#26. "RE: Will Higher Resolution of D800 Be Visible in "Native" Resolution Prints?" In response to In response to 25 Thu 08-Mar-12 02:01 PM by nrothschild
All interesting information. The ruler test is not a very "ambitious" project. We all have a ruler, a roll of masking tape, and a suitable wall. I suspect that for many people that test is not so much "ambitious" as the results are "embarrassing" and I note that this thread is not exactly overwhelmed with additional reports from other members (which I would be VERY curious to see).
The main reason I think it might be an embarrassing test is that most people's real world resolution is nowhere close to the 60' resolution suggested by the Snellen chart. No offence to Snellen but as I alluded to previously, in the one endeavor I am familiar with where these things are computed with great precision (astronomical double star observation) very, very few people claim anywhere near that capability.
As an example for those inclined to research the point, the double-double binary Epsilon Lyra is nearly ideal for the purpose; it is bright and high in the sky all summer for Northern Hemisphere observers and a truly spectacular sight. Because of that there are many, many user reports to be found on the net. It is almost everyone's favorite binary (me included!). One of the major pastimes of double star observers is to see how little magnification they need to cleanly split any given pair.
They are measuring not only the quality of their optics but even more so their eyesight since most of these observers are using the highest quality optics and there aren't that many choices in that regard. These observers have generally spent many years training their eyes to separate these two dots
The two E Lyra pairs are separated by about 2.3 and 2.6" (arc-seconds). The simple math of magnification is such that if you use 100x of magnification the resulting visual separation of those tiny dots is then about 230-260 arc-seconds, which is very consistent with the COC used in standard depth of field tables (~225") and not consistent with the Snellen chart at 60".
Under ideal conditions and known diffraction limited optics most visual observers of E Lyra need about 100X.
I know that at 160x (384") I can unambiguously split it, and I am pretty sure I can do the same at 100x (230-260"). At 80x (184") it may be real or it may be imagined (and that is the nature of this work since we know exactly what we are trying to see!). That puts me in the 200-225" range and that is as close as I'll try to nail it here, especially since I haven't done this under even near ideal conditions since last summer.
I've seen reports as low as 45x (equivalent to about 112" or just under 2 minutes or half the Snellen 20/20 number). These are rare individuals and there is simply no way to positively confirm this; we just have to take their word for it.
I don't ever recall anyone suggesting they could split it at 30x (just about the Snellen 20/20 limit), but neither have I tried to track down every posted observation claim.
I'm just trying to objectively analyze the reference you linked to which suggests you "need" over 400PPI to achieve the Snellen number, as if that is "typical" of your targeted audience. Let's see if anyone here can resolve a millimeter ruler at 11.3 feet. My number is more like 3-4 feet than 11.3 feet, and very consistent with my double star observations.
What I'm saying is that if 400 PPI were readily at hand, I would overkill it and use it just like I try to overkill other things, in order not to think too hard about it . But would I spend $3K on a camera just because 1 person in 10 or 20 (who is not me!) *might* perceive a difference? This is the context of my thoughts.
Your print test results are basically in agreement with what I just said. I'm just putting some numbers behind it to suggest your results are not necesarily atypical.
I do have one question about the sample images you prepared. Were the stated PPI image resolutions the actual image file resolutions as sent to the printer or were they the native resolutions subsequently interpolated up to some greater resolution to make the printer happy?
There is a huge difference. For example, the "stair stepping" you described is often called "pixelation" and that is the basis for determining a minimum acceptable image file resolution for a given printer. But, for a somewhat extreme example, an image at a native resolution of 50PPI, which would likely pixelate badly if printed as is, would simply be somewhat soft (but not pixelated!) if up-sized to "make the printer happy". And of course we can then play with post up-sizing sharpening and the interpolation method to try to change that perception . And we can do this for free with any image processing software. We don't need a $3K camera to do that
_________________________________ Neil Nikonians Team My Gallery