Printer resolution

  • Thread starter Thread starter Pper
  • Start date Start date
P

Pper

In the past, I have printed color photos at 300dpi. This has worked well.
I am going to upgrade my old HP color printer to something newer. Looking
at the market, the newer printers print at higher dpi. Most state 4800dpi.
When I print at 300dpi, I get a really nice looking picture. What's with
the 4800dpi printers? These should look better?

Assuming I have a jpg that has 2400 (or higher) dpi, will I get a
significantly better picture printed from the newer printers?
 
Looking at the market, the newer printers print at higher dpi. Most
state 4800dpi. When I print at 300dpi, I get a really nice looking
picture. What's with the 4800dpi printers? These should look better?
Click to expand...

Personally, I don't think there's any point in going above 1440 dpi and
720 dpi is often adequate.

Jon.
 
Pper said:
In the past, I have printed color photos at 300dpi. This has worked well.
I am going to upgrade my old HP color printer to something newer. Looking
at the market, the newer printers print at higher dpi. Most state 4800dpi.
When I print at 300dpi, I get a really nice looking picture. What's with
the 4800dpi printers? These should look better?
Click to expand...

I was slightly surprised to find that I _did_ see an improvement when I
printed the same image at 300dpi on a 4800 dpi capable printer when compared
to a 300dpi capable printer. After some thought I concludede that this was
for two reasons...

a) To achieve 4800dpi the printer must be capable of producing very small
droplets of ink. It still uses these small droplets (but more of them) when
you print at 300dpi. Several small droplets probably don't spread/smear/blur
as much as the larger ones used in a 300dpi printer.

b) 4800 dpi printers are more modern - so the firmware inside is better.
 
Pper said:
In the past, I have printed color photos at 300dpi. This has worked well.
I am going to upgrade my old HP color printer to something newer. Looking
at the market, the newer printers print at higher dpi. Most state 4800dpi.
When I print at 300dpi, I get a really nice looking picture. What's with
the 4800dpi printers? These should look better?

Assuming I have a jpg that has 2400 (or higher) dpi, will I get a
significantly better picture printed from the newer printers?
Click to expand...
You are confusing dpi (dots per inch) with ppi (pixels per inch). Dots and
pixels are not the same thing at all. Keep sending 300 ppi files to your
printer. I can seel little improvement with higher than 1440 dpi prints.
Jim
 
You are confusing dpi (dots per inch) with ppi (pixels per inch). Dots and
pixels are not the same thing at all. Keep sending 300 ppi files to your
printer. I can seel little improvement with higher than 1440 dpi prints.
Jim
Click to expand...
Actually, if you have a 1440 dpi printer, the likely native resolution
is 720 dpi, so you're better off with a 360 ppi image. (i.e a fraction
of the native resolution).
 
Pper said:
In the past, I have printed color photos at 300dpi. This has worked well.
I am going to upgrade my old HP color printer to something newer. Looking
at the market, the newer printers print at higher dpi. Most state 4800dpi.
When I print at 300dpi, I get a really nice looking picture. What's with
the 4800dpi printers? These should look better?

Assuming I have a jpg that has 2400 (or higher) dpi, will I get a
significantly better picture printed from the newer printers?
Click to expand...

Resolution on modern printers is not important in any more. The most
important thing is the number of inks (6 or 8 in the latest HP ones - I
don't know about the others) and the number of pixels per inch.

The reason that more inks are important is that this means more
effective colours. The reason that pixels per inch is more important
means that you can get more of these millions of colours into a pixel by
overlaying the inks.

These days, with the right paper, there's very little difference between
the top three vendors, HP, Epson and Canon. HP trys to match colours as
best as possible, Epson seem to go for oversaturated eye-catching
colours to catch the uninformed home user. Canon too.

In my opinion, the new HP 8 ink printers (using the Vivera inks) with
the HP Premium Plus photo paper which will gives over 100 years
light-fastness are far and away the most complete closed-loop system. YMMV.
 
This is a continually confusing area for most people, and the
manufacturers, rather than making it easier to understand, seem to add
to the confusion because people are always impressed with big numbers.

Before going any further, let me just say, that if at all possible,
allow your eyes to make the decision. Try to look at samples of the
printer output. Most companies provide free samples if requested. You
will note that most companies tend to NOT print the image at the highest
resolution the printer is capable of. The reason is the printer speed
is quite reduced when the printer is used in it's very high resolution
modes, and the quality is only slightly improved. In general, any
printing mode over 1200 dpi is barely visible on the output.

OK, now to try to help regarding input and output resolutions.

The input resolution, is the resolution of the image you are getting
ready to print, which is identified as XXX ppi or dpi at the final
printing size. So, if one is to use 300 dpi (which is most adequate for
even smaller prints which you will view quite closely), a 8" x 10" image
should have the total pixel dimensions of 300 x 8" or 2400 pixels by 300
x 10" or 3000 pixels, or 8" x 10" x 300 ppi/dpi.

Input resolutions of between about 250 and 350 ppi/dpi provide
photographic quality for even smaller closely viewed prints. The
average one hour lab image is about 200-250 dpi equivalent. A custom
color print is typically between 300-350 dpi. A very high quality large
format print may be up to 400 dpi.

ALso, keep in mind that the larger the print is, and therefore the
further away you will view it, the lower the input resolution you can use.

Now, as to output resolution. This is the resolution the printer
uses to create the image onto the paper.

Each printing technology requires differing resolutions depending upon
the mechanics of how the image is created. A dye sublimation printer,
which usually can produce millions of colors per "dot" can represent an
image pixel with just one dot. After all, any pixel in an image is only
of one luminosity and one color. Dye sub printers, use three of four
"layers" of transparent dye "vapor" that are transferred onto a special
receptor paper. Most have at least 128 different density levels for
each of dye colors, making literally millions of colors for any one dot.

Inkjet printers don't work the same way. They typically have only one,
or maybe two levels of any ink color, so even if they were completely
transparent colors and overlaid perfectly, they may be able to create
anything from a few dozen to maybe a hundred color combinations with all
the inks. That's not enough to make a photographic-like image.

So the way inkjet printers create colors is by a process of placing a
lot of very small dots near one another. That, mixed with the
background paper white, provide an illusion of hundreds of thousands or
even millions of colors. In order to do this, the addressable position
density for any one dot has to be much finer, and so, these printers
need over 1000 dpi and more importantly very very small dots, to get
photo-like results.

When a printer claims 4800 dpi, or even high as some do, that's the
potential number of addressable places a dot can be positioned on the
paper, but if the printer actually printed 4800 or more dots right next
to one another, they would overlap each other, because the dots are
still larger than 1/4800th of an inch.

In general, 1200-1500 dpi is just about the maximum that your eye will
call photographic, and going higher will mainly just slow the process
down, with little improvement, it also requires more memory for the
spooler process.

If the dots finally get small enough, there may be some advantage to
slightly higher resolutions, but the question comes down to if you will
be viewing your images with an unaided eye or with a photo loupe.

Art
 
OK, let's define some terms...

What color inkjet printer prints output at 300 dpi? I don't think any
have done so for about 8 or more years.

Yes, the 4800 dpi printer probably has a fairly small dot, although it
is unlikely to be 1/4800th of an inch.

Much older inkjet printers started out with about 20 picolitre dots.
Newest printers can use as small as 2 picolitre dots (many printers use
variable dot size, which can be changed on the fly for upward of 8
sizes). A picolitre is a volume of ink, not a size, since the size can
be different depending on how it lands on the page, and how much dot
gain occurs based upon the paper type.

Art
 
Hecate said:
Actually, if you have a 1440 dpi printer, the likely native resolution
is 720 dpi, so you're better off with a 360 ppi image. (i.e a fraction
of the native resolution).

--
Click to expand...

There has been a lot of empirical testing done regarding this issue, and
the consensus, at least with Epson printers, is that with the early 720
x 720 dpi printers, using 240 or 360 dpi tended to reduce some printing
artifacts. Since the newer drivers in the 1440 and higher printers, it
seems to make no difference if you use "sweet" numbers or not. The
drivers for these printers use 720 x 720 dpi spooling, while the wide
carriage printers now use 360 x 360 dpi. Possibly, if you used 720 x
720 ppi input on a consumer model, you might reduce some artifacting,
but the drivers today are so good that it makes little, if any,
difference.

