how come need so high dpi for printing photos?

"acl" <achilleaslazarides RemoveThis @yahoo.co.uk> wrote in message
news:1168952302.469729.107090@a75g2000cwd.googlegroups.com...

> But does it? My printer resolves around 480ppi on glossy paper. If I
> print at 85ppi (say), the result looks horrible.

Suppose you want 256 levels of red printed.
You need to print 0-255 dots for that red pixel.
That would be a square 16 x 16.
A 480dpi printer cannot do that at 85ppi. It could at 30ppi.

If your printer can do variable drop sizes it can do that 256 different
levels in less dots.. say 64 or 8x8.

This is a simplistic view as printers seldom actually use square pixels.
 >> Stay informed about: how come need so high dpi for printing photos? 

"Bucky" <uw_badgers.TakeThisOut@email.com> writes:
>A 15" monitor at 1024x768 resolution is about 85 pixels per inch.
>Viewing a photo on the monitor at that resolution looks great, right?

Compared to what? It really isn't all that sharp, compared to a monitor
that can achieve 200 PPI or even 150 PPI.

It's also not as sharp as a good print, but here the comparision is
confused by contrast. A decent monitor can achieve a white-to-black
contrast of several hundred to one or better, while a colour print can
manage maybe 30:1 and a good silver B&W print can do 100:1. Because
of this and other differences, images on monitors *normally* have higher
contrast than they did in the wild, while images on paper generally have
the same contrast as the original.

More precisely, if you capture an image containing two areas of similar
colour but one is 2X brighter than the other in the scene, when you
display that image on a monitor the difference will be more than 2X -
more like 2.4X brighter. On a print, the one will be only 2X brighter,
accurately reproducing the original scene contrast.

>So how come if you print out a photo with the same resolution, size,
>density (1024x768, 12"x9", 85 dpi), the print will look very poor? Even
>if you hold the print at the same viewing distance as the monitor.

Are you comparing resolution (meaning "ability to resolve detail", not
"pixel count") or contrast when you judge the print as poor?

Dave
 >> Stay informed about: how come need so high dpi for printing photos? 

"Dave Martindale" <davem.DeleteThis@cs.ubc.ca> wrote in message
news:eojk4p$d10$1@swain.cs.ubc.ca...

> It's also not as sharp as a good print, but here the comparision is
> confused by contrast. A decent monitor can achieve a white-to-black
> contrast of several hundred to one or better, while a colour print can
> manage maybe 30:1 and a good silver B&W print can do 100:1.

Glossy paper reflects about 95% of light.
The black level reflects a few percent too.

Even a good print is probably 20:1 contrast ratio at best.
To get 100:1 it would have to absorb all the light in the black areas and
reflect >99% in the white which they just don't do.
 >> Stay informed about: how come need so high dpi for printing photos? 

acl wrote:
> Bill Funk wrote:
>> On 16 Jan 2007 14:12:51 -0800, "acl" <achilleaslazarides DeleteThis @yahoo.co.uk>
>> wrote:
>>
>>> But anyway. I was not talking about how many colours each pixel must be
>>> (I don't think this is important for what we are discussing, although I
>>> might be wrong). I was rather talking about the fact that if I print at
>>> 85ppi, then it looks bad, while at (say) 300dpi it doesn't, while on a
>>> screen, it looks perfectly ok at around 80-100ppi.
>> The "dpi" in there should be "ppi".
>> The reason 85ppi on your monitor looks good, while it doesn't on
>> paper, is because the media is different. It's really that simple.
>
> Typo, sorry (well, I think that it was obvious that it's a typo, but
> what do I know).
>
> As for the reason a monitor looks ok at 85ppi but a print doesn't, I
> think the whole thread is an attempt to understand why these different
> media look so different. Saying that they look different because they
> are different is like saying that things fall because they tend to
> fall. It's true but tautological.
>
>> In printing, dpi refers to the number of dots of ink per inch being
>> put down, and is in no way related to the number of pixels per inch.
>> When printing, a pixel may be represented by zero dots of ink (white)
>> to many different drops (say, for a dark green pixel). There is just
>> no way to correlate dpi with ppi.
>
> Right. I think I understand this, but thanks.

Hi...

Let me try another way, if I may, with the hope that it may make
it a little clearer...

First, please allow me to "make believe" that the printer prints r, g,
and b (instead of the correct c, m, y, k) as it will make example
easier.

Now let's take a special picture. Take your grand kid - a red headed
one. Put on a pink hat, a pink shirt with different shades of red
printing on it, pink shorts, pink socks, and bright red shoes.
All different tints of pink, of course. Now put her on a sandy beach,
wait till the sun is almost set (red again), and take the picture.

