On Jun 15, 2006, at 4:29 PM, Don Sanderson wrote: > Godfrey that's a great explanation, I understood most of it! > Could you simply define "spatial resolution" for me?
Thank you for the compliment, Don. While the subject matter is rich in details and can be quite complex to understand in implementation, I believe that the basic concepts are not all that difficult when explained clearly. Much less so for me than understanding the mechanics of chemistry ... I'm a Mathematician, not a Chemist. ;-) :-) ... Spatial resolution in this context is the [x,y] pixel coordinate position of the photosites relative to the subject target that they recorded. Larger shifts in relative pixel position usually only occur with resampling (aka: resizing of the pixel grid for different resolutions) because you're interpolating discrete pixel positions in the [x,y] grid to new positions [x',y'] through some typically Real- valued function, which causes round off error. JPEG compression, when set to high compression/low quality, can create artifacts due to the way the algorithm works on [NxN] blocks of pixels. These artifacts influence spatial resolution not so much through intentionally moving pixels relative to one another but by changing the tonal values to where the original pixel data altered spatially. Chroma interpolation can affect spatial resolution in a similar way but by a different mechanism. (As an aside, This is one reason the Fovean folks make such a big deal of how each photosite in a Fovean chip captures all three colors ... they claimed gains in spatial resolution for the same [x,y] grid. However, in practical terms, the spatial resolution degradation caused by chroma interpolation is pretty minimal. What the Fovean sensor theoretically provides is more accurate RGB tonal capture, but this has not worked out quite as well in practical terms as the hype would lead you to believe.) Godfrey -- PDML Pentax-Discuss Mail List [email protected] http://pdml.net/mailman/listinfo/pdml_pdml.net