In fact, people found that manipulating the input ppi/dpi to try to up
or downsample to get to the "sweet" numbers did more damage to the file
than just leaving the image at whatever resolution it ended being at the
output size, unless it was great overkill (like 800 ppi/dpi or
something). I know of some cases where tests were done with even
leaving the ppi/dpi at odd numbers like 334.7, as an example, and that
produced a very slightly better result than upsampling to 360, for
instance. In conclusion, with modern drivers, it probably is not
worthwhile trying to make the image meet sweet numbers.

Art
 
There has been a lot of empirical testing done regarding this issue, and
the consensus, at least with Epson printers, is that with the early 720
x 720 dpi printers, using 240 or 360 dpi tended to reduce some printing
artifacts. Since the newer drivers in the 1440 and higher printers, it
seems to make no difference if you use "sweet" numbers or not. The
drivers for these printers use 720 x 720 dpi spooling, while the wide
carriage printers now use 360 x 360 dpi. Possibly, if you used 720 x
720 ppi input on a consumer model, you might reduce some artifacting,
but the drivers today are so good that it makes little, if any,
difference.

In fact, people found that manipulating the input ppi/dpi to try to up
or downsample to get to the "sweet" numbers did more damage to the file
than just leaving the image at whatever resolution it ended being at the
output size, unless it was great overkill (like 800 ppi/dpi or
something). I know of some cases where tests were done with even
leaving the ppi/dpi at odd numbers like 334.7, as an example, and that
produced a very slightly better result than upsampling to 360, for
instance. In conclusion, with modern drivers, it probably is not
worthwhile trying to make the image meet sweet numbers.
Click to expand...
I would never suggest anyone should upsample anyway. Most of the time
I am downsampling to those resolutions from 120Mb files.

Other than that, I would have to disagree in general. With the Epson
printers I have seen to suggest that aiming for a multiple of the
native resolution doesn't lead to a better print. If you have that
information, I'd appreciate it if you shared it.
 
Hecate wrote:

Other than that, I would have to disagree in general. With the Epson
printers I have seen to suggest that aiming for a multiple of the
native resolution doesn't lead to a better print. If you have that
information, I'd appreciate it if you shared it.
Click to expand...

I'm sorry, but I am unclear what your last paragraph (above) is stating.
This is not a flame of any kind. I'm suspecting it's just the
sentence construction which is confusing me.

Are you stating that you do find that using even multiples of native
resolution provide you with better results with most current Epson
printers/drivers?

My information was from a series of tests that a number of people did on
one of the Epson printer groups a few years back. These were scanned at
high dpi (I don't recall the resolution anymore) and enlarged and
scrutinized by a number of people including myself. I no longer have
the files on my system. The consensus was as I stated.

It is certainly possible your results could be different. Subject
matter, saturation, and contrast can all alter the nature of the
results. If you are using a digital camera image as your source, that
might influence the result (we were working with analogue grain film
images). Heck, even changes in the manner Epson used in their printer
drivers may be unique to a certain model. As I stated in my earlier
posting, Epson printer drivers have certainly evolved.

At the time, the group involved in the comparisons was pretty much in
agreement that the natural multiples were no longer relevant, as long as
adequate input resolution was provided.

Perhaps a more important question is, with the nature and complexity of
error diffusion printing is if any methods can completely eliminate
printer induced artifacting of some sort, and also, whether anyone
viewing such an image at anything approaching normal viewing distances
could really tell.

I certainly am not trying to suggest what it is you are or aren't
seeing, as perception is quite subjective at a certain level.

In the end, what matters is how the print is perceived, by both the
originator and others. After a certain point, for instance, unsharp
masking creates no new real information, but many people still prefer
higher frequency sharpening even when it introduces error artifacts.
Who am I to suggest they are "creating" information that wasn't in the
original image or that it matters?

If you have found resampling (down or up) to a native multiple of the
driver's 720 works in providing a more pleasing, or more accurate print
result, then that is probably worthwhile pursuing. Others have not
encountered the same results, but quite honestly, in our comparisons, we
found we were splitting hairs at high magnification and there might have
been other confounding issues, like the scanners.

