scan dpi for photos
T. Parker
What's the minimum scanning dpi (dot per inch) for photos
or x-ray films? I saw a picture with a 72 dpi or 100 dpi and
the photo seems to be ok. I thought normal scanners
have 300 dpi as minimum. How many percentage image
quality loss would a 72 dpi picture produce? Any url for
comparisons between 10,000 dpi vs 72 dpi?
Parker
D.Mac
"T. Parker" <tompa...@gmail.com> wrote in message
news:c6250d31-c0b6-45cc...@y34g2000prb.googlegroups.com...
The scanned DPI is relative to the use you have for it. 72 DPI is for screen
resolution. 140 DPI is suitable for a Fax and anything from 180 DPI to 360
DPI is suitable for inkjet printing.
If you are scanning a document for a magazine page (Offset printing) it
could be as low as 100 DPI and still look OK when printed.
The highest resolution scan you want for a negative is 3200 DPI which is
about the actual resolution of most films. You can then enlarge the scanned
image to about 11"x14" for printing. As you increase the size of the
picture, the DPI you scanned it at will get lower.
T. Parker
> "T. Parker" <tomparke...@gmail.com> wrote in message
>
> news:c6250d31-c0b6-45cc...@y34g2000prb.googlegroups.com...
>
> > Hi,
>
> > What's the minimum scanning dpi (dot per inch) for photos
> > or x-ray films? I saw a picture with a 72 dpi or 100 dpi and
> > the photo seems to be ok. I thought normal scanners
> > have 300 dpi as minimum. How many percentage image
> > quality loss would a 72 dpi picture produce? Any url for
> > comparisons between 10,000 dpi vs 72 dpi?
>
> > Parker
>
> The scanned DPI is relative to the use you have for it. 72 DPI is for screen
> resolution. 140 DPI is suitable for a Fax and anything from 180 DPI to 360
> DPI is suitable for inkjet printing.
Why. What resolution does 72 dot per inch correspond to? To scan
a picture. Does it have any difference to scan it at 72 dpi versus
to scan it at 600 dpi and then downscale to 72 dpi?
Parker
Neil Ellwood
> The scanned DPI is relative to the use you have for it. 72 DPI is for
> screen resolution. 140 DPI is suitable for a Fax and anything from 180
> DPI to 360 DPI is suitable for inkjet printing.
>
> If you are scanning a document for a magazine page (Offset printing) it
> could be as low as 100 DPI and still look OK when printed.
>
> The highest resolution scan you want for a negative is 3200 DPI which is
> about the actual resolution of most films. You can then enlarge the
> scanned image to about 11"x14" for printing. As you increase the size of
> the picture, the DPI you scanned it at will get lower.
The DPI you scanned the image at will stay the same no matter what size
you print it.
--
Neil
reverse ra and delete l
Linux user 335851
T. Parker
wrote:
For book publishers who want to include x-rays pictures
in their book. How many DPI of scanners do they use?
In 1980. What DPI are available for book publishers?
Anyone?
Parker
dj_nme
> understanding the basic concepts involved regarding digital images. A
> good place to start might be Wayne Fulton's on-line guide at:
>
> http://www.scantips.com
The DPI at which you scanned a picture at is not changed by what you do
to it afterwards: the past cannot be changed, as it's already gone.
You can how-ever resample the image to a different resolution, but
information cannot be created (ie: resampling to a higher resolution
does not add any extra detail).
T. Parker
After more than an hour scanning different materials at
different DPIs. This is what I found out. Scanning at DPI
of 72 will always produce very quality. It's better
to scan at 300 DPI then change the image size to
smaller (matching the 72 dpi screen size). That way,
you would retain the quality of the 300 dpi yet the
image smaller to fit the screen (72 dpi). The file
size of pure 72 dpi scan is equal to 300 dpi with
image size down to same resolution as 72 dpi.
Yet the latter is much much clearer. So 72 dpi
scanning is pretty much useless unless you just
want a 5mm (1/5") ID picture.
Don Stauffer
It depends so much on the size of the print (I am assuming you are
talking about prints here, not scanning negatives. Small prints need to
be scanned at a MUCH higher pitch than you mentioned. The values you
gave would be for almost poster-sized prints.
You need at least about 1500 to 2500 pixels in the horizontal direction
to give much detail. Say you have a six inch print. Then 2000/6 is
about 350 ppi.
HEMI-Powered
...
Parker, to intelligently answer your question, you must say what
your intended purpose of the scans will be. But, here's a basic
guideline I use:
I attempt to scan for the ultimate image size I am interested in
BUT clearly NOT more dpi than the original prints, slides, or negs
actually contain. Usually, I overscan by about 3X the dpi I really
need than reduce the image before further processing. This tends to
smooth out the many defects in my scan material including noise,
film grain, and the like.
So, suppose I have an old snapshop that maybe is 3" x 5", suppose
it is reasonable in quality, and suppose I want to end up with a
1280 x 960 image, about all that size print can support. The simple
math says 1280 / 5 = 256 dpi, but I would like scan at 600 dpi.
One tip/warning, NEVER use a higher dpi than your scanner supports
optically. All scanners have higher dpi's than they really can
resolv but essentially, they are resampling upward to fit the
higher dpi, and generally result in similar problems to your
earlier questin about trying to make a silk purse out of a sow's
ear.
10,000 dpi is simply absurd! Even for 35mm negs or slides, and
assuming a high quality scanner and not a $50 Best Buy machine, few
older films can utilize more than about 2500-3000 dpi, but some may
be good enough for 3500-4000 dpi.
I don't want to insult you more than I already have, but these
questions smack of a newbie who has drunk the poison Kool-Aid of
the digital camera industry which has brainwashed consumers into
thinking that automagically have more mega pixels means better
images.
My final advice to you is to think about the purpose(s) you intend
to use these images for, such as on-screen slide shows, E-mail,
web, Facebook, Usenet and the like, quality prints in thee 8.5" x
11" size or larger, etc. Once you have a clear idea what you're
trying to accomplish, the math becomes quite simple.
Again, best of luck.
--
Jerry, aka HP
"If you are out of work and hungry, eat an environmentalist" -
Florida billboard
HEMI-Powered
...
>> > What's the minimum scanning dpi (dot per inch) for photos
>> > or x-ray films? I saw a picture with a 72 dpi or 100 dpi and
>> > the photo seems to be ok. I thought normal scanners
>> > have 300 dpi as minimum. How many percentage image
>> > quality loss would a 72 dpi picture produce? Any url for
>> > comparisons between 10,000 dpi vs 72 dpi?
>>
>> > Parker
>>
>> The scanned DPI is relative to the use you have for it. 72 DPI
>> is for scr een resolution. 140 DPI is suitable for a Fax and
>> anything from 180 DPI to 36 0 DPI is suitable for inkjet
>> printing.
>
> Why. What resolution does 72 dot per inch correspond to? To scan
> a picture. Does it have any difference to scan it at 72 dpi
> versus to scan it at 600 dpi and then downscale to 72 dpi?
Resolution as measured by dpi has NO, repeat, ZERO meaning
whatsoever except in two cases: the resolution of your PC monitor
to display the images and the largest size you want to print. In
the case of printing, the ONLY correct measure is PPI for a "good"
print which most people believe is in the 200-300 PPI range
although good results can be obtained by less in many instances.
See my reply to your OP for more on this, but you seriously need to
consider what you want or need to do with the end images and NOT
get tangled up in some useless game of statistics.
