OM-D E-M5 vs E-5 (build quality)

Started May 13, 2013 | Discussions thread
Great Bustard
Forum ProPosts: 23,182
Like?
Re: Printing big: the specifics
In reply to John King, Jun 2, 2013

John King wrote:

Specifics:

Six prints. Two each taken with E-1, E-510 and E-30. Subject matter was 2 each of portraits, cars, 1 modern building internally, 1 outdoor bamboo sculpture. All images had parts with extremely fine detail and edges, both centrally and towards the edges and corners.

Printed on Epson R3880 in highest quality print mode.

All images start life as a 12 bit RAW file, developed using ACR using ProPhotoRGB 16 bit colour space. Then taken into CS5 PS for further upsampling, USM and printing. Minimal PP is applied to any image I print at A2 size, as data loss easily destroys resolution, colour, and overall impact of the printed image IME.

All complete prints for framing printed at A2 size (actually 17 x 22 inches; printed area size = 54.65 x 40.99 cms).

Printing done on Ilford Smooth Pearl paper/s, using the Ilford ICC profiles for their paper and the R3880.

A2 prints were examined before glazing at the nearest focusing distance for naked eye using whatever spectacles are usually used by the observers for working at this distance. I had 20/10 vision at the time of the tests in question (when wearing bifocals that correct my vision to that visual acuity). My visual acuity with glasses has since declined to around 20/12 since then. This is still somewhat better than average visual acuity of around 20/16 (see http://en.wikipedia.org/wiki/Visual_acuity ).

The prints were examined in natural light, but not direct sunlight, adjacent to north facing windows on a sunny day - i.e. bright enough to critically examine the prints without being so bright as to damage them or to cause maximal pupillary contraction in the viewers.

Prints were also examined using a 4x magnifying glass by the observers. It was during this examination that slight lack of sharpness was noted with the E-1 prints. This slight difference in resolution was not noticeable without the magnifying glass by any observer.

So, at 17x22 inches, you were unable to see any differences in resolution between 5 MP and 12 MP without the aid of a magnifying glass.  Is this correct?

I am not suggesting, and never have suggested, that there is no difference in IQ between the camera bodies and lenses. Lenses used varied from an OM mount f/3.5 28 mm, f2/50 macro, 14~54 MkI, 14~42, 40~150 MkII and 14~54 MkII.
That is not, and never has been, the point of this examination.
We were looking for differences in resolution between the 5, 10 and 12.3 MP cameras when printing on a very high resolution output device.

Well, aside from resolution, what other differences does the pixel count have to do with IQ?  However, since you bring it up, what were the IQ differences that you did see, if not resolution?

BTW, I mention again that there is no monitor made, or at least commercially available, that has resolution even approaching the resolution of any modern, high quality pigment ink printer.

As we all know, we cannot get more resolution than what the camera captures.  The question, then, is how much of that detail the display medium can represent.

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.

Let's discuss these points in more detail.  5 MP at 17 x 22 inches results in 116 PPI (pixels per inch), whereas 12 MP results in 179 PPI.

So, if we cannot perceive a difference between 116 PPI and 179 PPI, then we will not notice a difference between 5 MP and 12 MP.  Alternatively, the lenses used might not be sharp enough to result in any visible difference in resolution between 5 MP and 12 MP.

On the other hand, we can display the photo at any PPI we like by upsampling.  So, we could upsample both the 5 MP and 12 MP files to, such as 300 PPI (the "gold standard" of viewing resolution).  So, it may be that the additional false detail introduced in the 5 MP file over the 12 MP file when upsampling to 34 MP (300 PPI for a 17 x 22 inch print) cannot be visually distinguished from real detail.

Lastly, it might be exactly opposite.  That is, the resolution lost in printing is such that the 12 MP loses much more detail in the printing process than 5 MP, just as 12 MP will lose more detail than 5 MP at narrow apertures.  In other words, while 12 MP will always retain more detail than 5 MP, it will asymptotically lose that detail advantage as the aperture narrows, just as it may lose much of the detail advantage when being printed.

A good analogy to this last point in dynamic range.  Assuming that the darkest inks have a reflectivity near that of charcoal (1%) and that the paper itself is essentially 100% reflective, then the greatest continuous range of DR a print can display is 6.6 stops, which we can compare and contrast with the 10+ stops of DR that *all* modern sensors are capable of recording.

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.

My own ASUS ProArt PA246Q has a resolution of 1920x1200 pixels. This (nominally) 24" monitor thus has a display of some 2,304,000 pixels total. This translates to approximately 27,000 sub-pixels per square inch.
This monitor will display almost all of the aRGB colour space, with a colour lookup table of 12 bits, and a 10 bit panel.

The R3880 prints 2880 dpi, regardless of the ppi of the image sent to it. This is approximately 8,294,400 droplets per square inch ... This printer (and its predecessors back to about 2004) will print colours that are well outside an aRGB colour space. The printer uses 4 black cartridges (three in use for any given paper) and five colour cartridges (Cyan, light cyan, magenta, light magenta and yellow. It is therefore an 8 colour printing process.
Thus the printer has both a far higher resolution than the monitor, and a far wider colour gamut.

May I suggest that you visit the Smugmug site if you want to find out what they think? After all, they process millions of prints a year. Here:
http://help.smugmug.com/customer/portal/articles/93358

Specially at the very bottom of this page:
http://help.smugmug.com/customer/portal/articles/93359

Please bear in mind that Smugmug specifically state that they are using commercial printers for commercial prints, and that specialist and expert users will often use wide gamut colour spaces and high bit depth, because they know what they are doing, and are not prepared to sacrifice highly saturated colours that get lost in commercial printing processes. For further on this issue, see (e.g.) Schewe & Fraser "Real World Camera Raw with Adobe Photoshop CS5" at pps.11-12 (et. al.). Smugmug also provide a link to this site regarding the use of wide gamut colour spaces such as ProPhotoRGB:
http://www.naturephotographers.net/articles1203/mh1203-1.html

Hope that this answers some of the questions you have raised, or alluded to ...

All of the above has been explained to you before, and recently.
I am quite surprised that you appear to have forgotten it so quickly.

Not forgotten, but rather irrelevant to the question I was asking.

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Hmm.New
Yup!New
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow