[photoptimist]

😁 Yes, but this is also a 720 PPI or even 1440 PPI printer making 1mm section 28 or 57 pixels wide respectively. Other variable factors that may help to clarify listed below

The 6x4 print was just a quick test or headclean print I had to hand and a massive downsize of a sample from a 5 or 7 frame pano stitch from a 36 MP camera. So it would most likely have been printed at 720 ppi due to an over abundance of pixels at this size not sure what quality settings so DPI/quality unknown. Paper type glossy so this would also have a bearing on DPI/quality settings.

This particular printer has a printhead with 180 nozzles x 8 colour and monochrome heads. Max resolution 2880x1440 DPI

Don’t forget that the droplets will potentially vary in volume from minimum of 3.5 picolitres to an undisclosed volume up to 3 different sizes per line. Plus spread and absorption of liquid onto/into various substrates makes accurate size calculation awkward. You may also need to factor in ambient temp and humidity of environment and media for absolute accuracy - OK a step too far! 😩

Indeed! There are many possible steps that seem to be too far. And yet if we are draw lines in the printer resolution sand, it's important to understand where that line should be drawn and how it might be easy to draw the line in the wrong place.

In all these discussions, there is the distinct possibility that some of the judgements about the maximum useful printing resolution come from images, printers, or printing conditions that are not capable of producing a higher resolution image.

For example, a soft photo may well look no better at 300 PPI than it did at 150 PPI because the original image simply lacked any detail at the higher resolution. Or a truly sharp photo may look no better at 300 PPI because the particular printer's 300 PPI spec is bogus marketing. Or a sharp image on a high-performance printer might not look better at higher resolution because the choice of paper or temperature/humidity conspired to blur the droplets.

There's a very long chain of steps from original subject to subjective judgement of the print quality that includes:

• The Subject:
• subject matter (existence of contrast, edges, and fine details)
• atmospheric conditions (haze + turbulence)
• The Gear:
• lens (& accuracy of focus)
• sensor (low pass filter + Bayer sampling)
• The Software:
• RAW development (de-mosaicing +JPG)
• post processing
• printer driver rendering (resampling+dithering)
• The Printing:
• printhead (nozzle pitch, droplet spacing, droplet volume)
• paper+temperature+humidity (dot gain and absorption)
• The Final Assessment:
• the human eyeball

With few exceptions*, each step in the chain degrades the resolution of the image. In general, the MTF curves of each step multiply down the chain so that even if each step passes a resolution test at say 300 PPI, the chain might fail the test because the highest frequencies in the image were steadily degraded at each step. The point is that deciding the print resolution hinges on knowing all the other resolution-degrading effects in the rest of the chain.

It's one thing to say my images with my lenses, my printer and my eyeballs don't benefit from resolutions greater than XXX PPI. But establishing that no possible image, lens, camera, printer, and viewer combination benefits from resolutions greater than XXX PPI is a much stronger assertion. For myself, I can only say that some images (e.g., B&W text) look better at 600 PPI.


*NOTE: post processing can increase resolution with some limits defined by the Bayer sampling and dynamic range of the sensor. "Sharpening" has a high-frequency MTF value that is greater than 1 but has the side effect of amplifying noise in the image

[TonyW]

Indeed! There are many possible steps that seem to be too far. And yet if we are draw lines in the printer resolution sand, it's important to understand where that line should be drawn and how it might be easy to draw the line in the wrong place.

In all these discussions, there is the distinct possibility that some of the judgements about the maximum useful printing resolution come from images, printers, or printing conditions that are not capable of producing a higher resolution image.

For example, a soft photo may well look no better at 300 PPI than it did at 150 PPI because the original image simply lacked any detail at the higher resolution. Or a truly sharp photo may look no better at 300 PPI because the particular printer's 300 PPI spec is bogus marketing. Or a sharp image on a high-performance printer might not look better at higher resolution because the choice of paper or temperature/humidity conspired to blur the droplets.

