EOS 5D MII REview

[marcfs]

Bob,

Thanks for writing a thorough and very informative review.

Additionally it was an enjoyable read!!

Marc Schoenholz

[Bob Atkins]

Thanks. My only regret is that I had to send the camera back!

http://www.bobatkins.com/photography/digital/canon_eos_5D_mkii_review_1.html

[munder]

Hi Bob,

I've had a D30 since 2001 and I'm considering upgrading to the 5D II. Thanks for your review on it - very informative and helpful.

In particular I was glad to see the issue of resolving power with regard to total sensor pixels and pixel density qualified for once. I've read many online articles and forum discussions/arguments on this subject from many websites and noticed next to none have qualified their statements as well as you have here. Since resolving power is related to many factors, it's important that unqualified absolute statements be avoided.

Some comments/questions though ... Do you think it would be better to say up front that resolving power is dependant upon many factors and to distinguish between actual and apparent resolving power (i.e. due to photographer's application) rather than making such a strong statement that "the EOS 40D, EOS 50D and even the Digital Rebel XSi and the lowly Digital Rebel XS actually "outresolve" the EOS 5D MkII (and the EOS 1Ds MkIII)."?

Would it have been better to state that although in many applications the 5D II outresolves the 40D/50D etc, there are some cases where the 40D/50D can outresolve the 5D II? (This seems to be the case, but only from what I've gleaned from online sources.)

If two cameras capture the same field of view on their sensors won't the camera with a larger number of sensor pixels generally out resolve the other (all other controllable variables being the same), especially if the former's sensor also has larger pixels that introduce much less noise into the image, which seems to the case with the 5D Mark II? Is it correct that pixel density per "composed image" matters more than pixel density per sensor?

I understand your point that if the final application is small prints the increase resolving power may not be apparent as you stated. But doesn't the 5D II still outresolve the others.

Wouldn't the article be more accurate if you turned the discussion around? As it is, it sounds like this section of the article is saying the increased resolving power of the 5D II is of very limited benefit. (Although by the end of the section it does seem to turn things around.)

My reason for all this is simply that I've been stuggling to throughly understand this issue among others in my camera upgrade decision and this one is still a little confusing.

Am I way off base on what I thought I read or what I think I understand? Please let me know. (BTW, this is my 1st forum post anywhere, anytime. So I appologize for its length.)

Any feedback would be appreciated?

[Bob Atkins]

My main point with the resolution argument was to debunk the myth that the 5D MkII needs sharper lenses than the Rebel XSi to take full advantage of the sensor - and that's just not true. The XSi actually pushes the lens harder than the 5d MkII does.

However you are correct that for a given view (which of course would require a different lens on the XSi and the 5D MkII), the image with more pixels (i.e. the 5D MkII) will be better and the prints will show more detail.

[KeithB]

I think that "pushing the lens harder" needs some qualification, since the FF camera needs a much bigger image circle it "pushes the lens harder" in a different direction, since lens defects are more pronounced in the farther corners of the FF camera.

[Bob Atkins]

True. I'm really talking about central region resolution. Full frame cameras do need high quality over a larger image circle to take full advantage of the larger sensor.

Coma, astigmatism, lateral chromatic, distortion and field curvature) all get worse as you move further from the opical axis.

[munder]

   Thanks for you reply Bob. Now I see more clearly where you were coming from.

While were on the resolution topic ... I was wondering about something that I haven't seen discussed anywhere yet.

Assuming the pixels of any digital camera were square and that the horizontal and vertical spacing were equal, the distance between pixels on the diagonal would be 1.41 times greater than the horizontal/vertical pixel spacing. Doesn't this indicate that the resolving power of a sensor at angles other than 0, 90, 180 and 270 degrees would be significantly lower than the resolution (and subsequently resolving power) along the vertical/horizontal?

I haven't come across any pictures that support this. Yet it would seem that such a significant difference in resolution along other angles should be readily noticeable - right? Or is the reason it goes unnoticed possibly that it has something to do with the acuity of the human eye/brain in the horizontal vs. vertical vs. other angles?

[KeithB]

Interesting discussion by Ctein about how much resolution is required to make no discernable difference in a print - note that this is *not* "as good as 35mm film":
http://theonlinephotographer.typepad.com/the_online_photographer/2009/02/why-80-megapixels-just-wont-be-enough.html

[klindup]

Bob what is the equivalent number of pixels for a full frame 35mm negative?
Ken Lindup

[Bob Atkins]

Ken - it's a bit of an "apples and oranges" situation when comparing digital pixels and an analog 35mm film image. They degrade in different ways as you approach the resolution limit.

For all practical purposes, in terms of real world prints, something like the 12.8 MP EOS 5D will equal or better 35mm film.

If you were to take the slowest, finest grain, highest resolution B&W film and use a really good lens, you can calculate that you'd need something like 35MP to equal it, but in practice you need a lot less.

[Bob Atkins]

...and as for diagonal resolution, in fact it should be higher than vertical or horizontal by a factor of about 1.4, not lower. If you look at the geometry carefully you will see that on the 45 diagonal there is actually a row of pixels between what you might first expect would be neighboring rows (see attached image).

Fuji used this reasoning in their "Super CCD HR" sensor, where they placed pixels on a 45 diagonal grid rather than horizontal and vertical as in most other sensors and claimed higher resolution.

[klindup]

....sio I guess that using a 6X6 balck and white film of say ISO 80, a high resolution developer and my Rolleiflex would put me ahead of the game for a really big print.  But then I have to worry about dust on the negative carrier and water marks.  I'll stick with my 40D.

Ken

[munder]

I must not be interpreting the image correctly. Are there more than 9 pixels laid out in a square pattern in that image? Are you saying there is a pixel at each intersection of the two blue diagonal lines?

[Bob Atkins]

The red arrow shows the spacing between nearest horizontal rows. The blue arrow shows the spacing between nearest 45 degree diagonal rows. As you can see, the blue arrow is shorter than the red arrow, i.e. the spoacing is closer which leads to higher resolution (in theory...) for closley spaced 45 degree diagonal lines than for similar horizontal or vertical lines.

[munder]

Bob, I was thinking about the point you mentioned about debunking the myth ...

The myth you describe sounds different from what I've come to understand it to be in reading about the need for higher quality lenses for the 5D MkII vs. the XSi or similar sensors. I understood the reason to be not about resolution but sensor size due to vignetting, since better lenses exhibit less vignetting at larger apertures and only full frame sensors will capture the vignetting exhibited at the edges of the image circle.