Best way to reduce image dimensions in PS?

Snapper

Greetings from New Zealand!

Now that we routinely have 10MP images, I'm wondering what the consenus is on the best way of reducing image dimensions in PhotoShop for everyday storage and viewing in albums like iView etc. (While keeping the originals elsewhere, of course.)

Seems like "Bicubic Sharper" might be the best algorithm, but what about the relationship of the original dimensions to the finished dimensions? Would one get a better result by always reducing by some whole-number factor?

For example would there be better/different pixel interpolation by reducing my D80's 3676 x 2451 images by a factor of exactly 2 to 1838 x 1225.5 (??), rather than "rounding them off" to 1800 x 1200?

--Ian

gfxartist

You are correct in that Bicubic Sharper is better for downsampling. However, be careful that you don't get an over-sharpened image. If that occurs, you may just want to use the standard Bicubic.

If the images you are uploading are just for viewing on screen and not for ordering prints, I'm not sure it really matters what you resize your image to. I doubt it will be a noticable difference on-screen in a photo album.

I did find this interesting write-up: http://www.xs4all.nl/~bvdwolf/main/foto/down_sample/down_sample.htm. I have to confess my eyes started glazing over while reading it.

Snapper

gfxartist wrote:

You are correct in that Bicubic Sharper is better for downsampling. However, be careful that you don't get an over-sharpened image. If that occurs, you may just want to use the standard Bicubic.

If the images you are uploading are just for viewing on screen and not for ordering prints, I'm not sure it really matters what you resize your image to. I doubt it will be a noticable difference on-screen in a photo album.

I did find this interesting write-up: http://www.xs4all.nl/~bvdwolf/main/foto/down_sample/down_sample.htm. I have to confess my eyes started glazing over while reading it.

Yes, that certainly is very dry reading! And I'm not sure what the conclusion is!

I guess I was just kinda wondering, seeing as how we have to throw pixels away when downsizing, whether there's any advantage in throwing out (say) every second one (reducing by exactly 50%), rather than reducing by 54%, or 62%, or some other "random" number. I agree the differences are most likely not noticeable for snap-shot album viewing, but there may be comfort in knowing!

Stu Engelman

Hello Again Snapper!!

Although the Adobe online Help system suggests using Bicubic Sharper for downsampling, I think you'll find Bicubic Smoother works better here (Adobe may have made a mistake in saying the reverse). The basic idea is that when upsampling, pixels have to be extrapolated, resulting in loss of contrast at color edges, creating the need for sharpening. The reverse applies when downsampling. After downsampling with Bicubic Smoother, you can add/reduce sharpness with Smart Sharpen or Gausiian Blur, respectively.

Use of round numbers for the reduced size image dimensions has no impact (good or bad) on quality. While it's true that an image already inside Photoshop will look best if viewed at "rounded" zooms like 25%, 50%, and 100% (due to the nature of Photoshop's display engine), the physical output quality of a reduced size image is independent of this consideration.

Stu

Snapper

Stu Engelman wrote:

Hello Again Snapper!!

Although the Adobe online Help system suggests using Bicubic Sharper for downsampling, I think you'll find Bicubic Smoother works better here (Adobe may have made a mistake in saying the reverse). The basic idea is that when upsampling, pixels have to be extrapolated, resulting in loss of contrast at color edges, creating the need for sharpening. The reverse applies when downsampling. After downsampling with Bicubic Smoother, you can add/reduce sharpness with Smart Sharpen or Gausiian Blur, respectively.

Use of round numbers for the reduced size image dimensions has no impact (good or bad) on quality. While it's true that an image already inside Photoshop will look best if viewed at "rounded" zooms like 25%, 50%, and 100% (due to the nature of Photoshop's display engine), the physical output quality of a reduced size image is independent of this consideration.

Stu

Thanks Stu. That's what I wanted to know. Regarding the Smoother/Sharper thing, your reasoning makes sense and I'll play around with the options. --Ian