Rant 23

[pegelli]

They must have some use ;-)

Well, of course they do. I never wrote otherwise. This completely misses the point of what I've written though.

Michael, glad you explained, but in my opinion you did write they were of no use:

From what's new:
QUOTE: We've all done it. There's a photograph or a 100% crop of an image online, taken with a camera or lens that we're interested in, and we judge its image quality capabilities (at least in part) on the basis of that screen image.

Wrong!
UNQUOTE

Unless I'm really wrong interpreting the words (at least in part), which is possible since English is not my native language.
The way I understand that sentence is that even partly judging sensor and lens qualities from internet pics and 100% crops is wrong, or did you have a different thought in mind when you wrote that  

[Eric Myrvaagnes]

Michael, glad you explained, but in my opinion you did write they were of no use:

From what's new:
QUOTE: We've all done it. There's a photograph or a 100% crop of an image online, taken with a camera or lens that we're interested in, and we judge its image quality capabilities (at least in part) on the basis of that screen image.

Wrong!
UNQUOTE

Unless I'm really wrong interpreting the words (at least in part), which is possible since English is not my native language.
The way I understand that sentence is that even partly judging sensor and lens qualities from internet pics and 100% crops is wrong, or did you have a different thought in mind when you wrote that  

Pieter,

You are correctly parsing what Michael said in English (although I think the rest of the essay clarified his views). The problem is, in English we have a peculiar custom of insisting that the listener/reader should always understand what I meant regardless of what I actually said.

[Jack Flesher]

Well, there you have it. A good monitor, properly calibrated, is not a bad device for getting a *fair idea* of what your print will look like, using proof setup in PS of course, and a good paper/ink profile.

As long as you leave the bolded modifier in I'd agree, however/but another thing is equally certain and I believe more to Michael's point:

When and where you can see differences between two images being proofed on your monitor you will almost certainly see those same differences in a good print, the converse is NOT necessarily true; you often detect differences in prints you cannot preview on your monitor!  The corollary point is that anybody who makes larger, wide-gamut prints with regularity knows this quite well, while folks who prefer theorizing about linear sensor response curves before the AD converter and post processing get applied, for whatever reason seem to refuse to acknowledge such differences can possibly even exist in the print side. You can lead a horse to water...  

Later,

[pegelli]

Pieter,

You are correctly parsing what Michael said in English (although I think the rest of the essay clarified his views). The problem is, in English we have a peculiar custom of insisting that the listener/reader should always understand what I meant regardless of what I actually said.

Maybe that explains it, however a few points:

• Allthough it's an English language forum it's really international, so forcing English habits on non native English speakers might fail (as it did here)
• Since it's a forum and not a face-to face discussion we're missing the "non-verbal" part making this even more difficult
• putting it between brackets puts a lot of emphasis on it, and it doesn't come acroos as just an idle remark in passing

I think I've found another cause for the confusion. In the essay he says:

QUOTE: But what about looking at images at 100% on-screen? There one can see the differences – right? Well, yes, one can see the difference of how they look on-screen at 100% magnification. The fact that this has little to nothing to do with how an image will appear in a print seems to escape many people. This applies to judging high ISO noise, resolution and more. UNQUOTE

While the "What's new" text talks about judging lens and camera image qualities being judged from 100% crops.

So in my mind the confusion lies between the difference between how an image will actually look in print (which I agree you can only see in a print) and judging certain lens and camera image qualities, which in my mind you can still "in part" do from 100% crops.

[Ray]

When and where you can see differences between two images being proofed on your monitor you will almost certainly see those same differences in a good print, the converse is NOT necessarily true; you often detect differences in prints you cannot preview on your monitor!

Jack,
You seem to be implying that two images that may look identical on the same monitor in the same color space, with same embedded profile and same rendering intent in proof setup, may print differently on the same printer using the same paper and ink. Is this true?

If it is true, then that is surely a printing problem which would require a test run before each print. I can appreciate that the latest Epson printers with improved color gamut may print subtle hues that are not visible on a monitor that may not be able to display even the full gamut of ARGB. However, in my experience such differences really are subtle and should not invalidate the comparison of image quality on the monitor.

[ErikKaffehr]

Hi,

Screens and printers are different. The color gamut volume of screens is actually larger than the gamut volume printers. Printers can reproduce some colors outside the screen RGBs but this does not apply to very bright or very dark colors. Especially LCD screens have large DR compared to print. Resolutionswise it's a different game. In my view an actual pixel view is much more demanding than print.

