Print quality considerations 180PPI and 360PPI where are they coming from…

[ErikKaffehr]

Hi,

There are some well known rules of thumb, one is that 180 PPI is enough for an excellent print, but we also know that Epson printers are capable of 360PPI or 720PPI native resolution. Here are some perspectives on that issue. It is not necessary knowledge, but it can be nice to know…

An important factor is human vision. Normal 20/20 human vision has maximum resolution at 25 cm. That resolution is pretty exactly corresponding to 360 PPI.

Moving closer does not increase resolution with normal vision. Some people have myopic vision, so they can see increasing detail at short distances. Ideal reading distance may be say 40-60 cm, or 15"-25". Assuming 50 cm distance the eye would normally resolve 180 PPI. There are of course variations of human vision.

The next factor is the contrast sensitivity of human vision. The 20/20 vision corresponds to a resolution of one minute of arc, or 30 (?) cycles per degree, the maximum contrast sensitivity is around 7 cycles per degree:

What do we consider to be sharp?

Without referring to sources, research has found that we need about 35% MTF to achieve visual sharpness. So my guess is that would like to have 35% contrast on fine detail (at vision limit) and probably much higher contrast at lower frequencies.

Another factor is "vernier acuity", human vision is very sensitive for broken lines. So very small shifts along a line can be observed, although the resolution is far beyond the normal 1 minute of arc resolution for 20/20 vision.

Here is a good discussion of the underlying issues: Acutance and SQF (Subjective Quality Factor) | imatest

This presentation by Larry Thorpe of Canon explains a few things about MTF: https://youtu.be/iBKDjLeNlsQ

Mr. Thorpe's presentation is about motion pictures but it also applies to stills…

John Galt of Panavision expands a bit on that issue: https://www.youtube.com/watch?v=v96yhEr-DWM

Best regards
Erik

 

[Paul2660]

I guess it all depends on your own experiences with printing and what you determine is the correct ppi for a print based on the printer and driver.

For me, the matter is quite simple, as I use Epson driver and not a RIP, thus I would prefer to have my work printed at a ppi that Epson's driver is going to use, i.e. 360 or 720. It's a know fact that work not sent at one of these ppi settings will be up sampled to one of them by the driver and the driver uses by far the worst possible up sampling algorithm, nearest neighbor, and you have no visible control on the output. Unless Epson has changed some of this in later printers, like the P9000 or P800 I believe it's still the norm.

Thus, I use LR and allow it to upsample with it's own algorithm before sending this to the driver and printing from windows. This technique has worked fine for me on prints up to 36" x 96, the largest single print I have made on an Epson on canvas and the 9900.

For work larger than this, 75" x 120" and 6 feet by 12 feet, I will send the work at a lower ppi, in this case 180 as to me the attempts to up sample to 360ppi such large images, is pretty much a waste as the 180ppi images appear to my eye and printed swatches of 30" x 40" to look better than the swatches of 30" x 40" from 360ppi, but again this is an individual preference, and one that has been suitable for my customers, (hospitals and office complexes that have required such large images).

For me the answer is understanding your printer and it's driver and the best way to work with them to get the best possible results. Spending an afternoon reading through the "Digital Print" by Jeff Schewe, is well worth the time. Jeff clearly explains the relationships between ppi, the printer driver, and the final print.

The other option is of course to use a RIP, which by passes totally the Epson printer driver, but that adds a lot more cost by the time you reach a 44" printer and have multiple printers operating.

Printing is something just like taking a photograph, in that no two people are going to do exactly the same way. In my case I have taking the knowledge of people like Jeff Schewe and Bill Atkinson (both accepted experts in the field) and developed my workflow.

Paul C

 

[ErikKaffehr]

Hi Paul,

The idea with my posting is to give some background to the numbers often used. So it is more about "why" than about "how".

There is a reason I posted it on the MFD forum and that reason is that I think that MFD users are often interested about how large you can print from a certain size of image.

Best regards
Erik

I guess it all depends on your own experiences with printing and what you determine is the correct ppi for a print based on the printer and driver...

/QUOTE]

 

[jlm]

eric: you are un-necessarily repeating an entire post that on my screen is only about 3" away.

just use the "reply to thread" button, not the "reply with quote" button:grin:

 

[ErikKaffehr]

Hi,

Lesson taken… I have edited my response.

On the other hand, I seriously hate going trough 20-30 postings to find out what a poster is responding too, so I think a quote is a better way to respond to a specific posting then replying to the thread.

"He, who is without sin, may throw the first stone…"

Best regards
Erik

eric: you are un-necessarily repeating an entire post that on my screen is only about 3" away.

just use the "reply to thread" button, not the "reply with quote" button:grin:

 

[jlm]

3" again

 

[MGrayson]

Actually, just quoting a line is enough, as the quote box contains a link to the original. One can quickly click back through a conversation to find the beginning.

