Skip to main content
5 events
when toggle format what by license comment
Apr 15, 2011 at 21:26 comment added mgkrebbs I did the exercise. Here's the 1-scale version and here's the 3x scaled version. You can see loss of detail in the fine lines of the lower middle. With zoom, you can see similar in many places. This was a cubic interpolation via Gimp, and used PNG files to avoid lossy compression.
Apr 15, 2011 at 20:30 comment added mgkrebbs @Jojo: I don't have a theoretical proof to offer (and a proof would depend on which of several interpolation algorithms was used in the scaling). I think you could practically show it: take a nontrivial image say 400x400, scale it to 380px, that to 360, & that to 340. Compare it to one scaled from the original to 340px. I haven't tried this, but I think there will be some greater blurring, although it may not be all that visually perceivable. Note the scaling suggested is not using nice ratios.
Apr 15, 2011 at 19:36 comment added horatio I just took a screencap of my multimonitor setup and then used photoshop to scale one version to 25% and one 50 then 50. The two results were pixel identical. I was surprised
Apr 15, 2011 at 18:56 comment added JoJo You say incremental scaling is worse, but is there a theoretical proof?
Apr 15, 2011 at 7:09 history answered mgkrebbs CC BY-SA 3.0