.

For the 10MP XTi, the return in resolution is still there, as seen in the dpreview test.

delhi wrote in post #2352006
i thought the general consensus was that the 350D has better IQ than the 400D due to some exposure issues on the 400D? Or was that resolved?

See dpreview.com for the XTi/400D test. The conclusion is, 400D is better (but as many argued and I agree, 8MP vs 10MP is not a big deal).

I have the 30D and the 5D. I don't print much larger than 8 X 10 with either. I have on occasion gone bigger with both. I did some very nice 16 X 20's with the 30D and they turned out excellent. I don't think I'll be doing any billboards lately. The 30D's 8 megapixels is probably all I will ever need. Its nice having the 5D's resolving power if needed, but I am not quite sure I have enough computer to handle files larger than what the 5D can produce. Maybe a 5D, 30D with digic III.

pieq314 wrote in post #2352415
I was in a hurry earlier. Here is more details:

If you have a lens at 80 lpm, and a sensor also at 80 lpm. The recorded resolution is not 80 lpm, but something a little less (let's say 70 lpm for the sake of discussion).

But if you have a sensor that is 100 lpm, then, the recorded resolution will be higher but still less than 80 lpm (let's say it is now 75 lpm).

If you have a sensor that is 1000 lpm, then, the recorded resolution will be 79.9 lpm (for the sake of discussion).

If you have a sensor that is infinite lpm, then, the recorded resolution will be the lens resolution of 80 lpm.

In other words, you still see improvements in recorded resolution, but there is a limit in sensor lpm where the return is diminishing.

For the 10MP XTi, the return in resolution is still there, as seen in the dpreview test.

Yes, I think this is basically what I was saying as well, i.e. sensors with better resolving power will record the transmitted detail with more fidelity. And since most people are choosing from among different APS-C cameras, but not among different sensor sizes (or even significantly different lenses), then those 2 megapixels are sure to have an impact on outcome.

In the grand scheme of image quality, there are already big constraints on detail capture using this kind of camera. The smaller the sensor (physically) the more desperately dependant the image becomes on high glass quality, and the more dependent the image becomes on high sensor resolution just to capture the detail that's transmitted.


Regarding the DPReview, I haven't found anything in their review demonstrating an advantage of the 400D over the 350D (with respect to resolution)... in fact the 350D looks better on this first page. I think part of it may be the unavoidable methodological limitation of resizing for web output -- resizing from 10 MP to web size will be more degrading to the image than resizing from 8 MP. Still, the resolution tests look like a wash to me, and the noise at ISO 800 and 1600 look better on the 350D.

http://www.dpreview.co​m …s/canoneos400d/​page24.asp
http://www.dpreview.co​m …s/canoneos400d/​page27.asp
http://www.dpreview.co​m …s/canoneos400d/​page28.asp

DrPablo wrote in post #2352638
Regarding the DPReview, I haven't found anything in their review demonstrating an advantage of the 400D over the 350D (with respect to resolution)... in fact the 350D looks better on this first page. I think part of it may be the unavoidable methodological limitation of resizing for web output -- resizing from 10 MP to web size will be more degrading to the image than resizing from 8 MP. Still, the resolution tests look like a wash to me, and the noise at ISO 800 and 1600 look better on the 350D.

http://www.dpreview.co​m …s/canoneos400d/​page24.asp
http://www.dpreview.co​m …s/canoneos400d/​page27.asp
http://www.dpreview.co​m …s/canoneos400d/​page28.asp

See:
http://www.dpreview.co​m …s/canoneos400d/​page28.asp

400D picture looks worth because you are looking at resolution/pixel, but since 400D has more pixcels, the total resolution is higher.

Measurable findings (three measurements taken for each camera):
Canon EOS 400D
Horizontal LPH 2200 2300
Vertical LPH 1800 2300

Canon EOS 350D
Horizontal LPH 1850 2100
Vertical LPH 1650 2100

Isn't that sort of a statement of equivalence? I.e. more pixels = more pixels. A given detail will again, by definition, be recorded by more pixels on a camera that (all else being equal) has more pixels.

But that doesn't answer any kind of practical question about resolving power -- recording power is more accurate. Their example doesn't make a case for improved detail resolution -- only a case for more absolute pixels, which we already know. In other words, they don't make a case for this being relevant in practice.

I'm not making a general statement by this -- I'm just skeptical that the test shows anything that would help one discriminate between the two cameras, or that says anything more than that there is a numerical difference in pixels.

DrPablo wrote in post #2352878
Isn't that sort of a statement of equivalence? I.e. more pixels = more pixels. A given detail will again, by definition, be recorded by more pixels on a camera that (all else being equal) has more pixels.

But that doesn't answer any kind of practical question about resolving power -- recording power is more accurate. Their example doesn't make a case for improved detail resolution -- only a case for more absolute pixels, which we already know. In other words, they don't make a case for this being relevant in practice.

I'm not making a general statement by this -- I'm just skeptical that the test shows anything that would help one discriminate between the two cameras, or that says anything more than that there is a numerical difference in pixels.

The number is the resolving power. If you use a junk lens with a resolution of 1 lpm, then, both 350D and 400D will give essentially the same resolving power.

See the crop photos from these two cameras. The one from 400D can see details further down the chart.

pieq314 wrote in post #2352942
See the crop photos from these two cameras. The one from 400D can see details further down the chart.

But the one from the 400D is also enlarged more than the one from the 350D.

That might explain some of the difference (at least has to be accounted for).

DrPablo wrote in post #2352638
Still, the resolution tests look like a wash to me, and the noise at ISO 800 and 1600 look better on the 350D.

http://www.dpreview.co​m …s/canoneos400d/​page24.asp
http://www.dpreview.co​m …s/canoneos400d/​page27.asp
http://www.dpreview.co​m …s/canoneos400d/​page28.asp

Hmm, the noise chart looks to me like the 400D has less noise at all ISO than the 350D. Perhaps I am reading it incorrectly.
http://www.dpreview.co​m …s/canoneos400d/​page17.asp

DrPablo wrote in post #2353329
But the one from the 400D is also enlarged more than the one from the 350D.

That might explain some of the difference (at least has to be accounted for).

It is not enlarged after the photo is taken. 400D has more pixels, so for the same subject area, it is larger on the screen.