Art
 
Thanks for the explanation. Not too long ago, 600dpi printers were
out-performing 2400 (and higher) Epson printers. Epson was touting XXXX
dpi. "It must be better! ..." HP was touting 600 dpi with drop overlaps.
Putting pictures side-by-side and visually examining them (without a loupe)
showed they were nearly identical. But, HP was loosing sales to the
"higher" dpi crowd. Now HP has joined the "others" with the 2400/3600/4800
dpi to sell to the person who looks just at the dpi ratings.

Examining printouts from Epson, HP, and Canon in the $150-$200 range all
appear to be similar. Printers with 6 ink colors appear to be better than
those with 3.
 
Hecate wrote:



I'm sorry, but I am unclear what your last paragraph (above) is stating.
This is not a flame of any kind. I'm suspecting it's just the
sentence construction which is confusing me.
Click to expand...

Oh, I realise you weren't flaming ;-)

I was trying to say that, generally, I have found that matching the
native resolution of the printer provides better prints than not doing
so. And I was asking that, if you knew different (i.e if you had
information that could show I was being subjective) I'd be glad to
look at it. :)
 
Arthur Entlich said:
Hecate wrote:



I'm sorry, but I am unclear what your last paragraph (above) is
stating. This is not a flame of any kind. I'm suspecting it's just
the sentence construction which is confusing me.

Are you stating that you do find that using even multiples of native
resolution provide you with better results with most current Epson
printers/drivers?

My information was from a series of tests that a number of people did
on one of the Epson printer groups a few years back. These were
scanned at high dpi (I don't recall the resolution anymore) and
enlarged and
scrutinized by a number of people including myself. I no longer have
the files on my system. The consensus was as I stated.
Click to expand...
Art, you may be referring to some of the tests that I was involved in
and the conclusion then certainly was that the opposite of what you are
suggesting. Avoiding resampling by the printer driver is ALWAYS better,
although the difference may often be imperceptible.

Epson printer drivers only apply nearest neighbour resampling if the DCC
option (if present) is unselected and bilinear resampling if this option
is selected. Particularly when downsampling, but also visible on high
contrast near vertical/horizontal edges with upsampling, this is visibly
inferior to the alternative resampling options available in even modest
image editing applications.
 
Hi Kennedy,

The tests I was referring to I do not believe involved you at the time,
to the best of my knowledge. As I recall, beer bottles and their labels
photographed at a fair distance were involved.

You may have missed my first response in this thread. The question that
came up was if using 360 or some other exact multiple of the native
printer driver resolution was beneficial over other numbers.

I stated that the conclusions of the tests I looked at were that doing
no resampling prior to printing, and just allowing the resolution to end
up where it may (and therefore allowing the printer driver to do its
upsampling to 720 dpi as most of the consumer Epsons do) appeared to
leave fine detail less altered than sampling to something like 360 or
240 dpi or whatever, and then leaving the printer to resample again to
720 dpi. I also stated that earlier Epson's seems to benefit more from
these multiples. I don't know if older Epson desktop printers (like 1st
and 2nd generation) used 360 or 720 dpi driver resolutions).

Perhaps down or up sampling to 720 dpi prior to printing might have an
advantage, for those who work with files that large. Most people
neither have that type of resolution from a scan or digicam (other than
for quite small prints) nor do they wish to work with such large files.

Art
 
Arthur Entlich said:
Hi Kennedy,

The tests I was referring to I do not believe involved you at the time,
to the best of my knowledge. As I recall, beer bottles and their labels
photographed at a fair distance were involved.

You may have missed my first response in this thread. The question
that came up was if using 360 or some other exact multiple of the
native printer driver resolution was beneficial over other numbers.

I stated that the conclusions of the tests I looked at were that doing
no resampling prior to printing, and just allowing the resolution to
end up where it may (and therefore allowing the printer driver to do
its upsampling to 720 dpi as most of the consumer Epsons do) appeared
to leave fine detail less altered than sampling to something like 360
or 240 dpi or whatever, and then leaving the printer to resample again
to 720 dpi. I also stated that earlier Epson's seems to benefit more
from these multiples. I don't know if older Epson desktop printers
(like 1st and 2nd generation) used 360 or 720 dpi driver resolutions).
Click to expand...
I may indeed have missed your original response, the earliest I have
seen was posted on Mon, 11 Oct 2004 20:08:18 MDT, and this is where you
refer to the "empirical tests". You may also recall some empirical
testing that was also conducted which proved conclusively that the
recent (eg. printers with >=1440dpi resolution) Epson desktop printer
drivers not only resampled to 720ppi (360ppi on large format printers)
but, unless set to cope with low resolution material, used crude nearest
neighbour resampling to do so. These proofs were achieved by printing
specific images at and near "sweet spot" resolutions and observing the
resulting aliased effects.

