Note that, to the best of my recollection, the lighthouse original is quite strongly sharpened.In an ideal toolchain, you don't sharpen before downsampling.P.S. If this is what you used to evaluate EWA LanczosSharpest 4, this pre-sharpening may have skewed the evaluation.
Bart:With the lighthouse, you use gamma 2 blend, or straight through linear RGB?P.S. Irregardless, this is a score 1 for EWA LanczosRadius.
Ok, thank you.
If you add final sharpening, I would be very tempted to use the now "standard" EWA LanczosSharp, overcoming the softness with post-sharpening.It's amazingly good at anti-aliasing.
I'll give it a try. Initial results look interesting, slightly different from LanczosRadius, both are nice. I'll try it with my 'Rings' image abd see what happens.Cheers,Bart
If I read http://www.imagemagick.org/discourse-server/viewtopic.php?f=1&t=25945&p=113584#p113553 correctly, Adobe RGB is a "straight" gamma space, which means that going to some linear RGB is as easy as applying gamma. (See Section 4.3.1.2 of the linked tech report.) Then, the Adobe RGB to linear and linear to gamma 2 (say) conversions can be merged into one single gamma, which can be undone at once. This should improve the acccuracy in a 16 bit toolchain. The "merged gamma" should be2./2.19921875
It makes sense to me to use 2.19921875 as "standard gamma" in which case we already have it when using Adobe RGB, and we only need to undo it to move things through linear RGB.In the case of sRGB, the toolchain remains as before, except for the value of gamma.P.S. I really don't like the "linear segment" of sRGB: Fine for compressing, not fine for manipulating the image.
P.S. I really don't like the "linear segment" of sRGB: Fine for compressing, not fine for manipulating the image.
...P.S. The verdict is in, LanczosSharp looks slightly cleaner and has an overall higher resolution (see logarithmic FFT attachment, amplitude is higher for LanczosSharp, and slightly less ragged). It really required this FFT analysis to spot the differences, so we are fitting a few more angels on a pinhead.
Would you be kind enough to add EWA Lanczos's plot?
P.S. Note: The rings image is very smooth. I believe that with sharp natural images, the difference between, say, LanczosRadius and LanczosSharp (and Lanczos) are more pronounced.
...No worries. The original Lighthouse PhotoCD file (IMG0019.PCD) is not (over)sharpened (it's basically a filmscan) at its highest packed resolution (16BASE = 3072x2048 pixels)....
...To show a bit more of what it's about, I've attached a sample of the down-sampling quality as it is right now. I've used the well known Lighthouse image from the PhotoCD sampler CD, and reduced the full resolution to 25%, with sharpening (make sure you watch it at real 100% zoom resolution, by downloading it). Aliasing is very well controlled, fully automatic...
It looks really good, although it has too much haloing, dark esp., for my taste.However, I dislike haloing more than many.
Indeed, however, if one allows the ImageMagick installer to do it (see third attachment), that is already taken care of as it prefixes the new Path to Imagemagick to the existing startup file. The drawback of using that option is that new installations will keep prefixing new paths to the PATH environment variable, so one needs to occasionally edit the older locations out (always tricky, so don't make mistakes).
To recap the (current) method choices:For general resizing, but especially upsampling: a blend between a gamma 1 and a gamma 1+2 resampling with the LanczosDetail filter.For down-sampling without sharpening: a gamma 1 resampling with the Lanczos4Sharper filter recipe.For down-sampling with sharpening: a gamma 1 resampling with the LanczosSharp filter.
(...)"hit me over the head with a baseball bat but hold the second halo"(...)