Also, ignorance of physics does not make for a better photographer but not necessarily a worse one, just keep in mind that we would not have the art if we did not have the physics.

Or to put it another way: Photography may be art but it's based on physics. We don't need to understand the physics in order to create art. On the other hand. art does not suffer from knowledge and actually understanding the process may even be helpful in improving the execution of the art.

Best regards
Erik

As long as you leave the bolded modifier in I'd agree, however/but another thing is equally certain and I believe more to Michael's point:

When and where you can see differences between two images being proofed on your monitor you will almost certainly see those same differences in a good print, the converse is NOT necessarily true; you often detect differences in prints you cannot preview on your monitor!  The corollary point is that anybody who makes larger, wide-gamut prints with regularity knows this quite well, while folks who prefer theorizing about linear sensor response curves before the AD converter and post processing get applied, for whatever reason seem to refuse to acknowledge such differences can possibly even exist in the print side. You can lead a horse to water...  

Later,

[cmi]

...and actually understanding the process may even be helpful in improving the execution of the art.

Absolutely Erik!

The artistic approach is about achieving practical results, it is not exact and purely result-centered. (One example, since the issue came up earlier: It is not important if sharpening increases resolution or edge contrast. You just adjust the slider and decide with your eyes.)

The engineering/scientific approach on the other hand is about scientific correctness, repeatability, exact comparisation. It is essential e.g. for Software and Hardware developers - applied science.

An artistic discussion in order to be useful should revolve strictly around the image and be held by real photographers. Technical explanations need to take the standpoint of the artist and come from someone with real working experience in the field. IF explanations come from a only theoretical side they must at least be understandable by an artistic mind, formulated towards that. So theoreticians and engineers should not post if they are not able to relate to the artistic standpoint and are not able to craft good and intuitive explanations. On the other hand, people who ARE able to give good technical explanation in the artistic context are always welcome, at least thats my experience.

Cheers,

Christian

[cmi]

Oh and I forgot someting, what we have here at LL is some mixup of the both. The analytical side arguing why the artists are ignorant.  //edit:// And sometimes also the other way around //edit end// This is unfortunate. Obviously the artist can make his image without being a physicist. At the end I am not required to be an engineer to do photos. Quite the contrary. Also the artist is not obliged to take the position of a scientist into account. He will do his art without it. Intuition and analytic thinking, different beasts. Not do downtalk the importance of the technical happenings in the camera through. And no doubt the best people out there in the business of computer graphics utilize both their artistic and scientfic knowledge, thats out of dispute. But what we had happen here in the past where no enlightening technical discussions. Everything else but that. These where to large parts absurd, academic and childish disputes. But I dont want to expand on that.

At the end no question both sides scientific and artistic, have each their merit in their own respective context. But most importantly they should respect - and hopefully also value each other (through the latter being maybe a bit of a far call currently).

Christian

[bjanes]

When and where you can see differences between two images being proofed on your monitor you will almost certainly see those same differences in a good print, the converse is NOT necessarily true; you often detect differences in prints you cannot preview on your monitor!  The corollary point is that anybody who makes larger, wide-gamut prints with regularity knows this quite well, while folks who prefer theorizing about linear sensor response curves before the AD converter and post processing get applied, for whatever reason seem to refuse to acknowledge such differences can possibly even exist in the print side. You can lead a horse to water...  

Later,

Wide gamut is a misnomer for prints. As previously pointed out, a decent monitor has a much wider dynamic range than a reflection print. Likewise, the color gamut of prints is restricted. Bruce Fraser admonished that you should temporarily look away from the monitor when displaying a soft proof lest you be shocked by the reduced gamut of the soft proof.

When differences are apparent on the print that are not apparent on the screen, it is likely that the tonal range or color gamut has been mapped to fit into the limited gamut of the print. Such rendering is not usually done for screen output.

What digital displays lack is resolution. It is not possible to view a 60 megapixel image all at once on screen. However, the field of view for sharp human vision is quite limited, and one can not view a large print all at once, but must move the eyeballs so as to view small portions at a time. However, if one views the screen at a suitable distance so that the angular resolution in pixels per degree rather in pixels per inch is the same on both media, then a better comparison can be made. However, panning of the screen is not as easy or natural than merely moving one's glance around.

Has Mr. Flesher ever compared a high resolution large format transparency to his prints?

[soslund]

Comedic interruption.....

I was reading this "rant" in between seeing patients in the Emergency Department where I work, when a disgruntled patient heaved his "petard" (aka "pee bottle" aka urinal) --yes, it was full of the warm yellow liquid--at no one in particular.  Thank goodness urine is sterile (mostly!!).  

