3
\$\begingroup\$

I felt that my 8x10 enlargements were not as sharp as I would have hoped. My original images are super sharp and I asked the printer whether it was because of the finish that I had picked or something else and they said this:

"I'm sorry to hear your prints came back soft looking. I think what might have happened was some pixel compression. Your files are a bit larger than the size we printed these test prints at, so compressing that larger file can cause prints to come back looking soft or blurry. Your files are about 13.3x9.9 at 350 ppi, while the prints were 8x10 at 300 ppi. I would suggest cropping your images to as close to the size you're printing as possible. So if you were printing a bunch of 4x6s, I would suggest cropping your files to 4x6 at 300 ppi."

Is that a correct explanation? I would think that whatever I do do adjust the size and resolution would be the same as they did. But I could be wrong

\$\endgroup\$
1
  • \$\begingroup\$ Thank you everyone for the very insightful and detailed answers. I had no idea that there was this much to it. \$\endgroup\$
    – pitosalas
    Commented Jul 26, 2022 at 23:54

4 Answers 4

8
\$\begingroup\$

Not really. But there is a grain of truth in it.

What they are saying is that your image had excess resolution and they had to downsample it by about 30%. This is not an extraordinary amount; all but the most professional labs resample images, often much more than that. (Professional labs sometimes require you to supply the images with exact dimensions/resolution, but this would be too onerous a demand for most people).

Yes, any resampling (up and down) loses a bit of sharpness in most circumstances. There are images that may visibly suffer from it (notably line art and drawings). But for typical photos 30% downsampling should not be too noticeable. If you had enough sharpness (not to be confused with resolution!) to start with, the resulting downsampled image should still look sharp.

Nevertheless, downsampling is destructive, and there are methods to compensate for it. Generally, this means applying some form of sharpening after downsampling (even if the initial image was perfectly sharp). This can be 'baked in' to the resampling algorithm, or be a standard unsharp mask. Also, when resampling by a large amount, it is usually better to do in several smaller steps. But, like with any post-processing, overdoing it may ruin the image.

For this reason, most labs don't bother with any smarts and just apply standard resampling, at best with some built-in sharpening parameters. So, you are better off doing it yourself. Not unlike preparing the image for a pro lab.

Now, I will assume your image is indeed "super sharp", such that if you zoom it to 100% it still looks reasonably sharp. (In practice, most bad prints are 'user error' and simply come from blurry images).

In most professional workflows there is a step usually called 'output sharpening'. Unlike the normal processing sharpening, its parameters primarily depend on the physical dimensions of the image and the medium, rather than on the image itself.

Even if your photo is already the right resolution (number of pixels for the print), it will still benefit from output sharpening. Without processing, prints will invariably look softer than on-screen image: they (usually) have less contrast, their 'pixels' are not as sharp, the surface is more diffuse, etc. To have good subjective sharpness on, say, matt paper, you may need to sharpen the image considerably, and it will look painfully oversharpened on screen (especially if you zoom in).

So, your final processing steps should be like this.

  • Crop the image to have the right proportions. (Note that 8x10 is more 'squarish' than the raw image from most cameras). How you crop depends, of course, on the subject and your artistic needs.

  • Do your regular adjustments, if necessary (colour, contrast, etc.)

  • Save the image as a high-quality jpeg or even lossless, if you want to keep it. What we are going to do next applies only for this print, will be saved separately and can be deleted after printing.

  • Resample the image to have the right physical resolution. This depends on what the lab advised. If they said 300 ppi, resample to 300 ppi, whatever the number of pixels this results in. Most decent photo editors let you specify this number (and the physical dimensions) directly when resampling, but more primitive tools designed for on-screen viewing may only work with pixels, in which case you'll need to calculate them yourself. (So, for 10" print at 300 dpi you'll need 3000 pixels along this edge).

    Note: if your software has an option to do a sharpening resampling (like 'Bicubic sharper'), use it. Otherwise, you might apply a very moderate sharpening with radius about 1 and amount 20-30%. This is only to compensate 'mathematical' losses of resampling. But given that we are about to apply a more aggressive output sharpening, this is not necessary.

  • Now the more difficult part. Zoom the image out to have approximately the same physical size on your screen as the intended print, maybe a bit larger. If possible, select an 'even' pixel scale like 25 or 50%. Yes, the pixel resolution of a computer display, even a 'retina' one, is usually lower than that of a print, and yes, we are undergoing an extra (low quality) resampling.1 But it's still a better representation of the effect than the pixel-peeping 100% that you use when doing your regular image sharpening.

  • Use a regular sharpening tool such as unsharp mask. Here, you need to choose the parameters that give some visible effect to the image as it's displayed. These parameters will be unlike those for regular sharpening. Most importantly, the 'Radius' will be larger. In normal image sharpening, you usually have it around 1. But now we need it to be of a physical, visible size. Say, for 0.5 mm you'll have about 5-6 pixels. The 'Amount' may be lower than usual, 30-50%, but may be higher: it depends on the image. You should just see the effect on your screen. (It is convenient to be able to toggle preview of the effect, if your editor allows it). The exact parameters will be gained with experience. You might also search specific guides for output sharpening: I'm sure there must be quite a few out there.