As a fast reply to the person you're really asking the 2nd question
of, I have a Canon Pixma 6600 photo printer. It is an outstanding
quality machine and Canon has improved on it considerably since I
bought mine. When I bought it several years ago, I benchmarked it
against similar priced printers - about $200 at the time - from
both Epson and HP and the Canon was clearly superior. But, my
intent is not to plug Canon, it is to qualify my next comment:
My Pixma 6600 can easily use a 300 PPI image to print to 8.5" x
11" which would equate to a pixel resolution of about 3300 x 2550,
but I have achieved good enough results for my purposes at 200 PPI
and even as low as 125-150 PPI. The main determining factors, for
simplicity, are the subject type(s), the degree of detail and
sharpness of the digital image, any image defects such as JPEG
artifacts, posterization, or aliasing (jaggies), etc. Generally,
the more fine detail in the image the higher PPI you will need for
"good" prints. My love is car pictures, and although I don't print
that many of them (I mostly do slide shows on-screen), I get what I
consider to be good resuls with images as small as 1280 x 960 but
1600 x 1200 is better and a 4-6 MP image would be better yet.
HEMI-Powered
jour ...
> On Sat, 04 Apr 2009 05:01:57 +0100, D.Mac wrote:
> The DPI you scanned the image at will stay the same no matter
> what size you print it.
>
Technically true, but dpi gets turned into pixels when the scan
outputs a digital image. When printing, the original dpi used for
scanning has zero meaning, the relevant number being the ppi which is
based on the size of the image in pixels and the intended print size.
Ofnuts
Two things to consider: at normal reading distance a good print quality
requires around 300dpi.
For large images which are looked at from a distance, 3000 pixels in the
larger dimension are usually enough.
So if you have "small" images (up to "Letter" size) you scan them a
300dpi (or, if you intend to enlarge them in the process, at
proportionally more DPI, say, 600dpi for a 2x size, up to 3000 pixels
on the largest side).
These figures are for direct scan-to-print. If you intend to do some
processing in the middle (especially de-noising), you can double the DPI
so that the processing software works with more information, and then
rescale the image afterwards.
--
Bertrand
HEMI-Powered
...
> For book publishers who want to include x-rays pictures
> in their book. How many DPI of scanners do they use?
> In 1980. What DPI are available for book publishers?
> Anyone?
>
Beats me what book publishers use, but I would expect the original
image size to be rather large. However, I have looked at digital X-
rays that my local hospital produces instead of the old-fashioned X-
ray plates and these images are in the 2 mega pixel range.
Again, it matters NOT what magazines, calendars, or books use for the
mega pixels of digital images they print or scans they print from. It
ONLY matters what the publisher wants to achieve and how much money
they will spend to get it. Among the many factors that affect printed
photos in books is the type of half-tone printing process used, the
book's paper type, and many other things. It just isn't possible to
generalize this, nor is it at all useful to you to try.
Again, please consider taking some basic and intermediate digital
imaging classes from your local camera store or public library system
and other educational sources. Before you can jump into all of this
as it appears you are trying to do, you need a much broader technical
grounding than a few questions in a Usenet group that are all
oriented toward "how to I create 10 mega pixels image?"
HEMI-Powered
...
> After more than an hour scanning different materials at
> different DPIs. This is what I found out. Scanning at DPI
> of 72 will always produce very quality. It's better
> to scan at 300 DPI then change the image size to
> smaller (matching the 72 dpi screen size). That way,
> you would retain the quality of the 300 dpi yet the
> image smaller to fit the screen (72 dpi). The file
> size of pure 72 dpi scan is equal to 300 dpi with
> image size down to same resolution as 72 dpi.
> Yet the latter is much much clearer. So 72 dpi
> scanning is pretty much useless unless you just
> want a 5mm (1/5") ID picture.
>
See my comment to you OP here, Parker. There is very simple math
involved here, it isn't a trial an error thing at all. The intended
use the scanned image is VERY important and the quality of the print
you are scanning is even MORE important. One simply cannot creat a
quality image from a substandard source, no matter the dpi you try.
You haven't mentioned noise removal yet, but if you've tried scanning
books, you undoubtedly are now familiar with the fact that the half-
tone dots used by printing houses results in a geometrically regular
pattern in the digital image that looks like noise. Scanners have
noise reduction schemes which vary in their capability, but the BEST
thing you can do is to over scan by attemping to use about 3X the dpi
you think you need and resizeing downward in your favorite graphics
app. What noise remains can be removed at that time.
Please flush your mind of sweeping generalizations like you made
above wrt what dpi is or is not useless and THINK about what you WANT
to do or what you NEED to do with the final image, consider the
quality of the source material you scanning from, and let your 4-
function calculator figure the dpi for you.
HEMI-Powered
> T. Parker wrote:
>> Hi,
>>
>> What's the minimum scanning dpi (dot per inch) for photos
>> or x-ray films? I saw a picture with a 72 dpi or 100 dpi and
>> the photo seems to be ok. I thought normal scanners
>> have 300 dpi as minimum. How many percentage image
>> quality loss would a 72 dpi picture produce? Any url for
>> comparisons between 10,000 dpi vs 72 dpi?
>>
>> Parker
>
> Two things to consider: at normal reading distance a good print
> quality requires around 300dpi.
I agree that 300 ppi - NOT dpi, wrong unit - is highly desirable,
butt at the "normal" viewing distances for common print sizes such
as 4 x 6 or 8.5 x 11, very good results can be obtained with 200
ppi, or even less, depending on the subject type of the image, it's
inate quality, and the degree of fine detail it contains.
> For large images which are looked at from a distance, 3000
> pixels in the larger dimension are usually enough.
Total overkill. The normal viewing distance for an 8.5 x 11, the
largest size most home printers can produce, is 1-2 feet. At that
distance, a ppi as small as 150 MAY suffice but 200 surely will.
These equate to 1650 x 1275 at 150 ppi to 2200 x 1700 for 200 ppi.
Again, more pixels will produce a better print IF and ONLY IF the
digital image is of sufficiant quality itself. As an easy example,
it is pretty useless to use 8-10 mega pixel images from a cheap
$100 P & S, but depending on one's definition of "good", even a
DSLR may not necessarily need to be as large as you suggest.
> So if you have "small" images (up to "Letter" size) you scan
> them a 300dpi (or, if you intend to enlarge them in the process,
> at proportionally more DPI, say, 600dpi for a 2x size, up to
> 3000 pixels on the largest side).
>
> These figures are for direct scan-to-print. If you intend to do
> some processing in the middle (especially de-noising), you can
> double the DPI so that the processing software works with more
> information, and then rescale the image afterwards.
>
--
T. Parker
Hi, the reason I asked all about scanning is trying
to figure out how google scan their books. There
is this book with the following page.
After getting it in jpeg. I opened it at photoshop
and use image size option, the pixel/inch says
it is 72. Another program also reported it as
72 dpi. But I know 72 dpi is low res. I was
deciding whether to buy the book at $450
if the original images are high clearer like
600 dpi. But just a while ago. I tried to open
any picture or images. Photoshop reported
it as ether 72 pixel/inch or 100 pixel/inch
(Also I presume dpi or dot/inch is equal to
pixel/inch since one just replace the
word dot with pixel to be more technical
sounding eh?) For jpegs or any pictures that
doesn't come from scanning, how come image
editor reported them as average of 72 pixel/inch.
How do you determine the true pixel/inch
or dpi in an unknown image or picture that
doesn't come from scanning? What
software to use or calculate? I'm not
interested in printing them, I just want to
know what dpi or pixel/inch Google scanned
the book.
BTW.. the following page in google has the
following information in photoshop:
Photoshop Image Size result:
"Pixels Dimension 2.67M
Width = 800 pixels
Height = 1168 pixels
Document Size
Width 11.111 inch
Height 16.222 inch
Resolution 72 pixel/inch"
By running another program like Nero
PhotoSnapViewer, the same jpeg
file as above mentioned 72 dpi.