If the "sweet spot" did not offer any visual advantage (ie. no aliasing
introduced) then these tests would not have been conclusive.
Consequently, it is clear that for some images the difference is
dramatic - where significant contrast is present between adjacent
pixels. Where this is not present the difference is less pronounced,
and this certainly can be the case with images scanned from film at
sufficient resolution, however the benefit of resampling to the "sweet
spot" at an integer division of 720ppi is never zero.
Perhaps down or up sampling to 720 dpi prior to printing might have an
advantage, for those who work with files that large. Most people
neither have that type of resolution from a scan or digicam (other than
for quite small prints) nor do they wish to work with such large files.
Click to expand...
That is a different question - one of convenience and effort. I
wouldn't care to make any recommendation on such a matter since I rarely
understand why many people cannot be bothered to make the effort to do
anything right in many fields.
 
That is a different question - one of convenience and effort. I
wouldn't care to make any recommendation on such a matter since I rarely
understand why many people cannot be bothered to make the effort to do
anything right in many fields.
Click to expand...


LOL! I think I'll print that out and frame it! ;-)
 
If only the issue were so clear cut.

The effort that goes into a process may relate to the intent of the
outcome, the time involved, the cost, and so on.

Assuming anyone knows what is acceptable or even desirable for someone
else is a bit of a gamble, at best.

Some people search for the holy grail (and some may even find it) but at
a cost of time, effort, expense, and perhaps even loss of creative
output to get there.

In some cases, it can even be pathological (Obsessive compulsive disorder).

To take some liberty with the concept behind "Zen and the Art of
Motorcycle Maintenance" one person may use a bottle cap to fix his
motorcycle and get it back on the road to continue his journey while
another might find himself without transportation for weeks waiting for
the exact part to come in from the manufacturer.

Scrutinizing an inkjet print output with a loupe may indeed prove
interesting, and even enlightening, but the differences between prints
may be completely insignificant to anyone seeing them at normal viewing
distance, in which case all the extra effort and expense involved in
getting the print beyond the point that the unaided eye can ascertain a
difference isn't much more than personal satisfaction for the technician.

Determining that Van Gogh used fast, broad brush and pallet knife
strokes, often without cleaning them between each use seems to have had
no negative consequence upon the value of his paintings today. In fact,
it may indeed be part of the charm that sets them apart.

Assuming one needs at least 720 dpi of real data to produce an image on
an Epson printer to make the image meaningful, with it's commensurate
extra cost for a high res scanner or digicam, a computer with more
memory and faster processor, and a larger hard drive to store it all in
order to make the image worthwhile, or before it reaches some level of
required sophistication no longer has anything to do with producing art,
and is instead about producing "image by specification". Sorry, but I
neither buy art on that basis, nor do I produce it that way.

Good technique is valuable, but secondary to the image content, and
microscopic "improvements" are nearly meaningless. Certainly, if basic
good practice can lead to better quality results without sacrificing
other matters, than why not, but splitting hairs over issues that
actually are not visible to the non-assisted observer, to me at least,
are of minimal, if any value.

I recall once hearing Zamfir (the pan flute player - anyone remember
him) commenting on a talk show that the music of the Beatles was with
little merit, because it was undisciplined and didn't conform to the
qualities of good music, or good musicianship.

Art
 
Arthur Entlich said:
Scrutinizing an inkjet print output with a loupe may indeed prove
interesting, and even enlightening, but the differences between prints
may be completely insignificant to anyone seeing them at normal viewing
distance
Click to expand...

The point is that you don't need a loupe to see aliased image content.
Spatial aliasing is its own magnifier. If you don't take steps to
control aliasing - and the Epson print drivers themselves don't - then
the detail that you would need a loupe to see can often become so
distorted and enlarged that it is readily seen at long viewing ranges.
 
Back
Top