pixel dimesions?

[nickmt]

Hi all, I'm writing here as I'm trying to understand up-resing small files, and starting with the question what exactly does the pixel dimension number in image size actually have to do with width and height of the image in pixels? For example I'm working with a silly little image that has a pixel dimension stated to be 318.4K. The, almost exactly square, width and height are 285W x 286H pixels. Multiplying W x H = 81,510 pixels. That's not 318.4K so what is this number telling me in relation to the "size" of my image? Do bytes, kilobytes, megabytes, etc have a direct relationship to pixels? Does that matter re my confusion??

This was a, obviously very quickly, scanned image I pulled from a folder I haven't visited for a while, so I can't remember exactly how it was scanned, but clearly not for printing. I of course realize rescanning is absolutely the way to approach this image if I want to print it, but my question about how to use the pixel dimension information to my advantage, and in the context of up-resing, still stands. The document size is 1.188 inches W x 1.192 inches H. Resolution is 240. Thanks so much for your consideration.

Nick

 

[Tom Mann]

Here's the arithmetic for the example you gave.

# of pixels = 285 x 286 = 81510 pixels = 81.5 Kpixels

From the numbers you gave, that was almost certainly a CMYK image, so you had 4 channels.
Note: RGB images would have only 3 channels, grayscale 1 channel.

Also, from the numbers you gave, it was almost certainly an 8 bit per channel (8 bpc) image, not 16 bits per channel.

Therefore the total number of bits per pixel was 4 x 8 = 32.

Therefore the total number of bits in the image was 32 x 81510 = 326040 bits = 326 Kb.

1 KByte = 1024 bits, which is abbreviated as 1 KB. (notice the change from lower case "b" to upper case "B").

Therefore the image contains 326040 bits / 1024 (bits per KB) = 318.39 KB.

I should point out that you used the term "pixel dimensions" incorrectly. The "pixel dimensions" for your image would conventionally be stated as "285x286", NOT as the product of those two numbers (as you stated).

The product of those two numbers is the number of pixels in the image. Divide the product by 1,000,000 and you have have the image size in Megapixels (ie, 0.08 Mpixels).

Hopefully, the above answers all the questions in your 1st paragraph.

--------------

With respect to your 2nd paragraph, I'm not sure exactly what you are asking. You correctly gave the physical size of the image at 240 pixels per inch (ppi), so you obviously know how to do the conversion.

If you are asking how to make a physically larger image without changing the data, your only choice is to decrease the number of ppi. In Photoshop terminology, this is "without resampling". Put differently, the number of pixels stays constant.

For example, if you went from 240 ppi to 24 ppi, your image would become roughly 12 inches on each side. However, unless you were viewing it from distances typical of large signs, ie 10 or more feet away, each pixel in the image would be clearly visible as a square of uniform color. This is NOT "up-rez'ing".

OTOH, if you asked for a 12 inch by 12 inch image, but checked the "with resampling" option in the Image Size dialog box, Photoshop would keep the ppi at 240 and attempt to guess (aka, interpolate) 10 new values between each of the old pixels. This IS what is conventionally called "up-rez'ing". The result would be a 12" x 12" image that is smooth to the eye when viewed from normal viewing distances. It wouldn't look blocky, but instead, it would look blurry. So, the name, up-rez'ing is a misnomer, but everyone knows what you mean.

That being said, if you only attempt to increase the physical size of the image by a factor of 2 or so (ie, double the number of pixels along each axis), and you started with a very sharp image, the typical viewer won't notice the blur, and you will have a larger image.

The bottom line is that "there is no free lunch". If you want an image of a certain size, you must make sure that you start with a large enough image (in pixel dimensions - the conventional definition, NOT your definition).

HTH,

Tom M

 

[nickmt]

Thank you Tom. This helps a lot. First I must apologize, and am very puzzled as the number I posted yesterday as "pixel dimensions", as 318.4K is a different number today!!! It led you to you figure out it must be cymk, which makes total sense... But today (!) upon opening the image the number is 79.6K, and that makes sense too(!), as the image is in fact a greyscale image, which you would have known if I'd given you the accurate number. In a way it really doesn't matter as I've absorbed your overall lesson. But to explain a bit, what I had been doing was tying to increase the file size by adding duplicate layers, wondering if doing so would add some sort of integrity to the file that might have it up-rez more successfully. I had no luck. I don't know how I got that number to change. Just now I again opened the image (which is just one layer), and made three duplicate layers, saved it twice, flattened, and not flattened, hoping the file size would increase, hoping I'd understand how I got from 79.6 to 318.4, but I could not duplicate the change. Weird.

Sorry.

At any rate following your example, but given it is a grey scale image, its total number of bits per channel would be 1 x 8 = 8, and the total number of bits in the image is 1 x 81510 = 81.5Kb

And 81510 bits / 1024 bits per KB = 79.59 KB and that's the "pixel dimensions" number I have. So it explains the how photoshop uses the term pixel dimensions. And that was my first question. I also understand your explanation that the conventional usage of the term 'pixel dimensions' indicates pixel WxH to the user, not it's product, or it's product x bits / 1024. It is confusing that ps does that while there's a conventional usage of the term.

As far as my second paragraph your advice confirms my suspicions - I need a new scan.

I really appreciate the help!! Thanks again!

Nick