Looking at it on your (analogue) TV set, one triad (a red dot, a blue
dot, and a green dot) can light the red dot to virtually unlimited
number of intensity levels. Looking at it with our 8(24) bit monitors,
we can light it to any of 256 different intensities. So of all of our
reds and pinks in the foolish picture example, it should look quite
nice, we have 256 different shades. And if any other colors snuck in,
if only one (blue for instance) we can have 256 * 256 colors and shades
for a little more than 65000 different. I two snuck in we get
256 * 256 * 256 colors and intensities for over 16000000 total.

Now let's print it, keeping in mind our make believe rgb printing, and
right away we have a huge problem, that being that we have only one
red ink supply. No different intensities at all - either print red
or print nothing (white). So all that we can do is dither red and
nothing (white) to produce different intensities, which does work
quite well, IF we print at a very, very large number of dpi.

The only solution is a larger number of ink intensities. Imagine,
I'm out of ink, I need 256 different red ink cartridges, and 256 blue.
and 256 green (really cmy)

Hope this might make it a bit clearer.

Take care.

Ken
 >> Stay informed about: how come need so high dpi for printing photos? 

Ken Weitzel wrote:
> Hi...
>
> Let me try another way, if I may, with the hope that it may make
> it a little clearer.

Hello. Let's try.

> First, please allow me to "make believe" that the printer prints r, g,
> and b (instead of the correct c, m, y, k) as it will make example
> easier.
>
> Now let's take a special picture. Take your grand kid - a red headed
> one. Put on a pink hat, a pink shirt with different shades of red
> printing on it, pink shorts, pink socks, and bright red shoes.
> All different tints of pink, of course. Now put her on a sandy beach,
> wait till the sun is almost set (red again), and take the picture.
>
> Looking at it on your (analogue) TV set, one triad (a red dot, a blue
> dot, and a green dot) can light the red dot to virtually unlimited
> number of intensity levels. Looking at it with our 8(24) bit monitors,
> we can light it to any of 256 different intensities. So of all of our
> reds and pinks in the foolish picture example, it should look quite
> nice, we have 256 different shades. And if any other colors snuck in,
> if only one (blue for instance) we can have 256 * 256 colors and shades
> for a little more than 65000 different. I two snuck in we get
> 256 * 256 * 256 colors and intensities for over 16000000 total.
>
> Now let's print it, keeping in mind our make believe rgb printing, and
> right away we have a huge problem, that being that we have only one
> red ink supply. No different intensities at all - either print red
> or print nothing (white). So all that we can do is dither red and
> nothing (white) to produce different intensities, which does work
> quite well, IF we print at a very, very large number of dpi.
>
> The only solution is a larger number of ink intensities. Imagine,
> I'm out of ink, I need 256 different red ink cartridges, and 256 blue.
> and 256 green (really cmy)
>
> Hope this might make it a bit clearer.

Thanks for the reply! OK, this was clear from before. I do not disagree
with any facts that you mention. But let me try again: 1) The aspect of
print quality that you are discussing is improved by a lower ppi. But
we are trying to answer the question "why is 100dpi bad and 300dpi good
on a print?". 2) Prints made by wet processes (eg at minilabs, Fuji
Frontier etc) do not work by dithering, but try to print at 80ppi and
see the results.

Cheers,
Achilleas
 >> Stay informed about: how come need so high dpi for printing photos? 

You're right, it's probably a combination of reasons. I had assumed the
original question was why a given number of dpi printed looked lower
resolution than a similar number of ppi displayed. On re-reading I see it's
a bit more general than that.

There is clearly an intrinsic difference between a light generating medium
such as a monitor and a selectively reflecting medium such as a print but I
don't know enough about human visual perception to know the implications of
that.

Keith
 >> Stay informed about: how come need so high dpi for printing photos? 

acl wrote:
> we are trying to answer the question "why is 100dpi bad and 300dpi good
Argh! ppi, ppi, ppi.... Typo (again!)
 >> Stay informed about: how come need so high dpi for printing photos? 

Wayne wrote:
> In article <1168985570.928701.323250 RemoveThis @38g2000cwa.googlegroups.com>,
> achilleaslazarides RemoveThis @yahoo.co.uk says...
>
> >Well, the printer is supposed to be 5760dpi in the horizontal
> >direction, which I interpret to mean 2880x1440dpi in reality (with some
> >kind of intelligent dot placement, as I can clearly see it's not
> >5760dpi if I photograph it using a macro lens).
>
>
> No, that printer rating is about where the carriage motor can position an
> ink dot, the geometry of the location of the center of the drop. However,
> the ink drops themselves are very much larger diamter than this maximal
> spacing, so many fewer of the possible locations are actually used. The
> number of distinguishable ink drops per inch is much lower, and is not
> rated. And even then, that ink drop is only the color of one of three ink
> colors, NOT the color of the pixel.

Hi. The Epson R800 is labelled as "effective 5760x2880dpi" (or
something like that). I would bet any amount of money that 5760 isn't
the accuracy of the motor, but is achieved by a 2880 or maybe even
1440dpi accuracy and repeated passes for intelligent dot placement. I'm
sure there are resources on the internet stating it, but you can see it
on the print easily (under magnification), so didn't look for them.

>
> The monitor looks better at low resolution because it has much better color
> depth (256 intensities - basically like 16 million colors of ink
> cartridges). Such color changes contribute to spacial detail, it is detail
> if it can be shown. The monitor is also a much brigher "ink" that "glows",
> with much more dynamic range. The monitor image is viewed by transmitted
> light instead of reflected light. Very much like why a projected film
> slide is much better than a print.

Yes, this has been repeated a few times in this thread. Regarding your
first point (more colour depth per pixel area for small pixel sizes),
this would imply that lower ppi would look better. But to get a good
print, you need more ppi. So, while I do not disagree with the facts
that you state, I do not think it is the answer to the question (see my
other post).

>
> Continuous tone printers like the Fuji Frontier can do the multiple
> intensities too, but light from reflected paper is still poor dynamic range
> compared to transmitted light. Most of those possible colors cannot be
> seen.
>
>
> >I meant that it
> >actually resolves 480ppi (ie if I create 1 pixel wide lines spaced by 1
> >pixel and send them to be printed at 480dpi, it actually resolves them,
> >although just).
>
> You must be speaking of black ink lines. No color dithering is required to
> print the black ink color. Black is effectively printing in line art mode,
> one ink dot per pixel suffices. Try red or green or blue lines
> (compliments of CMY ink colors). Then depending on shade, it must dither
> more ink dots per pixel. This is a vastly harder problem, because it cannot
> put all those ink dots into the limited pixel space of higher resolution.

No, I am not speaking of black lines. Look here: I took a chart of
Roger Clark's site, here,
http://www.clarkvision.com/imagedetail/printer-ppi/index.html
and printed it. I printed it at 1440ppi, and photographed the print
with a Nikon D200 at a magnification of, I think, 1:1.2 or thereabouts.
Here's a full resolution jpeg of the photo:
http://www.pbase.com/al599/image/73194593
(see the full size). Judge for yourself what is resolved and what is
not, but whatever the resolution is for black, it is also for eg blue.
I would say that the lines separated by 3px are just separated, thus
480ppi (maybe one could say 4px, this is debatable).

Or maybe I am misunderstanding something here, in which case, please
correct me.

> So we see an approximation color, with an offsetting color correction in an
> adjacent pixel. We confuse this with high resolution.
 >> Stay informed about: how come need so high dpi for printing photos? 

"Ken Weitzel" <kweitzel RemoveThis @shaw.ca> wrote in message
news:4Ghrh.690947$R63.493515@pd7urf1no...

> The only solution is a larger number of ink intensities. Imagine,
> I'm out of ink, I need 256 different red ink cartridges, and 256 blue.
> and 256 green (really cmy)

Cheaper to buy a dye sub printer though.
 >> Stay informed about: how come need so high dpi for printing photos? 

In article <1168985570.928701.323250.TakeThisOut@38g2000cwa.googlegroups.com>,
achilleaslazarides.TakeThisOut@yahoo.co.uk says...

>Well, the printer is supposed to be 5760dpi in the horizontal
>direction, which I interpret to mean 2880x1440dpi in reality (with some
>kind of intelligent dot placement, as I can clearly see it's not
>5760dpi if I photograph it using a macro lens).


No, that printer rating is about where the carriage motor can position an
ink dot, the geometry of the location of the center of the drop. However,
the ink drops themselves are very much larger diamter than this maximal
spacing, so many fewer of the possible locations are actually used. The
number of distinguishable ink drops per inch is much lower, and is not
rated. And even then, that ink drop is only the color of one of three ink
colors, NOT the color of the pixel.