--Matt

 

[kdphotography]

3" again

Geez, John----I thought only girlfriends were allowed to complain about 3"... :ROTFL: It was the pool. The water was quite cold and it was chilly outside.

Ok, back to work. I've had this new Epson P9000 in studio for a week and I still haven't installed it....

 

[Egor]

I am not sure how to cross-post from the thread about 60" prints from a Credo-60; but I gave some info on why the numbers or "rules of thumb" as Erik referred to are so often used. They are not relevant to modern FM screening techniques used and I think their history is worth mentioning.
They were derived from the early days of offset printing using AM screening at average of 150LPI (Lines Per Inch)

I will just paste my last response in the other thread here as it may be of help:

Try not to confuse PPI (Pixels Per Inch=camera resolution) with DPI (Dots Per Inch=printer resolution)

They are two completely different things utilizing two completely different color models and representations of continuous tone images. THEY ARE NOT RELATED

Frequency Modulated screening means all dots are the exact same size and the distance between them changes to produce density, color and gradation.

All inkjet printers use FM (Frequency Modulated) screening algorithms and there is no correlation between camera resolution (PPI) and print resolution (DPI).

We fixate on numbers like 180PPI and 300PPI for arbitrary reasons like old AM offset printing techniques (don't apply) or multiples of DPI print resolution (180PPI is half the 360DPI so its an even number and we think that matters...but it doesn't apply again...its irrelevant)

Whether your digital image is 1PPI or 1200PPI it is going to be interpolated into completely different values based on FM screening algorythms

 

[jlm]

I suppose the less sophisticated of us (me), make the assumption that we would want the same number for the image ppi and the printer optimum dpi, and that up-resizing the ppi number to match is a good idea, and is best done by software other than in the printer. Since new data needs to be created, bicuspid interpolating beats out nearest neighbor.

I'm leaving in spell correct for giggles

 

[ErikKaffehr]

Hi,

It may be a good thing to keep apart resolution of the image and the PPI of the printer…

According to Jeff Schewe, native resolutions of Epson printers are 360 PPI (in 1440 DPI mode) and 720 PPI (in 2880 DPI). My initial posting discusses a bit why Epson has chosen 360 PPI, and the reason is in all probability that 360 PPI corresponds pretty well to 20/20 vision at 25 cm.

Still according to Jeff, if an image is sent to the printer at PPI different from the native ones the printer will upres or downres using a weaker algorithm than bicubic, probably nearest neighbour. So Jeff recommends that the image is interpolated to 360PPI or 720PPI depending on pixel size, using a proper algorithm.

The interpolation does not add resolution, but can make transitions smoother. Interpolation will introduce artefacts, a good interpolation will introduce fewer artefacts. Bart van der Wolf used to have a lot of good info on these issues, but his web-site is moved so I cannot give links to that.

These articles may offer some insight Downsizing Algorithms: Effects on Resolution | Strolls with my Dog and Comparing downsampling algorithms — ISO 12233 | The Last Word .

As said before, interpolation does not add information just makes transitions smoother. The image printed needs to have the native resolution to match the print size and viewing distance.

Resolution of human vision (in lp/mm) maxes out at around 10" (25 cm) viewing distance, so looking closer the viewer actually sees less details. That distances increases with age, but that can be corrected by corrective lenses. 20/20 vision will resolve around 360 PPI at that distance.

My understanding is that practical evidence indicates that 180 PPI in image resolution is good enough for excellent prints. I would suggest that Leica S system owners can make excellent 42"x28" prints, due to the high MTF of the Leica lenses and they would claim that those prints are good enough for close scrutiny. That print size corresponds to the 180 PPI mentioned.

Best regards
Erik

I suppose the less sophisticated of us (me�&#56839, make the assumption that we would want the same number for the image ppi and the printer optimum dpi, and that up-resizing the ppi number to match is a good idea, and is best done by software other than in the printer. Since new data needs to be created, bicuspid interpolating beats out nearest neighbor.

I'm leaving in spell correct for giggles

 

[Egor]

camera sensors and monitors are AM (Amplitude Modulated screening which means a grid is laid down and the distance between dots is always exactly the same, only the size (amplitude) of the dot within the grid squares changes size. Within the additive color model of RGB that is how they create density, color, saturation and gradation.
This is why you can get moiré when the AM screen grid of your image is overlaid on top of the AM screen grid of your monitor, for example.

FM (Frequency Modulated screening which means that the dots stay the same exact size and just the distance between them (frequency) changes) is how FM, using the subtractive color model (cmy) produces the same density,color, saturation and gradation in reflective printing process. Moire is impossible with FM screening because there is NO grid to overlay.
This is also how film works, btw, so moiré impossible with film, until it went to standard AM offset printing methods anyway.