There's a very long chain of steps from original subject to subjective judgement of the print quality that includes:

• The Subject:
• subject matter (existence of contrast, edges, and fine details)
• atmospheric conditions (haze + turbulence)
• The Gear:
• lens (& accuracy of focus)
• sensor (low pass filter + Bayer sampling)
• The Software:
• RAW development (de-mosaicing +JPG)
• post processing
• printer driver rendering (resampling+dithering)
• The Printing:
• printhead (nozzle pitch, droplet spacing, droplet volume)
• paper+temperature+humidity (dot gain and absorption)
• The Final Assessment:
• the human eyeball

With few exceptions*, each step in the chain degrades the resolution of the image. In general, the MTF curves of each step multiply down the chain so that even if each step passes a resolution test at say 300 PPI, the chain might fail the test because the highest frequencies in the image were steadily degraded at each step. The point is that deciding the print resolution hinges on knowing all the other resolution-degrading effects in the rest of the chain.

It's one thing to say my images with my lenses, my printer and my eyeballs don't benefit from resolutions greater than XXX PPI. But establishing that no possible image, lens, camera, printer, and viewer combination benefits from resolutions greater than XXX PPI is a much stronger assertion. For myself, I can only say that some images (e.g., B&W text) look better at 600 PPI.


*NOTE: post processing can increase resolution with some limits defined by the Bayer sampling and dynamic range of the sensor. "Sharpening" has a high-frequency MTF value that is greater than 1 but has the side effect of amplifying noise in the image

Well, put

[biz-engineer]

I have an image printed with 125 PPI: 1 basic crop, 1 basic crop sharpened, 1 basic crop upsampled to 300 PPI equivalent and sharpened, printed by the same printer on the same glossy paper. It is hardly possible to see any difference between the 3 prints and there is no way my eye can see any printer dot or pixelization, each of the 3 prints on glossy paper look smooth with nice colors. BUT.... none of the 3 prints from the 125 PPI crop look as neat as the prints from the 180 PPI crop, and I think the reason is as simple as the higher enlargement factor from cropping less pixels in the original image: the sampling by the camera sensor and re-sampling by the printer have a bandwidth that is beyond the detail of the image itself when cropped and magnifying by the printing at a given size on paper. If I crop 2Mpixels and print them on A4 paper, or crop 1Mpixels and print on A4, the enlargement is not the same, up-sampling before printing doesn't change the enlargement factor. At some point, it doesn't matter how many PPI and DPI are used for the print, if the enlargement is such that the image is already soft.

---------- Post added 01-05-19 at 23:00 ----------

atmospheric conditions (haze + turbulence)

lens (& accuracy of focus)

That's correct. I selected 5 images for printing out of thousands of images taken over the last few years... a lot of photographs are just fine on a small computer display but they can't make a good A1 sized print, for various reasons:
- some images taken at f16 and f22 to get enough depth of field good enough for the foreground and background, can make a detailed A2 or A1 print, because muddy pixels after diffraction
- some images are cropped already (such as when the bird wasn't close enough to fill the frame)... the enlargement is limited
- some actions images where the subject motion created blur at the pixel level, fast shutter speed reduce the amount of blur of a moving subject but it does not eliminate the blur completely...

If somebody want to print a poster with high quality, the shooting conditions must be the best possible: e.g tripod, still subject, no wind, no crop, f-stop not too high etc..

[TonyW]

I have an image printed with 125 PPI: 1 basic crop, 1 basic crop sharpened, 1 basic crop upsampled to 300 PPI equivalent and sharpened, printed by the same printer on the same glossy paper. It is hardly possible to see any difference between the 3 prints and there is no way my eye can see any printer dot or pixelization, each of the 3 prints on glossy paper look smooth with nice colors

What you appears to describe is to be expected. You should not see any printer dots or pixelation between any of these due to the fact that the printer treated them all as required by the driver. If the printer Epson then all images would be upsampled to Epsons proprietary bitmap if 360 PPI and exactly the same dot pattern will be applied.

. BUT.... none of the 3 prints from the 125 PPI crop look as neat as the prints from the 180 PPI crop, and I think the reason is as simple as the higher enlargement factor from cropping less pixels in the original image: the sampling by the camera sensor and re-sampling by the printer have a bandwidth that is beyond the detail of the image itself when cropped and magnifying by the printing at a given size on paper.

The reason is likely much simpler and I am guessing the printer Epson? Best printing Epson is 360 PPI or multiples 720, 1440 or half the standard resolution, your 180 PPI

At some point, it doesn't matter how many PPI and DPI are used for the print, if the enlargement is such that the image is already soft

Absolutely

I posted some comparisons of printing at different resolutions on the cheapest Canon A4 printer sending image data at different levels. Prints were scanned together and illustrate the potential differences from sending non standard data to using the printers recommended settings.
https://www.pentaxforums.com/forums/32-digital-processing-software-printing/...r-best-iq.html

---

Last edited by TonyW; 05-01-2019 at 02:44 PM.

[biz-engineer]

I posted some comparisons of printing at different resolutions on the cheapest Canon A4 printer sending image data at different levels. Prints were scanned together and illustrate the potential differences from sending non standard data to using the printers recommended settings.https://www.pentaxforums.com/forums/32-digital-processing-software-printing/...r-best-iq.html

Ah but that is a comparison with inkjet technology. I've recently realized the difference between Inkjet and C-prints (Lambda, Lighthet on silver paper).
When comparing the cost of printing at home and printing at a lab, it became obvious that the cost of C-print tech is routhly half of the cost of printing with Inkjet, and resolution figures aren't the same. Apparently, modern Inkjet now gets the edge over C-print in terms of sharpness and color gamut. C-print is cheaper and mostly used by photo printing labs.

[TonyW]

Ah but that is a comparison with inkjet technology. I've recently realized the difference between Inkjet and C-prints (Lambda, Lighthet on silver paper).
When comparing the cost of printing at home and printing at a lab, it became obvious that the cost of C-print tech is routhly half of the cost of printing with Inkjet, and resolution figures aren't the same. Apparently, modern Inkjet now gets the edge over C-print in terms of sharpness and color gamut. C-print is cheaper and mostly used by photo printing labs.

The comparison is still valid, just different method and media using the same principles of PPI and DPI

Cost of printing differences may well be much lower (depends on the skill and service level of the lab!)

There is no particular reason for any printer to follow 300 - 360 PPI as input resolution (other than it is close to calculated visual acuity figures (20/20 vision) for seeing a contone image from dots)

The point is that for a specific file its native pixel count will allow you to print at a maximum certain size depending on printer and prior to any interpolation. So for a 6000x4000 ppi image resolution for different printers you will get a different maximum print size:

300 PPI printer e.g. Canon/HP = 20" x 13.3"
360 PPI printer Epson = 16.6" x 11.11"
400 PPI Durst Lambda (C Type) = 15" x 10"

Now of course we may require a print much larger or smaller than the max sizes quoted. In that case up or down sampling required resuting in lower or higher PPI figures. C Type or inkjet it does not matter as the current native resolution ppi will be up/downsampled to what the printer needs. DPI applied relating to quality settings


If you examine the specifications of C type printers you will find that they quote two resolutions. The first may be called the optimum input resolution 200-300-400 ppi (they may use confusingly dpi). This is the same as inkjet in as much as a particular printer requires a certain amount of pixels per inch to offer the best quality. The second figure is the output resolution as RGB continuous tone and these figures may be over 4000 dpi but quoted as equivalent (compared to inkjet). This is similar to an inkjet printer squirting many dots of different coloured ink to form a particular colour on paper, but in this case it will be shooting coloured light of varying intensity R, G, B to form the colour image of a pixel to the photographic paper.

[biz-engineer]

I've ordered and received some test prints on a few different paper types. The paper plays a role in the rendering of details. Glossy / metallic finish offer deeper blacks. One of the prints is of a photographs of a vintage car captured at an old timers car race with the K1 and DFA70-200 at f2.8, the focus on the car is spot on (blurred background), the car has some shiny chrome parts, it's printed black & white on metallic finish paper (silver paper process) and with proper lighting it really shines, more that on calibrated 4K monitor! Lighting of the print is essential to avoid glare, with a dedicated light at the top or bottom of the frame, it looks like the car in slightly out the frame (3D effect). The trouble is, I haven't printed the same vintage car photo on mat paper, so I don't know how much of the 3D effect is produced by the metallic finish vs what the 3D effect would be with a mat paper (the mat paper relaxed the constrain of angle of light... that's the advantage).

[TonyW]

Viewed in isolation either high gloss or matt print should look good for many subjects - just don't try to compare them side to side. As you have already identified glossy finish paper usually offers a greater Dmax and can appear to resolve more detail than a matt surface. My preference tends to be for semi gloss, lustre finishes as viewing angles and lighting not quite as critical as high gloss


If your lab offers icc profiles for its range of papers then you may get some idea of the contrast loss by soft proofing if you own a suitable application (PS or LR)

[biz-engineer]

As you have already identified glossy finish paper usually offers a greater Dmax and can appear to resolve more detail than a matt surface.