So based on the above information, what dpi
did google scan the book?
Parker
Parker
T. Parker
>
> - Show quoted text -
Hi,
Are there differences in quality between scanners say
of the same dpi like 300 or 600? I assume that what
scanners do is simply scan the image so there is one
to one corresponse unless some scanners can really
scan better??
Also if you just want 72 dpi output. How come the
image is worse when you scan at 72 and print
at 72 versus overscanning at 300 dpi and downscaling
the image to 72 and print at 72? Is this more of varieties
in scanner quality. I mean If a scanner is well
made and of higher quality. It can scan at 72 dpi
and print them at 72 dpi with the same quality as
overscanning at 300 dpi and then downscaling to
72 and print it at 72??
Parker
T. Parker
Yes. I just tried scanning at 100 dpi a family picture and
part of a book image. The book image shows
some kind of geometrical patterns. But the
original book doesn't have the patterns, how
come scanning it at 100 dpi produce the patterns
that I can't find in a normal camera picture? If it's noise,
how can noise be amplified in the 100 dpi
image?! If half tone dots. How come it only
shows in the scanned image and not the
original? Any clues? I know the solution is
scanning at higher dpi but I just want to
know the theoretical principle behind it.
Thanks dude.
Parker
HEMI-Powered
...
> Hi, the reason I asked all about scanning is trying
> to figure out how google scan their books. There
> is this book with the following page.
Excuse me for asking, but why?
> http://books.google.com/books?id=zb-3YzIn4ZcC&printsec=frontcover
> &dq=biologically+closed+electric+circuits&ei=Z4nWSbebKoqUkQTK_eiA
> Aw#PPA6,M1
>
> After getting it in jpeg. I opened it at photoshop
> and use image size option, the pixel/inch says
> it is 72. Another program also reported it as
> 72 dpi. But I know 72 dpi is low res. I was
> deciding whether to buy the book at $450
> if the original images are high clearer like
> 600 dpi. But just a while ago. I tried to open
> any picture or images. Photoshop reported
> it as ether 72 pixel/inch or 100 pixel/inch
> (Also I presume dpi or dot/inch is equal to
> pixel/inch since one just replace the
> word dot with pixel to be more technical
> sounding eh?) For jpegs or any pictures that
> doesn't come from scanning, how come image
> editor reported them as average of 72 pixel/inch.
> How do you determine the true pixel/inch
> or dpi in an unknown image or picture that
> doesn't come from scanning? What
> software to use or calculate? I'm not
> interested in printing them, I just want to
> know what dpi or pixel/inch Google scanned
> the book.
It really doesn't matter a wit what the designated dpi is, it ONLY
matters how many pixels there are and if it was a scan, the quality
and dpi of the original scanning effort.
You seem to be hung up on the reported dpi. PhotoShop, like all
editors, can change the dpi in EXIF to anything you like WITHOUT
resampling up or down or WITH resampling. This is why it is a
meaningless unit. Please understand, while in digital form, the
ONLY thing that matters is the number of pixels AND the overall
technical quality of the image, which stems primarily from the
quality of the original and the quality of the scan.
> BTW.. the following page in google has the
> following information in photoshop:
>
> http://books.google.com/books?id=zb-3YzIn4ZcC&printsec=frontcover
> &dq=biologically+closed+electric+circuits&ei=Z4nWSbebKoqUkQTK_eiA
> Aw#PPA6,M1
>
> Photoshop Image Size result:
>
> "Pixels Dimension 2.67M
> Width = 800 pixels
> Height = 1168 pixels
What does 2.67M mean? This image is less than a mega pixel. Is that
the uncompressed image size? Again, a totally meaningless number
unless you're saving the image to a format with no compression at
all.
> Document Size
> Width 11.111 inch
> Height 16.222 inch
> Resolution 72 pixel/inch"
This is simple math stemming from the pixel dimensions. 1168 /
16.222" = 72. Nothing magic about this. If all of this is accurate,
and the original photo that is essentially a poster size print is
of sufficient quality, the scan would be just fine. Personally, if
I had a scanner that big, which nobody but the pros do, I would
have tried to create a much larger image so as to be able to
recreate the large print.
> By running another program like Nero
> PhotoSnapViewer, the same jpeg
> file as above mentioned 72 dpi.
>
> So based on the above information, what dpi
> did google scan the book?
>
It looks like 72 based on what you've said, again, just doing the
simple math. But, I still have to ask you why this is of ANY
importance to you? The idea in computer graphics is to achieve some
kind of qualitative and quantitative set of goals depending on the
source image and your intended use(s). It is this and this alone
that should dictate the dpi chosen.
As a final note on scanning from published material, suppose that
the very same film print of the exact same size is scanned and
printed in a newpaper magazine in color, an inexpensive monthly
magazine, a calendar, an inexensive book with matte and porous
paper, and then books of varying quality ending up with a high
quality publication with excellent glossy paper. Each of these
examples will use a different set of criteria for the original scan
and a different set of specs for the half-tone or dot printing
processes.
In short, from newspaper to high quality book, the size and angle
of the half-tone printing process "dots" will be really coarse to
rather fine, resulting in in pretty crapping quality in the
newpaper but often quite lovely quality in a fine color photo book.
Parker, I am a pragmatist, not a theorist, so I can't personally
give you the technical detail on all of this, but perhaps it will
suffice to say that I've been scanning since the early 1900s using
a variety of scanning machines and various photo editors through
the years. I can tell you categorically that the ultimate quality
of the scanned image depends almost entirely on the quality of the
printed photo, and the rest on the quality of the scanner, it's
noise reduction software, and your skill post-processing the image.
And, as I've said a couple times before, the big factor is choosing
a scan dpi that is appropriate not only for the size and quality of
the original printed photo, but YOUR notion of what "quality" even
means, not to mention YOUR ultimate intended usage.
My last comment for now is that MY goal has always been two-fold:
to create a large enough image to fill my monitor screen to at
least 1280 x 960, preferably larger, and to allow for the
possibility of decent prints up to at least 8.5 x 11. I will
readily admit that over the 17 years I've been "practising" this
art, my skill has increased some and so have the capabilities of
the scanners I use. My early work was pretty dismal, to be sure!
Again, please disregard what others may have used for the dpi in
book scans UNLESS your true goal is simply to understand their
methodology. If your intent is to create the best possible images
for yourself from book photos, please strongly consider
overscanning to about 3X the dpi needed for your desired final
image size, resample down, then work on the image. But NEVER use a
dpi over what your scanner can produce optically without
interpolation.
And again, the best of luck in your endeavors!
HEMI-Powered
...
> Hi,
>
> Are there differences in quality between scanners say
> of the same dpi like 300 or 600? I assume that what
> scanners do is simply scan the image so there is one
> to one corresponse unless some scanners can really
> scan better??
Parker, it is just as true for scanners as for anything else you
buy - you generally get what you pay for. As you know, it is
possible to buy a letter size USB scanner for maybe $50, and
upwards from there to as much as you want to pay. You won't find
truly outstanding scanners in stores like Best Buy, though, because
they're tuned into the mass market which values only price.
Start by Googling for the major scanner manufacturers these days. I
haven't looked in some time, but HP and MicroTek used to be among
the best. If your material is letter size or smaller, you don't
need anything larger than that but if you're into scanning larger
books and calendars you will find it VERY helpful to have a legal
size scanner, or even a B-size (11" x 17") because you'll be able
to scan in one pass and avoid the many problems associated with
stitching 2 or more scanned images together.
> Also if you just want 72 dpi output. How come the
> image is worse when you scan at 72 and print
> at 72 versus overscanning at 300 dpi and downscaling
> the image to 72 and print at 72? Is this more of varieties
> in scanner quality. I mean If a scanner is well
> made and of higher quality. It can scan at 72 dpi
> and print them at 72 dpi with the same quality as
> overscanning at 300 dpi and then downscaling to
> 72 and print it at 72??
>
First, let's get the terminology straight, OK?
It is correct to say "dpi" for a scanner because in reality they
ARE scanner X number of "dots" on the page per linear inch across
and down the page. Each "dot" turns into a pixel, as you already
know. Once you have the image in digital form,and decide to print
it, the correct term/unit is "ppi", meaning "pixels per inch".
Printers do NOT output "dots" one-to-one, they use a variety of
techniques to lay down MANY "dots" to form the "pixels" on the
print by combining the colors of ink available, which will be at
least 4 (CMYK, cyan, magenta, yellow, and black), although photo
printers will use 6 or even 10 or more. Ink jet printers often have
published specs like 2700 dpi or even more but that is NOT the true
measure, it is how many INK "dots" the printer is laying down, and
not how many pixels. In the "old days" this was called dithering.
PPI means how many PIXELS do YOU want the printer and the pritner
driver to lay down per linear inch on the paper. Most people feel
that at least 150 ppi is needed, 200 is better, 300 better still -
IF the original scanned photo is of enough quality and so is your
graphics post-processing. Please also note that the type of paper
matters as does the kind of subject(s) you print. Also, the viewing
distance of the print is ultra important.
If you want to produce "good" quality prints yourself of images
with fine detail such as architecture, natures, machinery, cars,
anything with fine detail and sharpness, you'll want to use high-
quality GLOSSY print paper. If you're more into portraits, you
probably want to use matte paper. If you intend to frame your
prints without glass, you may want to use semi-gloss to avoid glare
on the wall. But, it is VERY important to use paper your printer
likes, preferably from the manufacturer, and ALWAYS paper that
doesn't readily absorb the ink which tends to muddy the print.
Hope this helps. If not, ask more questions. And, enjoy your
weekend!
HEMI-Powered
...
>> You haven't mentioned noise removal yet, but if you've tried
>> scanning books, you undoubtedly are now familiar with the fact
>> that the half- tone dots used by printing houses results in a
>> geometrically regular pattern in the digital image that looks
>> like noise. Scanners have noise reduction schemes which vary in
>> their capability, but the BEST thing you can do is to over scan
>> by attemping to use about 3X the dpi you think you need and
>> resizeing downward in your favorite graphics app. What noise
>> remains can be removed at that time.
>
> Yes. I just tried scanning at 100 dpi a family picture and
> part of a book image. The book image shows
> some kind of geometrical patterns. But the
> original book doesn't have the patterns, how
> come scanning it at 100 dpi produce the patterns
> that I can't find in a normal camera picture? If it's noise,
> how can noise be amplified in the 100 dpi
> image?! If half tone dots. How come it only
> shows in the scanned image and not the
> original? Any clues? I know the solution is
> scanning at higher dpi but I just want to
> know the theoretical principle behind it.
> Thanks dude.
>
Great, you're learning fast!
Briefly, prints made from film negs or slides used to use the same
silver hallide chemicals to produce the print image as the film
used to produce the neg. These days, it is done digitally. But, if
the print is from silver negs and prints, they you WILL see an
IRREGULAR pattern of "noise" in your scan depending on how large
the silver crystals are. If you're a film photographer, the term
for this is "grain".
OTOH, pretty much all books and the like are printed with a rather
old process called "half-toning." Think back to early B & W photos
in newspapers. You know that these are comprised of gillions of
little black dots on the paper and are NOT continuous tone like a
film print. The dots are spaced to trick the human eye into
THINKING it is looking at a gray scale but you are really looked at
just black dots or white paper. That's where the "half" comes from,
half black, half white.
Half-toning lays downs dots at an angle for efficiency in the
printing room and to better trick the eye into believing it is a
continous tone photo. A common angle is 45 degrees. Hopefully you
can see that the SIZE of the dots and the distance between the
angular "rows" of dots is the main determining factor for quality.
Please see one of my other replies to you for an example of how
this is perceived.
Color printing is almost the same, except that now the printing
presses lay down 3 colors of dots. Unlike computer monitors which
use the RGB color model, printers almost always use the CMY (cyan,
magenta, yellow) model so a conversion is required. Early color
printing formed black by combining all 3 ink colors and the
lightness or darkness of various parts of the print depending on
how much ink was sprayed on the paper. These days, though, more
colors of ink are used as variants to the basic CMY model plus
black to control light/dark and contrast.
But, whether it be a B & W or a color photo print in a book, there
WILL be for real dots on the page which your scanner picks up and
turns them into what most people call "noise" because these digital
dots tend to make the digital image look like it was painted on
sand or has a screen door screen over it. But, if you look closely,
the pattern is NOT random as it is with real noise, it tends to be
geometrically regular and you can usually see the angle of the
half-toning process.
Your scanner will be able to use various techniquues to minimize
this "noise", a process usually called "descreening." If your
scanner can control the amount of noise reduction, don't overdo it
because it will destroy the fine detail in the scan. Instead, use
PhotoShop or other quality editor to remove the rest of the
"noise."
Please be aware, though, that totally getting rid of the noise and
preserving the fine detail and sharpness in the image are mutually
exclusive, so your results are some sort of compromise for which
you must make a judgement. However, you can simulate some of the
lost detail and sharpness using a variety of sharpening techniques
in your post-processing.
One more time, because this is so important: a half-dozen years
ago, a friend put me on the the BIGGEST single thing to do to
improve my scans, which is to overscan by 3x the dpi you think you
need. Once I started doing that and resampling down before I
attempted any image fixing such as noise reduction, my quality went
way up. And, over the same time period, noise reduction algorithms
in software became better as well.
So, unless your goal in all of this is to simply figure out what
your book publishers are using to scan and print their photos,
don't get all hung up on what PhotoShop reports as the dpi. Worry
ONLY about how the get the best possible quality image at a decent
file size.
John McWilliams
> In article
> <c6250d31-c0b6-45cc...@y34g2000prb.googlegroups.com>,
> want to email it or share it on a web site, then 72dpi is fine. If you
> want to print it with good quality, you need at least 300 dpi. If you
> intend to crop the photo, more dpi is better.
Correct, except they're pixels, not dots. 72-100 ppi is normal for
monitor viewing, and many a fine large print is made at 240 ppi.
--
John McWilliams
HEMI-Powered
...
>> What's the minimum scanning dpi (dot per inch) for photos
> It depends entirely on what you want to do with the photo. If
> you just want to email it or share it on a web site, then 72dpi
> is fine. If you want to print it with good quality, you need at
> least 300 dpi. If you intend to crop the photo, more dpi is
> better.
>
As I'm sure you know, the real issue here is the number of total
pixels it is possible to create by the scanning process depending
on the technical quality of the orignal print and one's scanner.
But, I think that telling everyone they need 300 PPI - NOT dpi - to
get good prints does them the same kind of disservice that digital
camera makers do with their nonsense about "more mega pixels means
better pictures." It is simply NOT enough to have more and more and
more pixels unless the quality in the entire workflow is there to
support that. Depending on the type of subject attempted to scan
and later print, 300 PPI may be really required but if viewing
distance is kept reasonable, 200 works fine in most cases, often
even less.
As to cropping, you are technically correct but it would be better
to "crop" at scan time rather than by intentionally creating
millions of unnecesary pixels. I will strongly agree, though, that
if you already have the image, one's goal should be to be able to
preseve a decent after-crop image size for the desired print size
without needing to resort to resampling up.
For the OP or anyone who's not a theorist, the best way I know of
to test these ideas is to print one's fav subjects using a good
photo printer are several PPI numbers, mix them up, and give them
to a friend to look at. Unless your friend puts their nose on the
print or uses a magnifier, in all likelihood they won't seen any of
the obvious defects from too little PPI, which is usually aliasing.
Your friends will easily spot this if you show them where, but may
not otherwise.
So, the idea is to create images that meet YOUR definition of
quality, and not what the pundits say. The most pragmatic way to do
this is to run some test prints and decide for yourself.
An easy example: Suppose you want to scan book photos or maybe old
family snapshots. And, the max size of the book photos or the 1950s
snapshots is at best 3" 5". If you've tried this yourself, then you
know that it is pointless to overdue the dpi, which in turn blows
up the pixel size, in some vain attempt to get to the magic 300 PPI
because the inate technical quality of the originals in my two
examples just isn't there. This can be easily proven with a few
test scans, but I'm sure everyone is familary with looking at old
drug store snapshots where the quality was pretty dismal, both from
lousy processing but also from Kodak Brownies not exactly being
Leica or Nikons.
Cheers!
T. Parker
>
> - Show quoted text -
Thanks for your incredible details. I think you
must write a book about scanners or write
articles about them in magazines :)
After many hours of testing and trying different
images and software and printing them (that
is I bought a HP 910 Deskjet just to test
your theories). Well. I kinda got it already
and know what kind of quality of scan
google put into the books especially
the free ones allowed by publishers. And my
sole purpose of knowing it all was to decide
whether to buy a used book that costs $450
in the second hand market. After many
analysis. I decided not to get it because the
over 30+ x-ray pictures in the book is of
sufficient quality to be readable enough.
Again this is the book:
I live in the East. We know there is a extra
circulatory system that Western Medicine
simply ignore. And we can manipulate
the biochemistry of the body by manipulating
the extra circulatory system made of subtle
currents of Yin and Yang.
What has this got to do with image processing.
Nothing.
Thanks for all the help. Many useful adequate
Image related knowledge gained and
acquired in the last few days to guide the way.
Parker
HEMI-Powered
...
> Thanks for your incredible details. I think you
> must write a book about scanners or write
> articles about them in magazines :)
You're most welcome, Parker, glad you've found this treatise useful
to you. I appreciate the complement and suggestion to write a book,
but I truly lack the theoretical and technical information to do a
proper job of it. As I've said, I am a consumate pragmatist and
less interested in what books, web sites, and "experts" have to say
and more interested in what I see see with my own two eyes. Please
understand that this is NOT to say that I don't read nor do
research, I most certainly do. And, like you, when I am vexed by a
problem I can't solve or need a nudge in the right direction on a
buying decision, I will post here and other places.
> After many hours of testing and trying different
> images and software and printing them (that
> is I bought a HP 910 Deskjet just to test
> your theories). Well. I kinda got it already
> and know what kind of quality of scan
> google put into the books especially
> the free ones allowed by publishers. And my
> sole purpose of knowing it all was to decide
> whether to buy a used book that costs $450
> in the second hand market. After many
> analysis. I decided not to get it because the
> over 30+ x-ray pictures in the book is of
> sufficient quality to be readable enough.
> Again this is the book:
OK, I see. I have no clue what makes a used book worth $450 but I'm
guessing it is old and rare and you have well justified concerns
about whether you can capture decent scans of the photos in the
book.
> http://books.google.com/books?id=zb-3YzIn4ZcC&printsec=frontcover
> &dq=Biologically+Closed+Electric+Circuits&ei=Z27XSb_MOobikwS62bXq
> Ag#PPA176,M1
>
> I live in the East. We know there is a extra
> circulatory system that Western Medicine
> simply ignore. And we can manipulate
> the biochemistry of the body by manipulating
> the extra circulatory system made of subtle
> currents of Yin and Yang.
>
> What has this got to do with image processing.
> Nothing.
I took a quick look at the web site you provided, thanks. Without
more information, it's tough to comment intelligently on image
quality that (if I understand it correctly at all!) starts with
scans in some/many cases and ends up at least some of the time on a
web site. As you know, web sites tends to use minimal image sizes
to conserve server space and make access as fast as possible. I
have been to car photo web sites though, for example, that show you
a thumbnail picture, and give you a couple of choices for
downloading larger images. Occasionally, obtaining nice large
images requires one to register with the web site and/or pay a fee.
Whether that makes sense is up to the individual.
This DOES have something to do with image processing, though,
because you have a specific problem you're trying to solve. And,
although it sounds like you have decided not to buy this particular
book, the basic principles of what you and I have been discussing
will apply to the next book you buy.
To sum up, I'll take on faith that your HP scanner is up to the
task. Modern scanners generaly do quite well except maybe for the
really bargain basement kind. So, I'd say that when you're
investigating, you MAY want to ask the seller to give you some
insights as to the kind of paper the photos, X-rays, whatever are
printed on to make a decent judgment as to whether the price is
worth it or not.
Let me give you an entirely different way to look at this with a
somewhat obtuse example:
When I was still a teenager circa 1962-63 - yeah, I'm as old as
dirt! <grin> - I bought a book called ""Chrysler Performance
Handbook" by the editors of Hot Rod Magazine. I still have the book
today but it's pretty beat up. There's a truly beautiful photo of a
1963 426 Short Ram Max Wedge drag race engine is color on the
cover, but the rest of the book is printed on paper even worse than
newspapers use. As you might imagine, on a book this old and so
well handled by me so many times, the cover is wrinkled so scanning
was a time-consuming chore.
But, what was REALLY a challenge was getting anything at all decent
from the B & W photos in the text of the book! Imagine if you can
dots in the photos so coarse that I can see them with my naked eye,
and you can gain an appreciation of how much work it was to get
even marginally good scanning results. But, I was well motivated
since this book has been out of print for 45 years. I did an
extensive Google search to see if I could buy a copy in better
shape but they STARTED at $250 and likely were as bad as mine.
Plus, they'd have the same crap photos, right?
Now that I have a modicum of understanding of what you're trying to
accomplish, I can appreciate the struggle you're going through.
Using my obtuse example above to that $450 book you decided to pass
on, IF the photos are valuable enough to you to not only justify
the price but also a likely LONG time scanning and cleaning up the
photos, they go for it.
I'll drop two more examples: I've still got the brochure from the
1975 Plymouth Gran Fury I once owned. Unlike today's brochures, it
is printed like crap. And, I have a color large magazine ad of a
1962 Chrysler 300 hardtop like my father once owned, again printed
like crap plus this ad cost be half a C-note. Why do I tell you
this? Because these two sets of photos have special meaning to mean
and I'm willing to pony up the effort. Likewise, a last example. My
father was a WWII Marine who fought at Saipan, Tinian, and Iwo
Jima. I've got a number of Iwo books and an entire snapshop album
of his Marine experiences that my mother kept during the war. To
say that these photos are bad is an understatement! Yet, they have
so much sentimental value to me that I will spend whatever it takes
to get decent scans.
What all this blather has to do with you is that if you have a good
scanner, you have a good image processor in PS, and you have
special interests, go for it. Use the hints/info I've given you and
that of others to make a judgement on cost vs benefit but I think
that you'll be pleasantly surprised by the results you obtain once
you gain a little experience. And Parker, you won't need a 10 mega
pixel image to be satisfied.
> Thanks for all the help. Many useful adequate
> Image related knowledge gained and
> acquired in the last few days to guide the way.
