A very interesting question on resolution and prints

Started Jun 3, 2013 | Discussions thread
Great Bustard Forum Pro • Posts: 42,955
Re: More detail

rovingtim wrote:

I wonder if some manufacturers are encouraging 'false detail' in their processing?

If anything, just the opposite -- applying involuntary NR (noise reduction).

By the way, I noticed that the thread where I linked my reply to your OP is now gone.  Allow me to reproduce my reply here, which I feel has some pertinent points (the quoted sections are John King, not you):


IMO it is because all the cameras are still within the reasonable limits for  printing when the print size is at around 22 x 17 inches. The E-1 is capable of  printing at this size natively at 119 dpi - well above the 80 dpi limit set  by Smugmug, and as demonstrated by them on their own site.

So, is that the answer to my question?  That a print above 80 DPI is  "good enough" that any higher resolution will not result in any visible (to the  naked eye) difference in resolution in a print?

Up-sampling the files to the precise print size and using 360 ppi as the  benchmark "resolution" is just 'filling in' the space between pixels. IOW, it  does not involve "stretching" the image, more a matter of "filling in the  gaps".

What we are "filling in the gaps" with is false detail -- something you  denied in an earlier post (as always, links and quotes available per  request).

In addition, you seem to be using the terms "DPI" and "PPI" interchangeably  when, in fact, they are very different things.  PPI (pixels per  inch) is a function of the number of pixels in the image file and the size  the photo is displayed at.  DPI (dots per inch), on the other hand, is a function of the number of ink drops used by the printer to make a print.

The difference is significant.  Each pixel in the image file can have  millions of colors.  Each dot of ink from the printer can be at most one of  eight colors, so one dot from the printer most certainly does not represent one  pixel from the image file.

So, assuming you mean PPI, as opposed to DPI, are you saying that an image file that results in an 80 PPI print will resolve as well as a person can see  with their naked eye?

The E-1 also brings some wonderfully "good" pixels to the table. This may be  one reason why this camera is so liked - its output is very, very good. Compare  it with my 5 MP Nikon, and you will immediately understand what I am talking  about.

I take it you say that not all pixels are created equal.  That is, 5 MP from the E1 is superior to 5 MP from your Nikon (which, I presume, is a compact).  Well, of course this is true.  This is why I do not use  even PPI as a measure of resolution, but rather lw/ph (line widths per picture height).

For example, for a sufficiently sharp lens, negligible diffraction softening,  negligible motion blur, and for the portions of the scene within the DOF, 12 MP would resolve 55% more linear detail than 5 MP (i.e. the lw/ph for a 12 MP file would be 55% higher than the lw/ph for a 5 MP file under the given  assumptions).

In general, of course, these conditions are not met, so 12 MP will not resolve 55% more than 5 MP, but anywhere between 0% and 55% more, all else equal.

In the other thread, I posed the following possibilities to  you:

So, we are left with the following conclusions:

  • 5 MP records more detail than your eye can see at 17 x 22 inches.
  • The lenses used are not sharp enough to capture more detail on a sensor with more pixels.
  • The false detail introduced by upsampling is good enough to pass as real detail.
  • The effect of printing on resolution is similar to the effect of diffraction softening.

I then asked the following question that went unanswered, and, indeed, completely ignored:

So, which do you think it is?  Or,  perhaps, a combination of the above factors?  If there is an alternative I overlooked, please feel free to offer it and explain.

Perhaps now would be a good time to answer the question directly.

I have also found that the print quality appears to be marginally better when  up-sampling is used with all three cameras.

This is rather contradictory.  First, you say that there is no visible  (to the naked eye) difference in resolution between 17 x 22 inch prints from the  5 MP E1 and the 12 MP E30.  Now you say that upsampling, which merely  introduces false detail (albeit in a pleasing fashion if done well), produces a  noticeably, if not substantially, better quality photo.

The only resolution I can see for this contradiction is that you are claiming  that the false detail introduced by upsampling is superior to the real detail captured by a sensor with more pixels.  What say you?


Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow