% of non-transparent pixels?

[cliffaddy]

Is there some way to get the % of pixels in an image that are non-transparent? I'm trying to get an approximation of how much ink would be used in printing an image.

[JJMack]

How much ink will be used depends on the composite image content, image size, pixel opacity,  printer model, the paper used and print quality set.  Higher quality paper and high print quality settings use more ink.  It not just pixels.

[cliffaddy]

Yes, I'm perfectly aware of all that. As I said, an approximation that'll be based on other factors and my own empirical testing. All I need is the pixel count.

[rob day]

The Histogram panel in extended view will give you the pixel count of a selection along with the mean and median pixel value. You could set the Channel to Luminosity and get the Mean pixel value along with the count.

[cliffaddy]

Hmm, that's really interesting but doesn't seem to be giving the result I expect. For example, I have an image that's around 4500 x 4700, hence around 19M pixels. The artwork covers 60-80%, but the histogram output (Channel: Luminosity, Source: Entire Image) shows a pixel count of only 308K, i.e. about 1.5%

[rob day]

Try clicking the refresh icon in the upper right corner:

[Chuck Uebele]

If you have transparent pixels, make a selection by ctrl/cmd clicking on the layer icon, then use the histogram to get a pixel count.

[cliffaddy]

Ah, this works, with the caveat that you have to click the refresh button afterwards.

[rob day]

It doesn’t seem like transparency tells you much about ink coverage. A pixel could be completely transparent (white), or partially transparent (gray).

Obviously these flattened images would have very different ink coverages, and different Mean values:

[cliffaddy]

There are no gradients in any of the work. Anti-aliasing is a trivial amount. I don't need a precise answer, if I'm within 20% that's more than good enough.

[rob day]

There are no gradients in any of the work.

I was just showing a quick example of how the ink coverage would be reflected in the mean value—the ink coverage (and mean) would also change in a photographic image

[Lumigraphics]

Pixel count is inaccurate for this. CMYK with a large line screen will be vastly different than RGB on an inkjet. Different colors use different dot patterns (you are aware than CMYK can use different screens and get the same visual output?) I know its tempting but any print shop or service bureau can explain.

[rob day]

Right, it would be the output dimensions that matter, the OP’s 4500 x 4700 pixel dimension could be output as 45" x 47" or 4.5" x 4.7" obviously using very different amounts of ink.

If you know the maximum amount of ink the printer uses per square inch, the mean luminosty could help get an approximation—if the mean is 128 then the ink usage would be 50% of the max.