How to express level of detail within a given resolution?
Posted: 24 Jun 2013 23:44
Hi!
Have you noticed that phenomenon on YouTube that there are 720p videos that look crappy, while other 480p videos still look great? I think people (or the industry) are too single-mindedly focused on resolution, when it comes to photos and videos. As for cameras, the sensor plays an important part in how much detail there is per pixel. Crappy cameras can deliver high resolution images which are all blurry and washed out. As for videos, there are 720p videos that are too much compressed and often times you would rather have 480p with little compression. I'm particularly fascinated with pixels that carry the maximum of details, such as a high-quality image that was down-scaled with a powerful algorithm. And last but not least there is usually a clear difference between video frames and photos/screenshots. I noticed this with a 720p Let's Play. When I stopped the video at any point, the 720p frame had pretty little detail in comparison to the overall resolution (this was worse for movements than for slower passages). I made a test: how small could I size down and size back the image without noticing a difference? I ended up with 360p being almost of the same detail, that means one fourth in size! In any case, the detail of video frames was always considerably lower than that of screenshots of the same resolution - which keep the original detail to nearly 100 percent. So to sum this up: I've always wondered whether there are ways to measure, and express in technical terms the level of detail expressed in an image, compared to a source image. (Without that source comparison, a blurry image of a chaotic scene would have more detail than a detailed image of a smooth surface.)
Have you noticed that phenomenon on YouTube that there are 720p videos that look crappy, while other 480p videos still look great? I think people (or the industry) are too single-mindedly focused on resolution, when it comes to photos and videos. As for cameras, the sensor plays an important part in how much detail there is per pixel. Crappy cameras can deliver high resolution images which are all blurry and washed out. As for videos, there are 720p videos that are too much compressed and often times you would rather have 480p with little compression. I'm particularly fascinated with pixels that carry the maximum of details, such as a high-quality image that was down-scaled with a powerful algorithm. And last but not least there is usually a clear difference between video frames and photos/screenshots. I noticed this with a 720p Let's Play. When I stopped the video at any point, the 720p frame had pretty little detail in comparison to the overall resolution (this was worse for movements than for slower passages). I made a test: how small could I size down and size back the image without noticing a difference? I ended up with 360p being almost of the same detail, that means one fourth in size! In any case, the detail of video frames was always considerably lower than that of screenshots of the same resolution - which keep the original detail to nearly 100 percent. So to sum this up: I've always wondered whether there are ways to measure, and express in technical terms the level of detail expressed in an image, compared to a source image. (Without that source comparison, a blurry image of a chaotic scene would have more detail than a detailed image of a smooth surface.)