Which image is better?

owlice · Post by **owlice** » Thu Sep 20, 2012 9:34 am

I often have to reduce the resolution and size of images. There are different algorithms for scaling images; star trail images usually reduce well with one called "Smooth." But this time, I'm not so sure. Which of these scaled images is better? By "better," I mean which, of the two, do you like the best, would you choose to post on the submissions thread to best show off the photographer's work?

Copyright: Tyler Brower

Algorithm A:
[attachment=0]NorthstarQuartz_Brower.jpg[/attachment]

Algorithm B:
[attachment=1]Northstar.jpg[/attachment]

Beyond · Post by **Beyond** » Thu Sep 20, 2012 12:01 pm

Well, for me the picture on the left has a little more clarity on the left-hand side, than the picture on the right. Especially in the magnified view.

Chris Peterson · Post by **Chris Peterson** » Thu Sep 20, 2012 1:44 pm

owlice wrote:I often have to reduce the resolution and size of images. There are different algorithms for scaling images; star trail images usually reduce well with one called "Smooth." But this time, I'm not so sure. Which of these scaled images is better? By "better," I mean which, of the two, do you like the best, would you choose to post on the submissions thread to best show off the photographer's work?

I think the left image is significantly better, both in reduced and full views.

I find that the best reduction strategy is simple: always use basic bicubic resampling- no sharpening or smoothing in the reduction algorithm at all, then apply either an unsharp mask or Gaussian blur afterward, depending on your results (blurring is sometimes necessary when you have star trails, to minimize aliasing). This approach is easy, and gives the most control.

owlice · Post by **owlice** » Thu Sep 20, 2012 6:17 pm

Thanks for the input; much appreciated!

Chris, I barely have time to reduce and post; I definitely don't have time to apply a mask/blur, so I have to depend on the scaling algorithm to get as close as possible. That, I sometimes have time to select!

Rothkko · Post by **Rothkko** » Thu Sep 20, 2012 6:44 pm

A: In the area with less trace of stars is more apparent quality.

Moonlady · Post by **Moonlady** » Fri Sep 21, 2012 3:49 pm

I can't see any differents, maybe because I am myopic ?

geckzilla · Post by **geckzilla** » Fri Sep 21, 2012 4:15 pm

Click both of the images and view them side by side at full size, Moonlady.

Chris Peterson · Post by **Chris Peterson** » Fri Sep 21, 2012 6:06 pm

owlice wrote:Thanks for the input; much appreciated!

Chris, I barely have time to reduce and post; I definitely don't have time to apply a mask/blur, so I have to depend on the scaling algorithm to get as close as possible. That, I sometimes have time to select!

Well, what's going to work best is going to vary with the image. If you don't have the time for more critical processing, I'd probably error in the direction of sharpening rather than blurring.

Can't you just have rules for submission and simply ignore images that don't meet those rules? Seems like the image creator is the best one to figure out how an image should be reduced.

owlice · Post by **owlice** » Fri Sep 21, 2012 8:10 pm

Thanks, Chris! Some scaling algorithms are very hard on star trail images when reducing them, for the reason you identified earlier: aliasing. I usually stick to one algorithm as I've found it handles almost everything well; star trail images are usually an exception. Not in this case, however!

Many imagers are now posting their own images (which is great!) so reduce their own images when posting them. Still, there are plenty of imagers who don't post -- sometimes because they just don't know about Asterisk -- so I try to fill in the gaps as best I can.

Starship Asterisk*