Scott

[Ray]

Wide gamut is a misnomer for prints. As previously pointed out, a decent monitor has a much wider dynamic range than a reflection print. Likewise, the color gamut of prints is restricted. Bruce Fraser admonished that you should temporarily look away from the monitor when displaying a soft proof lest you be shocked by the reduced gamut of the soft proof.

Well, I can certainly affirm that I sometimes struggle to get an image, as viewed on the monitor after ticking the 'proof colors' box, to match the vibrancy of the image without proof colors enabled. It's always possible after much adjustment of contrast, saturation and brightness to get reasonably close. However, after making such fairly radical adjustments to get the proof image looking close to the way it previously looked before enabling proof colors, one always finds that the adjusted image then looks even more vibrant, and sometimes preferrable, when one unticks the 'proof colors' box and toggles between the two modes.

So I would agree that the transmissive nature of the monitor gives it an advantage which the reflective nature of the print cannot match, but the point Jack appears to be making is slightly different. He claims that, if two images of the same subject taken with different cameras, viewed side by side on the monitor, at print size, look identical, it does not necessarily follow that they will look identical when both are printed on the same printer using the same paper, ink and profile.

Even my old Epson 7600 claims to be able to print a few shades and hues which are contained within the ProPhoto RGB gamut and which cannot be displayed on the average monitor, or perhaps even any monitor. Epson have now apparently increased that gamut capability with there latest printers such as the 7900.

In other words, if the two images have an embedded ProPhoto RGB profile and are adjusted on the monitor in the ProPhoto RGB working space so that they both look as identical as possible, they may not actually be identical because there are subtle hues which the monitor cannot display but which a good printer, such as the Epson 7900, can reproduce.

I've never actually witnessed such an effect myself. It would be very troublesome and time-consuming to try and isolate such an effect. The mere fact that a particular print does not appear to quite match the appearance of the image on the monitor does not demonstrate the point Jack is making. One needs to compare print with print of apparently identical images taken with different cameras.

Two candidates for such a comparison would be the Canon 5D and the Nikon D700, both of which I own. They both have a very similar pixel pitch and pixel count so it should not be difficult, after ajustment of temperature, tint and levels etc, to get two images of the same subject looking very similar on the monitor.

However, if I were to go to the trouble of making such a comparison and were to discover that there was no significant difference between the two printed images, over and above any small differences apparent on the monitor, it would not disprove Jack's point. It's really up to Jack to demonstrate his point, but I don't see how he can if the reasons for such differences are due to the print containing hues that are not visible on the monitor.

[Jack Flesher]

Jack,
You seem to be implying that two images that may look identical on the same monitor in the same color space, with same embedded profile and same rendering intent in proof setup, may print differently on the same printer using the same paper and ink. Is this true?

No Ray, just like Michael in his rant I am talking about similar images from two different camera systems looking similar in certain areas on the monitor, but those areas printing out differently in a properly color-managed workflow.  

So now do you see that Michael is right?  Some folks simply refuse to get it, instead trying to twist the topic to some other discussion.

[Ray]

No Ray, just like Michael in his rant I am talking about similar images from two different camera systems looking similar in certain areas on the monitor, but those areas printing out differently in a properly color-managed workflow.  

So now do you see that Michael is right?  Some folks simply refuse to get it, instead trying to twist the topic to some other discussion.

A very subtle effect, I bet. Remember Michael's comparison of the P45+ and Canon G10 on A3+ prints? No-one could tell the difference, except for the fact the P45+ print had a slightly shallower DoF which sort of gave the game away.

[JohnKoerner]

A 100% view onscreen may tell you what you need to know about some aspect of image quality (except gamut, of course), but it tells you little about how the image from a particular lens or sensor will appear in a print.

I think your rant is aptly named "rant," because I think there is an enormous amount of exaggeration of what is essentially a triviality.

If the images we see onscreen really told us "little" about how they would appear in print, then what would be the point of having color-calibrated monitors, and then using our monitors to select from our entire montage of images WHICH photos we're going to print? There would be none.

If we took your logic of the unrelatedness of onscreen-image-to-print to the extreme, then we I guess we shouldn't use or depend our monitors to judge our photos at all; I suppose we should just print every image out that we ever take, and then base our analyses on how each image looks in print. This of course is ludicrous.