How are you correlating a given PPI of resolution using additive color model; to a random collection of FM dots using subtractive color model absorbing light on a given substrate?

So Epson likes 360DPI or 720DPI or 1440DPI...etc and Canon likes 300DPI and 600DPI for its CT FM screening but with all due respect, neither correlates to resolution of camera sensor; they are simply FM screening dot sizes all of which are too small to be perceptible by any human eye.

If you have an image that is only 25PPI at size it is printed you will most likely see the square grid of the AM screen it was used for exposure or manipulation in bitmap image format. Once you exceed about 50-100PPI at size it is printed there are very few humans who can make out the grid any longer without special lenses and magnification. Look at a set of calipers and tell me you can "see" .01 inch increments with your naked eye. Now tell me you can then make out .003 in increments...

 

[ErikKaffehr]

Hi,

I think you mess things up a bit. Moiré in a sampled image is mostly an aliasing artefact coming from discrete sampling. If you have moiré in you image you will also have moiré in your print, but it may be less visible. Calling it "colour aliasing" may be more correct, though.

The image below is a good example of that. The image on the left is shot at f/8 and the one on the right at f/16. All other factors are the same. The one on the left has significant aliasing artefacts, both colour and monochrome, while the on the right has no such artefacts. The reason is that diffraction at f/16 reduces MTF at Nyquist enough to avoid aliasing. Aside from the aperture, sampling is the same as both are JPEGs shown on the same screen.



Not all printing is done on inkjets. The few large prints I have made were done on Durst Lambda, which is continuous tone on photographic paper.

If you check the original posting, it is pretty clear that the intended discussion is about needed print sizes.

I have no issue with your posting but I don't think it is relevant for a photographer trying to find out how large he can print, or what resolution would be needed for a given print size.

Best regards
Erik

camera sensors and monitors are AM (Amplitude Modulated screening which means a grid is laid down and the distance between dots is always exactly the same, only the size (amplitude) of the dot within the grid squares changes size. Within the additive color model of RGB that is how they create density, color, saturation and gradation.
This is why you can get moiré when the AM screen grid of your image is overlaid on top of the AM screen grid of your monitor, for example.

...

 

[Egor]

The resolution needed for a given print size is what this is all about. The answer is the same as was discovered over 100 years ago for offset AM screening lithography. That is that the human eye stops making out differences in dot frequencies between 50-100DPI.

That is all, but that is not what was asked here. The title of the thread asks about "180PPI and 360PPI where they come from?"

Erik, not contesting your findings either, but without a grid screen there can be no moiré.

Your sample image is an example of two grids causing moiré (camera sensor and AM printed screen of dollar bill print)

If the dollar bill was shot using film instead of a sensor there could be no moiré. diffusion or otherwise.

I was trying to answer the post title question about "where do these numbers like 180PPI and 360PPI come from?"

The first thing I wanted to do away with was the conflation of PPI and DPI

180PPI is a resolution of AM screening in camera; whereas 360"DPI" is an arbitrary print dot resolution used by Epson Corp for their FM inkjet printers. There is no correlation other than arbitrary guesswork.

Your monitor screen is at best 96DPI and I know of very few who can see the individual dots that size, yet alone the average 72DPI. Its different for everyone of course, but most lose seeing the separation of these frequencies between 50-100DPI. That is why they are those resolutions and that is the answer to the question as to print resolutions.

 

[AreBee]

Erik,

...20/20 vision corresponds to a resolution of one minute of arc, or 30 (?) cycles per degree...

Given that there are 60 minutes in one degree, "30 (?) cycles per degree" could be expressed as: 30 arcminute-pairs per degree.

 

[fotohouse]

The resolution needed for a given print size is what this is all about. The answer is the same as was discovered over 100 years ago for offset AM screening lithography. That is that the human eye stops making out differences in dot frequencies between 50-100DPI.

That is all, but that is not what was asked here. The title of the thread asks about "180PPI and 360PPI where they come from?"

Erik, not contesting your findings either, but without a grid screen there can be no moiré.

Your sample image is an example of two grids causing moiré (camera sensor and AM printed screen of dollar bill print)

If the dollar bill was shot using film instead of a sensor there could be no moiré. diffusion or otherwise.

I was trying to answer the post title question about "where do these numbers like 180PPI and 360PPI come from?"

The first thing I wanted to do away with was the conflation of PPI and DPI

180PPI is a resolution of AM screening in camera; whereas 360"DPI" is an arbitrary print dot resolution used by Epson Corp for their FM inkjet printers. There is no correlation other than arbitrary guesswork.

Your monitor screen is at best 96DPI and I know of very few who can see the individual dots that size, yet alone the average 72DPI. Its different for everyone of course, but most lose seeing the separation of these frequencies between 50-100DPI. That is why they are those resolutions and that is the answer to the question as to print resolutions.

This.