The monitor looks better at low resolution because it has much better color
depth (256 intensities - basically like 16 million colors of ink
cartridges). Such color changes contribute to spacial detail, it is detail
if it can be shown. The monitor is also a much brigher "ink" that "glows",
with much more dynamic range. The monitor image is viewed by transmitted
light instead of reflected light. Very much like why a projected film
slide is much better than a print.

Continuous tone printers like the Fuji Frontier can do the multiple
intensities too, but light from reflected paper is still poor dynamic range
compared to transmitted light. Most of those possible colors cannot be
seen.


>I meant that it
>actually resolves 480ppi (ie if I create 1 pixel wide lines spaced by 1
>pixel and send them to be printed at 480dpi, it actually resolves them,
>although just).

You must be speaking of black ink lines. No color dithering is required to
print the black ink color. Black is effectively printing in line art mode,
one ink dot per pixel suffices. Try red or green or blue lines
(compliments of CMY ink colors). Then depending on shade, it must dither
more ink dots per pixel. This is a vastly harder problem, because it cannot
put all those ink dots into the limited pixel space of higher resolution.
So we see an approximation color, with an offsetting color correction in an
adjacent pixel. We confuse this with high resolution.

--
Wayne
http://www.scantips.com "A few scanning tips"
 >> Stay informed about: how come need so high dpi for printing photos? 

"dennis@home" <dennis.TakeThisOut@killspam.kicks-ass.net> writes:

>"Dave Martindale" <davem.TakeThisOut@cs.ubc.ca> wrote in message
>news:eojk4p$d10$1@swain.cs.ubc.ca...

>> It's also not as sharp as a good print, but here the comparision is
>> confused by contrast. A decent monitor can achieve a white-to-black
>> contrast of several hundred to one or better, while a colour print can
>> manage maybe 30:1 and a good silver B&W print can do 100:1.

>Glossy paper reflects about 95% of light.
>The black level reflects a few percent too.

>Even a good print is probably 20:1 contrast ratio at best.
>To get 100:1 it would have to absorb all the light in the black areas and
>reflect >99% in the white which they just don't do.

There's some bad math there. You could get 100:1 contrast from paper if
the white reflected only 50% of incoming light - provided that the black
reflected only 0.5%. If the black areas absorbed *all* the light, the
contrast ratio would be infinite. What is really required is for the
white areas to reflect 100 times as much light as the black areas,
whatever the absolute reflectances turn out to be.

Good silver B&W printing paper can have an optical density above 2.0 in
the dark areas. By definition, that means it is reflecting less than 1%
of the incoming light. A realistic reflectance of the white area is
perhaps 90%, so more than 100:1 is possible.

Now, the gelatin surface reflects a few percent of the incoming light.
If the surface is glossy, parallel incoming light is reflected out still
parallel. If you view a glossy print under point light sources (e.g.
the sun, most incandescent lamps), you will get very bright reflections
off the paper from those lights, and you will automatically tilt the
print until the light is reflected away from your eyes - so it does not
contaminate the darkness of the blacks.

(Pearl or matte finish paper scatters light from its surface in all
directions, so the angle you hold the print isn't critical, but the
blacks are contaminated by scattered light no matter how you hold it.
That's why glossy prints have better contrast.)

Dave
 >> Stay informed about: how come need so high dpi for printing photos? 

Dave Martindale wrote:

>
>> Glossy paper reflects about 95% of light.
>> The black level reflects a few percent too.
>
>> Even a good print is probably 20:1 contrast ratio at best.
>> To get 100:1 it would have to absorb all the light in the black areas and
>> reflect >99% in the white which they just don't do.
>
> Now, the gelatin surface reflects a few percent of the incoming light.
> If the surface is glossy, parallel incoming light is reflected out still
> parallel. If you view a glossy print under point light sources (e.g.
> the sun, most incandescent lamps), you will get very bright reflections
> off the paper from those lights, and you will automatically tilt the
> print until the light is reflected away from your eyes - so it does not
> contaminate the darkness of the blacks.
>
> (Pearl or matte finish paper scatters light from its surface in all
> directions, so the angle you hold the print isn't critical, but the
> blacks are contaminated by scattered light no matter how you hold it.
> That's why glossy prints have better contrast.)


You can get about 300:1 contrast ratio on glossy color paper
if exposed, developed, mounted and lighted correctly. This is
an actual measured number with a good spot light meter. It's
hard to keep it due to dust on the print of the cover glass.

*THE* biggest help to get a nice snappy contrasty looking
print is adequate lighting of it. Direct sunlight is good
in this regard (and very bad for the dyes, of course).

Doug McDonald
 >> Stay informed about: how come need so high dpi for printing photos?