The truth is, the view that we see onscreen (on a properly-calibrated monitor) tells us ALOT about how our image is going to look on print (not "a little"), and this is why 100% of all photographers use their computer screens for this purpose, including yourself.

What really deserves the monker of "little" in this discussion is the amount of variance there actually is post-print. Sure, there may be tiny nuances in the final printed images, not seen onscreen, but there will be far more similarities than differences. Maybe these subtleties in the final print do warrant a subject for discussion ... but they are hardly worth a rant and litany against all onscreen evaluations. In the end, the biggest fallacy of all is to say that the images we see onscreen "tell us little" about how they will look in print; this is nothing more than an over-exaggeration. If anything, many of the errors seen blown-up onscreen don't show up in the final print.

Jack

[ErikKaffehr]

Hi,

Ray, your reasoning seems absolutely correct to me. One way to achieve that could be to take an image ProPhoto RGB image convert it to sRGB and than reconvert it to ProPhoto RGB. That would give us two images one with ProPhoto Gamut and another one beeing in ProPhoto RGB but with sRGB contents. Perhaps they would be identical on an sRGB screen and they may print different.

Best regards
Erik

...

Even my old Epson 7600 claims to be able to print a few shades and hues which are contained within the ProPhoto RGB gamut and which cannot be displayed on the average monitor, or perhaps even any monitor. Epson have now apparently increased that gamut capability with there latest printers such as the 7900.

In other words, if the two images have an embedded ProPhoto RGB profile and are adjusted on the monitor in the ProPhoto RGB working space so that they both look as identical as possible, they may not actually be identical because there are subtle hues which the monitor cannot display but which a good printer, such as the Epson 7900, can reproduce.

[ErikKaffehr]

Hi,

There are many things to discuss. In a sense screens and prints are very different:

- Screens are high contrast while prints are low contrast
- Color spaces are vastly different
- Screen resolution is discrete there is no smaller detail than the pixels themselves

- If we look at low resolution images we don't see detail, there is no way to say if a Canon L lens is sharper than a "cheapo" zoom in 600x800 images.
- If we look at actual pixels we enlarge the smallest picture elements extremely, we would never observe that level of detail in prints because our eyes are not that very sensitive to detail. Our visual impression is dominated by feature sizes of perhaps 0.3 mm. For this reason minor differences visible on screen may just disappear in print, although they may show up when the print is inspected with a magnifier.
- The actual pixel view exaggerates the differences

Summing up:

- Small images contain very little information about pixel level image quality.
- Actual pixel view shows differences that cannot be see in print.
- IMHO an image that is sharper at actual pixel will never be worse in print than an image that is less sharp, if other parameters are constant, that is.

Best regards
Erik

I think your rant is aptly named "rant," because I think there is an enormous amount of exaggeration of what is essentially a triviality.

If the images we see onscreen really told us "little" about how they would appear in print, then what would be the point of having color-calibrated monitors, and then using our monitors to select from our entire montage of images WHICH photos we're going to print? There would be none.

If we took your logic of the unrelatedness of onscreen-image-to-print to the extreme, then we I guess we shouldn't use or depend our monitors to judge our photos at all; I suppose we should just print every image out that we ever take, and then base our analyses on how each image looks in print. This of course is ludicrous.

The truth is, the view that we see onscreen (on a properly-calibrated monitor) tells us ALOT about how our image is going to look on print (not "a little"), and this is why 100% of all photographers use their computer screens for this purpose, including yourself.

What really deserves the monker of "little" in this discussion is the amount of variance there actually is post-print. Sure, there may be tiny nuances in the final printed images, not seen onscreen, but there will be far more similarities than differences. Maybe these subtleties in the final print do warrant a subject for discussion ... but they are hardly worth a rant and litany against all onscreen evaluations. In the end, the biggest fallacy of all is to say that the images we see onscreen "tell us little" about how they will look in print; this is nothing more than an over-exaggeration. If anything, many of the errors seen blown-up onscreen don't show up in the final print.

Jack

[Ray]

Ray, your reasoning seems absolutely correct to me. One way to achieve that could be to take an image ProPhoto RGB image convert it to sRGB and than reconvert it to ProPhoto RGB. That would give us two images one with ProPhoto Gamut and another one beeing in ProPhoto RGB but with sRGB contents. Perhaps they would be identical on an sRGB screen and they may print different.