>
Good luck and have a great weekend. I have to ask. Your name is
similar to T. Boone Pickens' name. Did you ever go the nickname of
"T. Parker?"
Frank ess
T. Parker wrote:
[ ... ]
>
> --
> Jerry, aka HP
>
> "If you are out of work and hungry, eat an environmentalist" -
> Florida billboard- Hide quoted text -
>
> - Show quoted text -
[ ... ]
> I live in the East. We know there is a extra
> circulatory system that Western Medicine
> simply ignore. And we can manipulate
> the biochemistry of the body by manipulating
> the extra circulatory system made of subtle
> currents of Yin and Yang.
>
> What has this got to do with image processing.
> Nothing.
>
> Thanks for all the help. Many useful adequate
> Image related knowledge gained and
> acquired in the last few days to guide the way.
>
>
> Parker
Wow. An Eastern Mystical Circulatory System meets the Creature From An
Alternate Universe.
/Dulce et decorum est/, eh?
--
Frank ess
Don Stauffer
>
> But, I think that telling everyone they need 300 PPI - NOT dpi - to
> get good prints does them the same kind of disservice that digital
> camera makers do with their nonsense about "more mega pixels means
> better pictures." It is simply NOT enough to have more and more and
> more pixels unless the quality in the entire workflow is there to
> support that. Depending on the type of subject attempted to scan
> and later print, 300 PPI may be really required but if viewing
> distance is kept reasonable, 200 works fine in most cases, often
> even less.
>
>
Could we say that high resolution is a necessary but not sufficient
requirement for good results?
van dark
from DPI?
Thank you for your reply and explanation, for me, as ugly boy.
Don Stauffer napsal(a):
T. Parker
> I am very sorry, but what it is PPI, please? Is it a gread difference
> from DPI?
> Thank you for your reply and explanation, for me, as ugly boy.
>
> Don Stauffer napsal(a):
It's like this. When you scanned a material. It's
in DPI.. Dot Per Inch. When you print the material,
it's in PPI.. Pixel Per Inch. This is because the
DPI term is already reserved for the printer
dot capability. So when you scanned a material
at say 100 dpi and use a printer with 1200
dpi, you print at 100 ppi. So a pixel in the paper
printed is composed of 12 printer dots. Gets?
Parker
>
>
>
> > HEMI-Powered wrote:
>
> >> But, I think that telling everyone they need 300 PPI - NOT dpi - to
> >> get good prints does them the same kind of disservice that digital
> >> camera makers do with their nonsense about "more mega pixels means
> >> better pictures." It is simply NOT enough to have more and more and
> >> more pixels unless the quality in the entire workflow is there to
> >> support that. Depending on the type of subject attempted to scan and
> >> later print, 300 PPI may be really required but if viewing distance is
> >> kept reasonable, 200 works fine in most cases, often even less.
>
> > Could we say that high resolution is a necessary but not sufficient
> > requirement for good results?- Hide quoted text -
Ray Fischer
>I am very sorry, but what it is PPI, please? Is it a gread difference
>from DPI?
>Thank you for your reply and explanation, for me, as ugly boy.
DPI - dots per inch - indicates the number of dots of ink per inch
that a printer can put onto a piece of paper. If here are four
different color inks then then each dot can be one of those four
colors.
PPI - pixels per inch - refers to the number of arbitrary-colored
pixels there are per inch.
The important practical difference is that a printer cannot make each
dot an arbitrary color and it has to use multiple dots to build up a
given color. That makes the real resolution of the printer much less
than the DPI rating.
--
Ray Fischer
rfis...@sonic.net
Don Stauffer
A digital photograph is essentially an array, a grid of columns and
rows. Each element in the array is called a pixel, short for picture
element. There is a brightness value (and color values in a color
image) for each pixel.
"Dot" can mean a lot of different things. A halftone dot on a printing
press is the same as a "pixel". The dot in an inkjet, dot matrix, or
laser printer is NOT the same as a pixel. In inkjet, DM, or laser
printers all the dots are the same size. Lets talk about black and
white, because it is easier to see, but color is done with basically the
same idea.
An inkjet printer prints various brightness pixels by printing an array-
say a nine by nine array of dots for each pixel. If it prints 81 black
dots the resulting pixel is very black. For the brightest pixels it
prints nothing. One black dot is the shade of grey just darker than
white, two dots is the shade two steps darker than white, etc.
Now, the pattern within the nine by nine square is something else, lots
of various algorithms around to optimize the "shape" of the pattern to
make the most pleasing picture. Google dithering and error diffusion.
HEMI-Powered
jour ...
>> But, I think that telling everyone they need 300 PPI - NOT dpi
Well said, Dan! Yes, I agree with mathematicians who say exactly
that. In short, there is a HUGE difference between a 10 mega pixel
$100 P & S with an itty bitty plastic lens system and an image
processor aimed as speed rather than quality, and a decent EVF and
even MORE quality difference to a competent DSLR with other than it's
typically soft kit lens.
And, to agree with you even more, including specifically this thread
which is about scanning, in my view and in my long experience, it is
the ENTIRE workflow that has importance here, including but not
limited to, the camera or scanner, the inate "quality" of the
"original" scene or document, the quality of the image processor
involved, any noise reduction involved, lighting/exposure, et al, the
competence of the graphics app used to post process the images, and
of course, the skill of the user.
Have a nice day.
--
HP, aka Jerry
"Laid off yet? Keep buying foreign and you soon will be!" - popular
bumper sticker
HEMI-Powered
...
> I am very sorry, but what it is PPI, please? Is it a gread
> difference from DPI?
PPI means "pixels per inch" and refers to how many IMAGE pixels you
tell your printer driver to lay down per linear inch across and
down the page. It is ONLY applicable to printing, it has no real
meaning other than the actual number of pixels available via the
chosen resolution during image processing, in other words, the true
mega pixels of the image itself.
DPI means "dots per inch" and is generally only correctly used for
hardware devices like scanners which actually look at individual
"dots" across and down a page of material and turn each of them
into digital pixels. In scanning, the higher the DPI, the more
pixels one will get in the scanned image.
DPI is often INcorrectly used to refer to the image in digital form
or when printing, and this myth is often perpetuated by the
graphics editor developers themselves who use the term in EXIF.
But, as I discussed with Parker, the EXIF DPI can be changed by the
user either with resampling up or down or simply changing the
number itself.
It is therefore technically correct to use DPI for scanning and PPI
for printing, again, remembering that there are no "pixels per
inch" nor "dots per inch" in a digital image unless/until it is
transferred either from paper or too paper.
Hope this helps, if not, please ask a clarifying question and have
a great day.
--
HEMI-Powered
...
>> I am very sorry, but what it is PPI, please? Is it a gread
>> difference from DPI?
>> Thank you for your reply and explanation, for me, as ugly boy.
>>
>> Don Stauffer napsal(a):
>
> It's like this. When you scanned a material. It's
> in DPI.. Dot Per Inch. When you print the material,
> it's in PPI.. Pixel Per Inch. This is because the
> DPI term is already reserved for the printer
> dot capability. So when you scanned a material
> at say 100 dpi and use a printer with 1200
> dpi, you print at 100 ppi. So a pixel in the paper
> printed is composed of 12 printer dots. Gets?
>
You got it, Parker! My congrats to you for coming up to speed on
this often confusing esoteric subject so quickly. And, I sincerly
hope that your future needs for scanning rare books and making
sense out of photos on web sites is now right for you.
One comment, except back in the old dot matrix days, it really
isn't possible to know the true DPI of a given printer such as an
ink jet, laser printer, or any other popular modern technology
since the printer itself uses as many "dots" as it needs to work
it's magic of proving superior sharpness/detail and quality
fidelity of color and brightness/contrast to the original image.