  • Save the sharpened image to a separate file and send it to the lab for printing.

  • If you are not happy with the result, don't re-edit the lab file. Start off the file you saved in the middle step, resample it again and try different sharpening.

There are specialised tools (usually plug-ins) designed specifically for output sharpening (and generally processing). There you might enter your physical output properties such as size, printer type and paper type, and they will apply optimised parameters. But for one-off print this may be an overkill, and it's useful to try sharpening yourself for better understanding.


1 Some tools (primarily the specialised image viewers) may apply on-the-fly sharpening for the preview images. This is also a form of output sharpening. But in this case it will interfere with your editing, and if there such an option in your software, it needs to be disabled.

\$\endgroup\$
2
\$\begingroup\$

Yes, it is reasonable because it is a polite way of saying that you don’t know what you are doing.

There’s nothing wrong with that — not knowing what you are doing, I mean.

We are all still have a lot to learn because what we want to do is complex.

A file for an 8x10 print at 300dpi should be 2400 pixels (8x300 pixels) by 3000 pixels (10x300 pixels).

Except that’s only correct if the paper size is larger than 8x10 so that the image is neither cropped nor printed full bleed.

If the paper is 8x10 then the image file needs to be reduced to provide a border, or made larger, say 8.5 x 10.5 (2550 x 3150 pixels) to be printed full bleed.

Full bleed is always cropped either by printing on larger paper and cropping with a paper cutter or by using a larger image and printing beyond the edges of the paper.

The reason is that paper, unlike a screen, cannot be pixel perfect. Paper changes dimensions with changes in temperature and humidity. And it is not manufactured to 1/300 of an inch tolerances, anyway because it is made of wood.

Circling back, the way to get good prints is to work with your printer.

And to practice in order to gain experience.

If you don’t like the results of a print you have to print again (and often again and again). Printing is an art.

\$\endgroup\$
1
\$\begingroup\$

Assuming your images were as described, 13.3 x 9.9" at 350 ppi would make them 4665 x 3465 pixels.

What device were you viewing the photos on?

If it was a computer screen, its physical size is likely to be similar or larger than the 8 x 10" print, but the pixel dimensions smaller. So there isn't a 1:1 relationship between the photo pixels and the screen pixels. One pixel on the screen is an average of several pixels that you see on the print, so is likely to look sharper.

If it was a phone, the pixel dimensions will be similar to the image (of course: it was the camera too). But the phone's screen is only a fraction of the size of the physical size of the print, so it will look sharper to the eye.

So what they say is likely to be true, but not for the reasons given. Scaling 350 ppi to 300 ppi is unlikely to make a significant difference to what the eye can detect at 300 pixels per inch. Much more significant is the size and scaling of the device that you were viewing your nice sharp image on.

\$\endgroup\$
3
  • \$\begingroup\$ It was a large computer screen. So if I use my app to resize (and resample) to 8x10 inches with 350 ppi I would expect the print to look a little sharper on paper? \$\endgroup\$
    – pitosalas
    Commented Jul 21, 2022 at 19:34
  • \$\begingroup\$ I don't suppose you resizing the image will be any different to them resizing the image, but would presume that a professional printer will know the most appropriate algorithm to use (there are several options in image software). If the photo, when printed at a scale much closer to 1:1, looks less than sharp, it can only be recovered by imposing the illusion of a sharpening algorithm (which exaggerates local differences between pixels). \$\endgroup\$ Commented Jul 21, 2022 at 19:38
  • \$\begingroup\$ What does the detail of your image look like on the screen when you view it at 1:1 instead of 'fit to screen'? Is it as sharp as you supposed? \$\endgroup\$ Commented Jul 21, 2022 at 19:52
1
\$\begingroup\$

All of the answers are giving good and important information. Let me come at it from a different angle.

A properly sharpened image, when the junctions of light and dark ares viewed at a high magnification, will show a slightly darker line(s) of pixels on the dark side of the junction and a slightly lighter line(s) of pixels on the light side. These lines are the result of the sharpening step and specifically emphasize the junction between dark and lighter shades and make the image seem 'sharper'.

An overly sharpened image will have visible light and dark 'halos' that indicate oversharpening.

If you resize the image, usually making it smaller in pixel dimensions, it is very possible that some of these light and dark lines get lost in the resizing thus before printing it is important to:

1)Crop to the desired aspect ratio and convert to the color profile the printer will accept.

2)Resize to the final pixel dimensions. (length in inches times 300) thus an 8 x 10 would be 2400 x 3000.

3)Resharpen so that the file is almost 'crunchy' or at the edge of being over-sharpened. This will give you the best shot at a sharp looking final print.

In general, learn how to sharpen, both globally and selectively.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.