Erik,
Good idea! What I've done is take the saturated swatches in Photoshop to create an image of colored dots in a new document with embedded ProPhoto profile. I then duplicated the image and converted the duplicate to sRGB. I then applied the Bill Atkinson profile for Epson Enhanced Matte in proof setup, ticked gamut warning and brought all colors within gamut by selecting each color with the magic wand and reducing the saturation of that color only, by an amount just sufficient, but no more than sufficient, to bring the color back within gamut.

I then reconverted the sRGB image to ProPhoto and printed them together, side by side on the same sheet of paper, using absolute colorimetric rendering intent.

What I've found is that the saturated CMYK colors (red, yellow and green in the lower portion of the image below) are very similar whether sRGB or ProPhoto. The RGB colors (red, yellow and green in the upper portion of the image) differ the most. ProPhoto RGB-yellow looks clearly a more solid and saturated yellow than sRGB. ProPhoto RGB-red seems to have less yellow and the green looks more blue.

These differences are equally visible on both my monitor and the print. I haven't come across any significant color differences so far that appear on print but are not apparent on my monitor. Is there a flaw in my methodology, or have I just got a good monitor well calibrated? It's an old-fashioned Sony Trinitron.

Do you see the differences that I see in the image below?

[attachment=16953:sRGB___P...to_space.jpg]

Edit: Forgot to mention. The above image has an embedded ProPhoto profile. You'll need to open it in Photoshop to see what I'm talking about.

[ErikKaffehr]

Ray,

I see a difference on the colors in the top and the middle on both of my color calibrated monitors, the first monitor is sRGB and the other one essentially AdobeRGB, I have no comments on your methodology and did not print.


Best regards
Erik

Erik,
Good idea! What I've done is take the saturated swatches in Photoshop to create an image of colored dots in a new document with embedded ProPhoto profile. I then duplicated the image and converted the duplicate to sRGB. I then applied the Bill Atkinson profile for Epson Enhanced Matte in proof setup, ticked gamut warning and brought all colors within gamut by selecting each color with the magic wand and reducing the saturation of that color only, by an amount just sufficient, but no more than sufficient, to bring the color back within gamut.

I then reconverted the sRGB image to ProPhoto and printed them together, side by side on the same sheet of paper, using absolute colorimetric rendering intent.

What I've found is that the saturated CMYK colors (red, yellow and green in the lower portion of the image below) are very similar whether sRGB or ProPhoto. The RGB colors (red, yellow and green in the upper portion of the image) differ the most. ProPhoto RGB-yellow looks clearly a more solid and saturated yellow than sRGB. ProPhoto RGB-red seems to have less yellow and the green looks more blue.

These differences are equally visible on both my monitor and the print. I haven't come across any significant color differences so far that appear on print but are not apparent on my monitor. Is there a flaw in my methodology, or have I just got a good monitor well calibrated? It's an old-fashioned Sony Trinitron.

Do you see the differences that I see in the image below?

[attachment=16953:sRGB___P...to_space.jpg]

Edit: Forgot to mention. The above image has an embedded ProPhoto profile. You'll need to open it in Photoshop to see what I'm talking about.

[MarkKay]

I think Michael makes some valid points in the Rant article. While monitors can help in certain comparisons, I agree with what Jack Flesher states in an earlier post.   However,  I am wondering whether or not there are hands on  reviews comparing monitors (after careful and accurate calibration),  which give the best results in terms of being able to display  some of the more subtle differences (e.g. widest gamut, tones, detail etc) discussed in this thread.  I suspect no single monitor is best for all. Perhaps I have just not seen such reviews.

[Jack Flesher]

If you want to see small gamut shortfalls in print, an easy way to do it is this:

1)  download this printer evaluation image (it's a 40MB, Pro RGB tiff): http://www.outbackprint.com/printinginsigh...i048/essay.html

2) It is in Prophoto RGB and already sized to fit on a letter-size sheet,  so save a native copy somewhere safe, then print it out properly using whatever your favorite paper and profile combination is -- I recommend Epson's Premium (or Canon's or HP's) Luster as it has a larger gamut than most matte papers and is readily available. Go ahead and use Epson's (or HP's or Canon's) canned profile for your printer if you don't have a custom one.

3)  Now convert the test image to sRGB and print the sRGB version on the same paper using the proper paper profile -- the only difference is this time you'll be printing from sRGB converted to your paper profile, not Prophoto RGB to your paper profile as above.  

4) Now compare the prints side-by-side.  You should easily be able to see the poor rendering of the sRGB print on the greens and blues of the test patch squares just under the grayscale image in the test image.  If you can't see any differences there, you are not printing properly or your printer/paper combo has an inordinately small gamut to begin with...

Cheers,