08-30-2010, 06:34 AM
[quote name='wim' timestamp='1283118956' post='2319']
Sorry, but you're wrong. I did have to use "bicubic smoother", and I still got sharpening artefacts nonetheless. Do note that normally you'd have to use "bicubic sharper" for downsampling, effectively as you state, and as I do with all of my other lenses, although not for every step, because with f.e. 24L II and 135L I also get into sharpening artefacts trouble occasionally that way. I also use steps of approximately 1.4X, 1.5X max (down obviously in that case, so 1/1.4, or 0.7, just in case I get another undeserved reprimand), so that the algorithm has to make a choice when downsampling (or upsampling for that matter, same thing, step wise), rather than just divide by 2 in the case of downsampling, or multiply by 2 in th ecase of uprezzing.
In short, I used "bicubic smoother", which is normally used for uprezzing, instead of "bicubic sharper", and still had to desharpen a bit.
BTW, there are nicer ways to say things than using the words you did. I'll be the first one to admit a mistake, so no need even to raise a flag, just ask.
[/quote]
Sorry if I offended you, that was not the intent. My opening remark was entirely tongue in cheek.
However, on the subject matter, I've never seen sharpening artefacts when downsampling, as downsampling inherently reduces sharpness (well, micro contrast, or whatever you want to call it), not increases it, and thus softens the image overall. So, unless you have sharpening artefacts in your original files already, it's hard to imagine how they could come about when downsampling. If on the other hand what you're seeing are sampling artefacts (which I haven't really seen with my files when using decent algorithms such as bicubic, either), as opposed to sharpening artefacts, I don't think you'd want to attribute them to the quality of the lens used (in fact, it would really bother me if I had to soften a downsampled image even further, leading me to question my downsampling routine...).
Sorry, but you're wrong. I did have to use "bicubic smoother", and I still got sharpening artefacts nonetheless. Do note that normally you'd have to use "bicubic sharper" for downsampling, effectively as you state, and as I do with all of my other lenses, although not for every step, because with f.e. 24L II and 135L I also get into sharpening artefacts trouble occasionally that way. I also use steps of approximately 1.4X, 1.5X max (down obviously in that case, so 1/1.4, or 0.7, just in case I get another undeserved reprimand), so that the algorithm has to make a choice when downsampling (or upsampling for that matter, same thing, step wise), rather than just divide by 2 in the case of downsampling, or multiply by 2 in th ecase of uprezzing.
In short, I used "bicubic smoother", which is normally used for uprezzing, instead of "bicubic sharper", and still had to desharpen a bit.
BTW, there are nicer ways to say things than using the words you did. I'll be the first one to admit a mistake, so no need even to raise a flag, just ask.
[/quote]
Sorry if I offended you, that was not the intent. My opening remark was entirely tongue in cheek.
However, on the subject matter, I've never seen sharpening artefacts when downsampling, as downsampling inherently reduces sharpness (well, micro contrast, or whatever you want to call it), not increases it, and thus softens the image overall. So, unless you have sharpening artefacts in your original files already, it's hard to imagine how they could come about when downsampling. If on the other hand what you're seeing are sampling artefacts (which I haven't really seen with my files when using decent algorithms such as bicubic, either), as opposed to sharpening artefacts, I don't think you'd want to attribute them to the quality of the lens used (in fact, it would really bother me if I had to soften a downsampled image even further, leading me to question my downsampling routine...).