This is partially the job of the printer driver to convert pixels
according to the PPI available and partly the type of technology of
the printer.
For ink jet photo printers that use the common CMYK 4 ink colors
plus 2 to 5 or more additional "photo" colors, the actual DPI of
the printer is even harder to know. For example, my Canon Pixma
6600 has 6 ink cartridges and advertises some wild number of DPI
which has no real meaning while newer, more expensive and higher
quality ink jets MAY have 10 or even more ink cartridges.
It's been said that the proof of the pudding is in the eating, and
so it is true for printing as well. I believe you've convinced
yourself by careful experimentation of what is and is not required
when scanning and how various pixel resolutions as translate to
varying PPI alter the quality of the print. Again, my congrats for
learning this so quickly.
>>
>> > HEMI-Powered wrote:
>>
>> >> But, I think that telling everyone they need 300 PPI - NOT
>> >> dpi - to get good prints does them the same kind of
>> >> disservice that digital camera makers do with their nonsense
>> >> about "more mega pixels means better pictures." It is simply
>> >> NOT enough to have more and more and more pixels unless the
>> >> quality in the entire workflow is there to support that.
>> >> Depending on the type of subject attempted to scan and
>> >> later print, 300 PPI may be really required but if viewing
>> >> distance is kept reasonable, 200 works fine in most cases,
>> >> often even less.
>>
>> > Could we say that high resolution is a necessary but not
>> > sufficient requirement for good results?- Hide quoted text -
>>
>> - Show quoted text -
>
>
--
HEMI-Powered
jour ...
>>> Could we say that high resolution is a necessary but not
>>> sufficient requirement for good results?
>
> A digital photograph is essentially an array, a grid of columns
> and rows. Each element in the array is called a pixel, short
> for picture element. There is a brightness value (and color
> values in a color image) for each pixel.
>
> "Dot" can mean a lot of different things. A halftone dot on a
> printing press is the same as a "pixel". The dot in an inkjet,
> dot matrix, or laser printer is NOT the same as a pixel. In
> inkjet, DM, or laser printers all the dots are the same size.
> Lets talk about black and white, because it is easier to see,
> but color is done with basically the same idea.
Yes to the first sentence, but no to the second sentence. In half-
tone printing, not only are more "dots" typically laid down than
digital pixels, the size and color(s) of the dots also vary. This
can be visually verified by looking at a typical printed photo in
various publications such as newspapers, magazines, calendars, and
books of varying quality with at least a 2x magnifier. Even more
can be discerned with a 4X-8X magnifier which allows you to see the
much smaller dots of a quality photo print.
> An inkjet printer prints various brightness pixels by printing
> an array- say a nine by nine array of dots for each pixel. If
> it prints 81 black dots the resulting pixel is very black. For
> the brightest pixels it prints nothing. One black dot is the
> shade of grey just darker than white, two dots is the shade two
> steps darker than white, etc.
How a given printer actually works has as much to do with it's
image processing system and number of ink colors available as it
does for the complex "dithering" you speak. For example, black
black can be simulated by over printing multiple CMY colors or by
true black ink or both. And, the overall brightness and contrast
across the image can be attenuated by mixing all of the ink colors
available with black. I can't speak to the specific grid that my
printer or any other actually uses, but I'd be quite surprised if
it were always a simple X by X array of dots.
> Now, the pattern within the nine by nine square is something
> else, lots of various algorithms around to optimize the "shape"
> of the pattern to make the most pleasing picture. Google
> dithering and error diffusion.
>
Good idea. The common defintion of dithering has become blurred -
no pun intended - over the years since it was originally
mathematically defined to mean the mixing of various inks to
simulate not only as broad a spectrum of colors as possible but
also attempts at anti-aliasing which may be part of the process or
optional. Dithering used to be a faux printing idea in reality
since printers in the 1970s through 1990s weren't capable of even
approaching 24 bit color to their print engine, much less the non-
noise bits of 48 bit color.
I am aware of the fundamentals of error diffusion but I expect that
you're technical and mathematical knowledge is much superior to
mine. So, I'll just end by saying that "dithering" is a very good
analogy to describe the process in a way that most people can
understand even if it is used much differently by modern photo
printers.
And, I won't even attempt to get into the very expensive commercial
printer technologies as I have essentially zero knowledge in those
areas.
J�rgen Exner
>On Apr 6, 6:42�am, van dark <van.d...@yahoo.co.uk> wrote:
>> I am very sorry, but what it is PPI, please? Is it a gread difference
>> from DPI?
>
>in DPI.. Dot Per Inch.
Sort of, see below.
> When you print the material,
>it's in PPI.. Pixel Per Inch.
No!
>This is because the
>DPI term is already reserved for the printer
>dot capability.
Exactly. So you are always printing in DPI, not in PPI.
>So when you scanned a material
>at say 100 dpi and use a printer with 1200
>dpi, you print at 100 ppi.
No, at least not necessarily.
>So a pixel in the paper
>printed is composed of 12 printer dots. Gets?
Maybe by coincidence.
First of all you got it right that DPI and PPI _ALWAYS(!)_ refer to
hardware, never to a JPEG file or similar data.
The easiest term is PPI. If you have a picture with 2000x3000 pixel and
when you display it (in its native size without any interpolation) on a
computer monitor that has 200PPI, then its size will be 10x15".
That is the native resolution of the monitor.
Key criteria is that each pixel on the display can display all colors in
the given colour space. Some manufacturers try to muddle the waters
quoting 'sub-pixel' (single-color elements on LCDs) density. That is not
what pixel are!
DPI refers to the physical resolution of a printer. A you mentioned
already on a printer a single pixel(!) has to be composed of many much
smaller dots(!) to create the proper colour. Terminology is somewhat
muddled by marketing, because often manufacturers are not clear about if
they are talking about positioning accuracy (you can place some dot
every 1/DPI inches), if different colours are staggered or if each
colour individually can be placed at 1/DPI inches, or if they are
talking about the minimum size of their ink dropplets.
In any case, key criteria is that a dot is a single drop of ink/single
impact of a hammer/single strike of a matrix needle/single charge of a
laser printer drum and therefore can only display one standard size
element in one of the basic colors.
As for scanners the proper term is SPI (samples per inch) and again it
refers to the physical resolution of the hardware devise. However SPI
never caught on and manufacturers have misleadingly been using DPI
instead.
Maybe the reason was that in ancient times you wanted to buy a scanner
and a printer with matching resolutions, such that you could reproduce a
black&white document in its original size at the highest quality without
interpolation. Of course nowadays with colour printers and scanners and
printers having much higher resolutions anyway that is a mood point.
Anyway, the term DPI stuck with scanned pictures, although its only
relevance is to indicate the physical size of the original document,
such that you can reproduce it at a given magnification. That EXIF value
in the JPEG file should be called "size of original document" or at
least "SPI of scanner that was used to create this picture". But it
isn't, so we need to learn that that value has nothing to do with the
normal meaning of DPI.
jue
HEMI-Powered
jour ...
>>> I am very sorry, but what it is PPI, please? Is it a gread
>>> difference from DPI?
>>
>>It's like this. When you scanned a material. It's in DPI.. Dot
>>Per Inch.
>
> Sort of, see below.
I disagree. Not sort of, always.
>> When you print the material, it's in PPI.. Pixel Per Inch.
>
>
> No!
Yes!
>>This is because the
>>DPI term is already reserved for the printer dot capability.
>
> Exactly. So you are always printing in DPI, not in PPI.
Exactly how? Printers do not understand "dots" they only understand
what is coming to them from their driver software, and that ONLY
understands pixels. Where would dots come from and how would a
printer calculate DPI?
