Prev: Is there a model identfier for SD memory?
Next: RAW vs. JPG: Fidelity vs. Convenience
From: UC on 23 Dec 2009 15:32
anyone here own it?

I have a smallish file that I have to use large in a calendar. It has
obvious pixellation. Is anyone willing to help me process this into a
larger size?

Thanks
From: David Ruether on 23 Dec 2009 16:04

"UC" <uraniumcommittee(a)yahoo.com> wrote in message
news:f5a4293d-375c-4059-b6a1-9d5bbfa753ca(a)22g2000yqr.googlegroups.com...

> anyone here own it?
>
> I have a smallish file that I have to use large in a calendar. It has
> obvious pixellation. Is anyone willing to help me process this into a
> larger size?
>
> Thanks

Ah, that brings back memories - I used to enjoy playing with it.
If you had the original program and knew the parameters you
had fed into it, it may be possible to generate a new similar image
with a higher resolution - but, sorry, I can't help...
--DR


From: NameHere on 23 Dec 2009 18:30
On Wed, 23 Dec 2009 12:32:21 -0800 (PST), UC <uraniumcommittee(a)yahoo.com>
wrote:

>anyone here own it?
>
>I have a smallish file that I have to use large in a calendar. It has
>obvious pixellation. Is anyone willing to help me process this into a
>larger size?
>
>Thanks

I've used Genuine Fractals a few times, but the paint-by-numbers appearance
of the detail enlargement isn't to my liking. It's also very costly for
what it claims to do. As well being rather convoluted in functionality, not
easy for a novice to use.

You might want to search out any utilities that use the S-Spline upsampling
technique. One of my favorites being BenVista PhotoZoom Pro.

While not having S-Spline methods, Qimage (for printing) and Cleanerzoomer
are both adequate upsampling utilities too. As are some editors that
include more than the common bicubic resampling methods. Some subjects work
better with certain resampling algorithms than others (Bell, Pyramid,
Triangle, Lanczos, etc.). There's no real cut & dried answer in this
regard. But S-Spline capable utilities are usually the safest bet for
upsampling.

You might like to also apply a Fourier Transform utility after upsampling,
commonly marketed as focus-sharpening utilities. One of my favorites is
marketed under the name of Focus Magic. When not applied too strongly it
can tighten up some of those soft edges created when upsampling an image
greatly. Creating some false detail where there was none before. Focus
Magic, when used together with S-Spline upsampling first, can create some
quite believable detail. Sometimes it's best to apply a Fourier Transform
in several smaller and weaker stages, rather than one large strong step in
focus adjustment. Applied to strongly and you'll get annoying, what they
call, "ringing artifacts". You'll learn to recognize what this is the first
time you play with a Fourier Transform tool. It's easy to get those ugly
artifacts if you aren't careful.

Any tool is only as good as the talent of the person manipulating that
tool. This applies to all editing software (noise removers, resampling
tools, basic editing, etc.) as well as the cameras in hand.

From: Bob Williams on 24 Dec 2009 03:09
UC wrote:
> anyone here own it?
>
> I have a smallish file that I have to use large in a calendar. It has
> obvious pixellation. Is anyone willing to help me process this into a
> larger size?
>
> Thanks


How big is your original? (X pixels x Y pixels)
How big do you want the Calendar image to be? (X inches x Y inches)
Do you have Photoshop? Any other image editor?
Bob Williams
From: Martin Brown on 24 Dec 2009 03:44
UC wrote:
> anyone here own it?
>
> I have a smallish file that I have to use large in a calendar. It has
> obvious pixellation. Is anyone willing to help me process this into a
> larger size?

GF is mostly marketting hype. You cannot magically blow up tiny bits of
image and obtain more detail from nowhere in the style of Bladerunner.

In some limited instances where natural textures like trees or rocky
crags that are roughly self similar fractals GF can invent plausible
looking "new" but fake detail. These conditions are seldom met in reality.

Your best bet is to upscale the image by factors of 2 with the best
interpolation function your image processing package provides (some
would argue for multiple upscalings of 1.1x) and then apply unsharp
masking to taste on the final image to bring the edges back to
sharpness. It might be good enough if you are *very* lucky.

You are unlikely to be satisfied with the results however it is done.

Regards,
Martin Brown
 |  Next  |  Last
Pages: 1 2 3
Prev: Is there a model identfier for SD memory?
Next: RAW vs. JPG: Fidelity vs. Convenience