You are correct. I've looked at that is because I wanted to display framed printed but without glass to avoid distracting reflection and reveal the character of the paper. What I also noticed is that mat paper still reflect light if the main light source comes from the back of the viewer. Then I looked at how lighting is done in photo galleries, using light spots fixed at the ceiling pointed at the print/frame, and the frame positioned around eye level of the average viewer. And yes, even high gloss paper doesn't show any reflection is the only light source is at an angle, but still pop. Having a dedicated lighting for prints also allow to use glass without reflections. Photo galleries have it setup the best way. Without proper setup for lighting of the prints, the choice of paper is more limited, it is trivial to find papers that feature both high Dmax and low gloss, luster paper is a good trade-off.

[Take-5-JB]

This discussion has been excellent...Photoptimist, TonyW, and Biz-engineer. Up to the point of non-published proprietary information, we have a nice breakdown of the physics, mechanical, and optical considerations for levels of viewing pleasure for printed images. I have filled in some information gaps in my understanding. I am a chemist by education and 39 years working as an Industrial Chemist. I tend to think about spectroscopy and spectral identities such as chromaticity curves, color wavelengths, and spectral lines which are the physical basis of color and its manipulation limits. Color is a spherical world and color "systems" or descriptions are line segments or curves transecting that sphere. When the color palette is chosen that is used to "mix" colors to result in colors generated in the image and displayed by the ink system, the physical/mechanical system must be adapted to handle how light interacts with the chosen color palette. And that does include the dot matrix of the printheads, substrate (paper, rag, canvas) architecture, viewing light temperature, viewing light angle, distance of the viewer, cover quality (polymer coatings for canvas, metal prints, or framing glass or plexiglass, with UV management, reflection control, and other considerations). Ultimately, the quality of the viewing pleasure rests with the individual perception, eye to brain translation, neural fill-in, and artistic attribute (left brain vs right brain). Thanks for your contributions to our overall knowledge. FYI, I use bicubic image resizing algorithm base software with good results.

JB

[biz-engineer]

There is another issue with printing on paper. Color cast from the paper and from the light source. The color temperature of the light source should preferably close to daylight, however, most light bulbs sold for housing are either to warm (<4000K) or too cool (6500K), it is possible to find 5000K bulbs but it's not very common and not necessarily cheap. As for the color cast from the paper, I'm printed a gray calibration chart , then photographed it under correct WB to correct for the light source, and reprinted color shifted B&W prints but it didn't work well with inkjet printing. The problem is that the amount of ink isn't constant over the print areas depending on the image content. For example, when a could is printed with inkjet tech (additive process), it is the absence of ink that leave out the white of the paper uncovered therefore impossible to correct with ICC profiles or a color shift applied on the image file before printing. The Lambda silver process is a substractive, active color correction would work better. With inkjet tech it is almost impossible to obtain pure B&W. So at this point, I'd say: B&W => silver process, Color => Inkjet.

[Fenwoodian]

So at this point, I'd say: B&W => silver process, Color => Inkjet.

.

I gave a thumbs up for this brilliant advice -- but, didn't feel a simple thumb's up went far enough!

The above comments are the best printing advice I've heard in years! Heck, it pretty much sums up the last 40 years of photography for me.

[TonyW]

Sorry but the very purpose of ICC profiles IS to accurately describe a printer, ink and paper combination. A good profile will do this and give you a very good print to screen match in a properly colour managed system using soft proofing. This includes evaluation of the print under proper conditions of illumination - many miss this important step.

Initial print illumination for evaluation is probably best using Soloux 4700k or similar, however if you know the exact end viewing conditions you can mimic this
during print run.

There really should be no issue printing C type or inkjet in B&W, these are colour materials.

