Question about photo printing, resolution, and DPI

I’m trying to print some photos in 4x6 forms using snapfish, who use a 300 DPI print process.

By default, photoshop loads my images at 180 DPI (not sure where that number comes from), and the camera’s native res is 4000x3000, resulting in a starting image that would result in a 22"x16" print.

Anyway - I’ve read that photos meant to be printed need to be sharpened more than what looks right on a monitor, but also that when you rescale an image, there can be artifacts introduced to the sharpening that make it look worse - and that you should size it first, and then sharpen it, to avoid that.

So if I’m working with 4x6 images at 300 DPI, do I simply resize the image to 1200x1800 pixels first? Do I also have to set the DPI within photoshop to 300 from the default 180? I’m not sure what effect this would actually have, since 1200x1800 is 1200x1800 regardless. Is it only relevant to printing and not to the actual image editing?

I guess I’m just not quite sure what my post processing order should be, to get prints that result in exactly what I’m looking at, without having snapfish automatically rescale them and introduce artifacts.

Don’t worry about Photoshop’s “180dpi, 22"x16"” - as you say pixels are pixels.

3000x4000 and 4"x6" are different aspect ratios, with the former being more “square”, so you’ll need to crop to the correct AR - say 2666x4000 (taking some off the top and/or bottom) if you want to keep the full image width.

You could resize your final image to 1200*1800, or just send it to them full size and let their equipment sort it out - I doubt you’ll see much a difference one way or the other (though the former would make for faster uploads, and if you upload the larger file it gives you more options for cropping and/or making larger prints later) - in either case you’re right that “output sharpening” should be your final step.

Hmm… seconding that at 4x6, 300dpi, JPEG-compressed… I doubt you’ll see much of a difference either way. Unless you’re an absolute perfectionist, you might just want to skip the Photoshopping altogether and either upload to Snapfish directly or use a simpler program like Picasa.

After some readig, it turns out that the DPI number on images is useless and fictional. When you take a picture, your camera sticks a random DPI in that exif entry that’s pretty much arbitrary. Canon uses 180 (hence my 180), but it could use 1 or 50000 and it wouldn’t matter. Why they even have this, I have no idea.

Similarly photoshop could have 1 DPI or 5 million DPI and it wouldn’t matter.

So those numbers essentially just exist to confuse people, they have no meaning in a digital sense.

Printing does have a DPI, but this has to do with the pixels of the image rather than any digital DPI setting. You need at minimum enough pixels to cover the DPI x physical size in inches of the image to give the printer enough information to have maximum detail. So for a 4x6 image on a 300 dpi printer, that’s 1200x1800 pixels. If you give them more, they’ll just scale it down.

So that’s all right?

But that scaling down process could potentially introduce artifacts, which was my concern. Probably an overblown concern because it won’t matter terribly much, but it turns out sharpening is a rather complex issue and I’d rather get started off on the right foot.

After a decade of printing and scaling, I think most places have eliminated the more blatant artifact/moire issues.

You can get passable pictures at 100dpi depending on the subject. Faces tend to have large flat relatively featureless areas of smooth tone change, so a lower DPI will not be as noticeable with portraits; if you have a landscape or cityscape, or really want the weave detail of the clothing in a protrat, then higher DPI will be better.

OTOH, most photos are viewed at half an arms length to arms length, unless they are poster size. Posters, we back off to see the whole picture so it might as well have the same number of pixels as an 8x10 at arms length.

I have done pretty good pictures from a 2.4Mp original at 8x10 and 11x14; as you note, today’s cameras have far more resolution than necessary, but unless it’s a more expensive one, either the lens quality or light conditions, or the process of taking the picture, may compromise the resolution.

In fact the way modern CCD sensors work, every pixel is more than half estimate and interpolation, as each pixel is only one colour sensor (R, G, or B) and the other colours for that pixel are filled in from the neighbour values that use those colour filters.

Short answer - the processor will not mess up your pictures, how much resolution you really need depends on the subject, unless transmission size for uploading is an issue, why resize or anything? They can do that.

If you took a picture of a high-contrast fine grid or stripes, do a test print first to avoid moire. However, you are as likely to get moire patterns from the camera.