>
>>So when you scanned a material
>>at say 100 dpi and use a printer with 1200 dpi, you print at 100
>>ppi.
>
> No, at least not necessarily.
>
>>So a pixel in the paper printed is composed of 12 printer dots.
>>Gets?
>
> Maybe by coincidence.
>
> First of all you got it right that DPI and PPI _ALWAYS(!)_ refer
> to hardware, never to a JPEG file or similar data.
>
> The easiest term is PPI. If you have a picture with 2000x3000
> pixel and when you display it (in its native size without any
> interpolation) on a computer monitor that has 200PPI, then its
> size will be 10x15". That is the native resolution of the
> monitor. Key criteria is that each pixel on the display can
> display all colors in the given colour space. Some manufacturers
> try to muddle the waters quoting 'sub-pixel' (single-color
> elements on LCDs) density. That is not what pixel are!
Computer monitors may have a theoretical PPI, but more so, a native
resolution and a number of lesser ones, where they support various
resolutions in a variety of manners according to one's video card
driver and it's complex software. But, one more time, it doesn't
matter a tinkers damn anything at all in the digital world wrt to
PPI or DPI unless and until one actually attempts to lay down
"dots" on a printed page by some sort of output media device. To
argue with that is futile as everyone including the theorticians
and elitists around here ALL agree that it is the number of PIXELS
the printer driver is able to send to the hardware that determines
ultimate print quality - all other factors being equal though they
seldom are. Thus, it is FAR easier to understand as well as
avoiding mental gear changes to calculate a requirement for INPUT
pixels divided by the OUTPUT paper size to see if there are or are
not enough of the little devils to meet one's own definition of
"quality."
> DPI refers to the physical resolution of a printer. A you
> mentioned already on a printer a single pixel(!) has to be
> composed of many much smaller dots(!) to create the proper
> colour. Terminology is somewhat muddled by marketing, because
> often manufacturers are not clear about if they are talking
> about positioning accuracy (you can place some dot every 1/DPI
> inches), if different colours are staggered or if each colour
> individually can be placed at 1/DPI inches, or if they are
> talking about the minimum size of their ink dropplets.
There is simply NO adequate defintion of a printer's DPI that uses
a mixing of inks to impart a printed image. Thousands of "dots" may
be laid down where only dozens or 100 pixels are intended to go.
> In any case, key criteria is that a dot is a single drop of
> ink/single impact of a hammer/single strike of a matrix
> needle/single charge of a laser printer drum and therefore can
> only display one standard size element in one of the basic
> colors.
>
> As for scanners the proper term is SPI (samples per inch) and
> again it refers to the physical resolution of the hardware
> devise. However SPI never caught on and manufacturers have
> misleadingly been using DPI instead.
> Maybe the reason was that in ancient times you wanted to buy a
> scanner and a printer with matching resolutions, such that you
> could reproduce a black&white document in its original size at
> the highest quality without interpolation. Of course nowadays
> with colour printers and scanners and printers having much
> higher resolutions anyway that is a mood point.
This is nothing more than semantics. It is technically correct, but
NOBODY understands sampling theory whether it is in statistics or
in scanning but EVERYBODY (eventually) understands the simplicity
of imputting X number of physical dots across and down a page and
converting each of them to a pixel. So, why muddy things up other
than to advance an elitist agenda?
> Anyway, the term DPI stuck with scanned pictures, although its
> only relevance is to indicate the physical size of the original
> document, such that you can reproduce it at a given
> magnification. That EXIF value in the JPEG file should be called
> "size of original document" or at least "SPI of scanner that was
> used to create this picture". But it isn't, so we need to learn
> that that value has nothing to do with the normal meaning of
> DPI.
>
This much I agree with, which is exactl what I told Parker, the OP.
DPI was originally both a scanner and a dot matrix printer unit,
and even had a reasonable correctness in early HP Laser Jet
printers. But, once printer technology advanced beyond output
devices that had a one-to-one relationship to digital picture
elements, it has made ZERO sense to continue to use an obviously
outdated unit.
I wonder if the OP here would think of you taking my thoughts apart
and attempting to dispute them all when all he wanted to know was
how to judge what is and is not "good"?
But, have a nice day anyway.
J�rgen Exner
>J�rgen Exner added these comments in the current discussion du
>jour ...
>
>>>> I am very sorry, but what it is PPI, please? Is it a gread
>>>> difference from DPI?
>>>
>>>It's like this. When you scanned a material. It's in DPI.. Dot
>>>Per Inch.
>>
>> Sort of, see below.
>
>I disagree. Not sort of, always.
Correct unit would be SPI (Samples Per Inch) but it is not commonly
used.
>>> When you print the material, it's in PPI.. Pixel Per Inch.
>>
>>
>> No!
>
>
>Yes!
>
>>>This is because the
>>>DPI term is already reserved for the printer dot capability.
>>
>> Exactly. So you are always printing in DPI, not in PPI.
>
>Exactly how?
Exactly as T. Parker wrote: "DPI term is already reserved for the
printer dot capability".
>Printers do not understand "dots" they only understand
>what is coming to them from their driver software, and that ONLY
>understands pixels. Where would dots come from and how would a
>printer calculate DPI?
The printer doesn't 'calculate' DPI. DPI is a physical, native property
of the hardware. And the printer doesn't understand anything but dots,
i.e. the smallest item of ink dots or matrix needles or hammer impact or
...
It is the printer driver's job to convert whatever format the picture
comes in (bitmap, JPEG, vector graphic, ...) into a sequence of
individual dots which can be printed by the printer.
>Computer monitors may have a theoretical PPI, but more so, a native
>resolution and a number of lesser ones, where they support various
>resolutions in a variety of manners according to one's video card
>driver and it's complex software.
Quite right. And if you take the width of that resolution and divide it
by the physical width of the displayed area then you get the PPI. It's
very basic mathematics.
>But, one more time, it doesn't
>matter a tinkers damn anything at all in the digital world wrt to
>PPI or DPI unless and until one actually attempts to lay down
>"dots" on a printed page by some sort of output media device. To
>argue with that is futile as everyone including the theorticians
>and elitists around here ALL agree that it is the number of PIXELS
>the printer driver is able to send to the hardware
Except that the hardware wouldn't know what to do with those pixels
because a printer can print dots only. And it's the task of the printer
driver to convert those pixels (or vectors or bitmaps or ...) into dots
such that the printer will be happy.
>There is simply NO adequate defintion of a printer's DPI that uses
>a mixing of inks to impart a printed image. Thousands of "dots" may
>be laid down where only dozens or 100 pixels are intended to go.
Quite right. That's why modern printers have 5000 DPI and more.
jue
Ray Fischer
>Jürgen Exner added these comments in the current discussion du
>jour ...
>
>>>> I am very sorry, but what it is PPI, please? Is it a gread
>>>> difference from DPI?
>>>
>>>It's like this. When you scanned a material. It's in DPI.. Dot
>>>Per Inch.
>>
>> Sort of, see below.
>
>I disagree. Not sort of, always.
>>> When you print the material, it's in PPI.. Pixel Per Inch.
>>
>>
>> No!
>
>Yes!
Printers do not, as a rule, print pixels. The driver will blend
pixels in order to create a smooth dither pattern.
>>>This is because the
>>>DPI term is already reserved for the printer dot capability.
>>
>> Exactly. So you are always printing in DPI, not in PPI.
>
>Exactly how? Printers do not understand "dots" they only understand
>what is coming to them from their driver software, and that ONLY
>understands pixels.
The printer doesn't know pixels. It knows dots. The driver converts
pixels to dots.
--
Ray Fischer
rfis...@sonic.net