The biggest problem most face in trying to match screen to print is that they are not printing themselves and many labs pay lip service to colour management. A good lab will supply an ICC profile and also allow you to soft proof and embedded the paper profile and importantly select the rendering intent in you data file

[biz-engineer]

A good lab will supply an ICC profile and also allow you to soft proof and embedded the paper profile and importantly select the rendering intent in you data file

I've printed from three different labs. Two of them provide ICC profiles for soft proofing. I've send the same B&W file to each of the three labs. I received three prints:

- None of the three B&W prints were aligned, I could see the difference immediately.
- For the two companies that provide ICC profiles, only one inkjet print is close to what the ICC soft proofing give to my display (my display is calibration over 4 hours, 100% matching of sRGB volume), the inkjet print from the other company contains a magenta tone. I've filed a customer complain sent them a shot with a gray card in front of the print to show the difference, but they didn't care much, they offered me a voucher for ordering more prints.(more B&W print with the same tone shift? laughs... I get the customer support lady did not understand my concern).
- The third company using silver base print , but provide no ICC profile , give the best B&W print with not tone shift, and their prints are cheaper...

---------- Post added 26-05-19 at 06:51 ----------

A good profile will do this and give you a very good print to screen match in a properly colour managed system using soft proofing.

The problem of soft proofing is that it relies on monitor calibration and viewer appreciation, which is very subjective. I've found another way to evaluate prints, using my camera with WB calibration on X rite gray card, and shooting an 18% gray print under the same light at the same moment, the histogram clearly show the tone shift but is it not possible to fully correct for it with inkjet if the tone shift come from the paper because ink will only correct the color in areas of prints where there is a lot of ink... That's why I wrote that the silver process is better.

---------- Post added 26-05-19 at 06:58 ----------

The above comments are the best printing advice I've heard in years! Heck, it pretty much sums up the last 40 years of photography for me.

I'm not sure if inkjet is best when the prints have only a few colors. However, by now, some inkjet printed have 9, 10, 12 cartridges, exceeding sRGB, better for color prints vs silver based photographs.

[TonyW]

You have already made errors in the way you think the print process works and PPI vs DPI and I am afraid you are also making errors with the following assumptions. However I will try but it is going to be difficult as you are not actually in control of the process passing to certain labs

I've printed from three different labs. Two of them provide ICC profiles for soft proofing. I've send the same B&W file to each of the three labs. I received three prints:

Providing ICC profiles is only the start, they need to actually encourage you to use and embedd the profile in your edited image where you will also have chosen a rendering intent. Did they tell you to do this or just ask you to supply an sRGB JPEG?

- None of the three B&W prints were aligned, I could see the difference immediately.

No surprises there then as they probably do not follow colour management practices preferring instead to send 'acceptable' prints

- For the two companies that provide ICC profiles, only one inkjet print is close to what the ICC soft proofing give to my display (my display is calibration over 4 hours, 100% matching of sRGB volume), the inkjet print from the other company contains a magenta tone. I've filed a customer complain sent them a shot with a gray card in front of the print to show the difference, but they didn't care much, they offered me a voucher for ordering more prints.(more B&W print with the same tone shift? laughs... I get the customer support lady did not understand my concern).

Display calibration of over 4 hours makes zero sense. Display calibration should not take long at all once you have set your required display standards. You monitor may be capable of displaying the full sRGB gamut however you do not calibrate to sRGB you calibrate to standards e.g. White Point (D65), Black Level, Gamma and luminence. The latter based on your ambient light and print viewing conditions including correct illumination of the print.
As a matter of interest what standard values are you using?
What is your ambient editing environment like, consistent, variable, dull, bright?
When viewing print next to monitor, how are you illuminating the print?

- The third company using silver base print , but provide no ICC profile , give the best B&W print with not tone shift, and their prints are cheaper...

If they were using C type photo paper then AFAIK this is still a colour process requiring equal RGB values to produce a neutral depending on paper substrate colour it should be easy to achieve



---------- Post added 26-05-19 at 06:51 ----------

The problem of soft proofing is that it relies on monitor calibration and viewer appreciation, which is very subjective. I've found another way to evaluate prints, using my camera with WB calibration on X rite gray card, and shooting an 18% gray print under the same light at the same moment, the histogram clearly show the tone shift but is it not possible to fully correct for it with inkjet if the tone shift come from the paper because ink will only correct the color in areas of prints where there is a lot of ink... That's why I wrote that the silver process is better.

There is nothing subjective about soft proofing if your colour management is working quite simply soft proofing works, unfortunately yours at least from what you say is broken

---------- Post added 26-05-19 at 06:58 ----------

I'm not sure if inkjet is best when the prints have only a few colors. However, by now, some inkjet printed have 9, 10, 12 cartridges, exceeding sRGB, better for color prints vs silver based photographs.

Inkjet prints can exceed areas of Adobe RGB