zeitgeist Posted May 25, 2012 Report Share Posted May 25, 2012 Whats a bi-cubic image? I'm assuming its sizing related, other than that I'm clueless. Cheers Z Link to comment Share on other sites More sharing options...
Coss Posted May 25, 2012 Report Share Posted May 25, 2012 Whats a bi-cubic image? I'm assuming its sizing related, other than that I'm clueless. Cheers Z Mr Wiki Link to comment Share on other sites More sharing options...
zeitgeist Posted May 25, 2012 Author Report Share Posted May 25, 2012 Mr Wiki That explains enlarging, thx. How does 'bicubic' relate to making an image smaller and not having the squashed effect that I sometimes get? Is it just due to resizing and the measurements no longer being linear? I'm not 100% sure I'm understanding this Link to comment Share on other sites More sharing options...
Coss Posted May 26, 2012 Report Share Posted May 26, 2012 The way I understand it is that when an image is compressed, or, downsized from a high resolution to a lower resolution, i.e. from 300 dpi to 72 dpi, the way it's done sometimes is to interpolate. I think interpolate in this sense means if there are a large amount of pixels that are almost the same in color, brightness, etc, then rather than describe each one, in the file, it is acceptable to describe a bunch of them as being the same. So if there are 1,000 pixels that are all nearly the same shade of grey, instead of doing something like this: 0001=grey10.01%, 0002=grey10.02%, 0003=grey10.32% and so on for the thousand, you can just say something like 1,000 pixels=10.22%. This goes for all the various parameters that describe the pixel. Now from Mr Wiki - "In image processing, bicubic interpolation is often chosen over bilinear interpolation or nearest neighbor in image resampling, when speed is not an issue. In contrast to bilinear interpolation, which only takes 4 pixels (2x2) into account, bicubic interpolation also considers the 16 pixels around it (for a total of 4x4 pixels) while computing an average. Images resampled with bicubic interpolation are smoother and have fewer interpolation artifacts." I take all this to mean, when bicubic does it's stuff, it evaluates 16 pixels around the one it's describing, instead of 4 for bilinear. This makes it more accurate and thereto better. When you get 2 black ones next to 2 white ones in a 4 pixel sample, you might get 50% grey for example. But if 16 were looked at you might find that there's still only 2 black ones and 14 white ones which would result in a paler shade of grey and this would look more like the original. The human eye/brain compensates this way all the time. Just look at a newspaper photograph with a magnifying glass, it's all just a mix of dots combining to give the appearance of solid colors and shapes. I think I've got that right, but then, as I get older, I finding lots of things that aren't true anymore. Like, you can trust bankers, or borrowing money is a good thing. But I rant. Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.