>> Bill Hilton <bhilton665@aol.com> wrote:
>>
>> Many people have come to the same conclusion. I had GF 2.0 (bundled
>> with a scanner) and tested it on several images and felt I could get
>> better results with Photoshop too, at least at the magnifications I used.
>Bill Tuthill wrote:
>
> That's interesting, because I have compared the Lanczos filter in Irfanview
> and ImageMagick with Photoshop bicubic and bicubic smoother. In all cases
> where upsampling aspect ratio remained 1:1, Lanczos produce sharper/smoother
> edges and better detail.
I haven't tried this filter but the examples I saw didn't seem to look
any better than what you can get with Photoshop, for example these two
pages using older versions of Photoshop ...
http://www.fredmiranda.com/SI/
http://www.americaswonderlands.com/digital_photo_interpolation.htm
Roger Clark has a page on upsamping using "Adaptive Richardson-Lucy
Iteration" and feels that works the best ... you can download the fox
image snippet and run your software and see if you get something as
good or better ... when I did this using just Photoshop and my normal
three-step workflow (which includes two sharpening passes) that I use
for printing Roger thought it looked about as good what he gets with
R-L, except instead of taking 90 minutes to run I could do it in 3
minutes ... at any rate if you can beat his example Roger would like to
see it. This is useful info because it's a standardized image.
http://www.clarkvision.com/imagedetail/image-restoration1/index.html
Note that different methods might give better or worse results
depending on the image (and the operator if sharpening is included).
For example here's a test (by someone selling GF, just as the guy
selling a Stair Interpolation action came up with an image that looks
good upsized using SI) that shows GF is better than Photoshop, at least
for this image ... http://www.inkjetart.com/news/gf/ ... but it's
irrelevant to me because they rasterized a vector image and then ran
the tests and none of my images are vector-like images. Few photos
are.
Bill
AAvK - 23 Sep 2006 11:35 GMT
> I haven't tried this filter but the examples I saw didn't seem to look
> any better than what you can get with Photoshop, for example these two
[quoted text clipped - 25 lines]
>
> Bill
That last link is very old by a few years. GF is now owned by OnOne
software and it's version 4, no saving stn files anymore. And it goes
up to 800%.
But thanks for the Clarkvision link, don't want to spend more money
these days. Currently I have GF 2.5 and have happily used it since I
bought it, with perfect results.
Because of this thread I tested CS1's bicubic work for both sharper
and smoother against GF 2.5 as follows:
800 x 600 x 72ppi to 4800 x 3600 x 300ppi (600% up)
I had never tested this before since PS 5, and was quite pleasently
surprized! PS's BC smoother is perfect. Should I upgrade GF for
$69? Maybe not!

Signature
}<)))*> Giant_Alex
cravdraa_at-yahoo_dot-com
not my site: http://www.e-sword.net/
Bill Tuthill - 23 Sep 2006 20:27 GMT
>> ... I have compared the Lanczos filter in Irfanview and ImageMagick
>> with Photoshop bicubic and bicubic smoother. In all cases where
[quoted text clipped - 6 lines]
> http://www.fredmiranda.com/SI/
> http://www.americaswonderlands.com/digital_photo_interpolation.htm
Fred Miranda has something to sell (stairstep) and the other page
is good but the images aren't similar to what I shoot.
I'm not saying Lanczos is the best upsampling filter-- Qimage and
SARS are better-- but Lanczos might be the best downsampling filter.
Here are samples showing its improvement over PS Bicubic Sharper.
http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00E7py
The reason I mention Lanczos is that it's free, which is somthing
you can't say for Photoshop Bicubic Smoother. And IMO (although
opinions are subjective) Lanczos is enough of an improvement over PsBS
to be worth downloading and using despite its non-plugin status.
> Roger Clark has a page on upsamping using "Adaptive Richardson-Lucy
> Iteration" and feels that works the best ... you can download the fox
[quoted text clipped - 6 lines]
> see it. This is useful info because it's a standardized image.
> http://www.clarkvision.com/imagedetail/image-restoration1/index.html
Thanks for the URL, which I had not seen before. I'll study it.
> Note that different methods might give better or worse results
> depending on the image (and the operator if sharpening is included).
Yes, certainly true.