And yes, the dpi will not tell us the final print size or quality, except that the higher the dpi, the larger the printed image can be, and still look good. Given the same page dimensions, the more pixels, the better, and that’s reflected in the dpi. If the image is fairly small, then the difference might negligible, or unable to produce on your printer.
And unexpectedly, as suggested I downloaded the most recent develoopment build, ran a NEF file through RT, only changing the white balance, and saved it. Previously, when I looked at the resulting output jpg file, the dpi as reported by the GIMP was 72 dpi. I didn’t notice what the dimensions in pixels were. This time, I looked at both. The original jpg files was 300 dpi at 5568x3712. The processed jpg had dimensions of 5576x3716 - and a dpi of 300!
So, now I’m a happy camper. The programmers corrected what I feel was a mistake, and for that I give them a long burst of applause - especially since now I am even less likely to plunk down the multi-dollars for PS!
Yes, in the print’s dpi but not in the metadata dpi. That’s what I tried to explain above: Given a page dimension and a minimum dpi value for the print will lead to a minimum number of pixels of the digital image, regardless of what is written into the file’s metadata, may it be 72 dpi or 300 dpi.
If you have a 300 dpi (metadata) image of size 100 × 100 pixels, you will not be able to print it larger than 1/3 inch × 1/3 inch if you need a 300 dpi print resolution. You would need at least 300 × 300 pixels for a one square inch print and 3000 × 3000 pixels for a 10 inch × 10 inch print. But the exact same is true for a 72 dpi (metadata) image that you have to print at 300 dpi. The absolute number of pixels is crucial, the dpi in the file’s metadata is just for convenience in your workflow.
And now please check the absolute number of pixels of the “old” 72 dpi file. Unless you changed some processing settings it should have exactly the dimension 5576 × 3716 pixels. That means it is exactly the same image.
I’ve had a couple of suggestions over the past few hours. The first was to refer to it as ppi instead of dpi, so as to reflect pixels per inch. So, if we have two images of the same dimensions, and one has 300 ppi and the other has 72 ppi, it’s going to have the same number of pixels? That’s like saying my back yard, the same size as yours, has 3 dogs per square foot, while yours has 10. Who has more dogs in their back yard?
I had a friend who tried to print out an image that looked good on the web, but lousy on paper. I got his file, changed the ppi (from 72 to 200, I recall), and printed the result. It looked good. I suppose that changing the ppi didn’t make any difference, since they were both the exact same picture. Bottom line: ppi (or dpi) is meaningless number. Right?
Oh, and the second suggestion I got about an hour ago from a friend who told me that the ppi was set by the program “for its convenience”. And, that discussing it further was pointless, because he “didn’t give a shit”. And now, since the installation of a new version of RT is outputting a file at 300 ppi, I’m going to take his advice,to stop beating it to death, and to no longer “give a shit”.
The image from your camera isn’t 300DPI, unless you are doing close-up photography(*). DPI in captured images are only relevant in scanners. AFAIK cameras either don’t even add a a DPI value to the exif, so processing software assume the default 72DPI.
(*) In astrophotograpy, they must be using dots-per-parsec…
Well, I once benchmarked various setups for macro photography (various lenses with various close-up add-on lenses) and my performance measure was in pixels per millimeter of subject
Makes sense, but it is only a useful performance measure when you limit it to the closest possible subject/focusing distance. I doubt that the camera writes the dpi value in the metadata based on the focusing distance and the angle of view, which should both be available to the camera. Maybe you could automate it with Magic Lantern if you are using a canon camera ;-).
The whole thing is based on the fact that the photographic industry, camera’s or softwaremakers, don’t know the difference between dpi (dots per inch) and ppi (pixels per inch). DPI means dots per inch. It is the number of inkspots wich a printer puts on an inch. For example, 1440 DPI and 2880 DPI are for fine printing. The dots don’t have any relationship with the pixels.
A sensor produces pixels. That’s why a 16 Megapixel file is always the same size in bytes, wether it is 72 ppi or 300 ppi.
The unit PPI is a hexadecimal word of 8 positions in the metadata. It is an indication how the pixes are linear placed in the output. So 300 ppi means that 1 inch of a picture contains 300 pixels. If you should change that to, lets say, 30 pixels on 1 inch this means that the picture will be linear ten times bigger. With the same total amount of pixels, the 16 Megapixelfile wil stay 16 Megapixel.
That’s why the scanoperators in the past asked if the customer was in need of a 25 Megabyte (A4) or a 50 Megabyte (A3) Tiff file.
A good demonstration is to set a file in 30 ppi (or if you still like DPI) and a copy of the same file in 300 PPI. Put them on two screens and enlarge both files to the same format (height and width) on screen. You won’t notice any difference, even when the enlargement is so big that you can see the individual pixels.
It’s like making a fence around your garden: do I place 1 pole on 10 meter? Or 10 poles on 10 meter (wich is 1 pole on 1 meter).
Rather than get into the discussion of which is the correct
expression (my Nikon says the image is 300 dpi, and the GIMP refers to
scaling a image referring to dpi). What I have found is if I do scale a
300 dpi image to 72, it makes it smaller and more suitable for uploading
to a website for viewing. But, if I choose to print it, the 72 dpi
version, as an 8x10 inch print, for example, it won’t look as sharp as
if I decided to print the 300 dpi version as an 8x10.
I don’t tell the printer to change the amount of dots per inch,
just to print the image. Perhaps it can read the data and see the ppi,
and adjust accordingly?
Sure. Using the GIMP, I used a drop-down menu item labelled “scale”, which gives you the option of changing the dimensions of the photo and/or the DPI. With my current camera, the DPI is shown as 300. To change it to 72, you just need to replace the " 300" with “72”. Then you have the option of saving the image in various formats, such as jpg, PNG, tif, etc.
I do need to change my previous answer to this question: When using
the drop-down menu (“Image”), when then selecting “Scale Image”, the
options shown are to change the width and height, and then the X and Y
resolutions, which are shown initially in pixels/in, not DPI. After
you’ve made that change to 72, you can then (in Windows) right-click on
the image, go to Properties, then choose Details, and scroll down to the
information on Image, and see the Dimensions, among which are the
Horizontal and Vertical resolutions, given as 72 dpi.
That’s a good clear explanation as far as it goes.
My experience is that I can use the scale image dialog to change the X and Y resolution and when I then export the image with a new name, the Windows file properties report the new resolutions - but with the same pixel dimensions.
However, you said that you end up with a smaller file which is “more suitable for uploading
to a website”, so what are you doing after using the Scale Image dialog?
I often change the width and height, too, to something like having
the smaller of the two dimensions being 2048 - and leaving the “chain”
intact, so as to not distort the result.
