Genuine Fractals

[sanking]

http://www.ononesoftware.com/landing/2010/05/gf/landing.php?utm_campaign=GF6_Demo_0510&utm_source=QH_GF6DemoGF&utm_medium=email

Is this a fair comparison of Genuine Fractals versus increasing file size in Photoshop?

Sandy King

[PenGun]

Her ya go:

http://carnagepro.com/photo/45/fractals/exult.jpg

Was 2048x2048 in the 'camera' so it's sorta large format, product of Fractint running on DOS 6.2

[Kirk Gittings]

IME its pretty close.

[Peter De Smidt]

At my former work place, the post production guys preferred interpolating in 10% steps with bicubric smoother, with a very slight unsharp mask filter at each step. They preferred this to any of the task specific software that they tried, including Genuine Fractals. They did this test about 2 years ago.

They do a lot of up-rezzing, as clients come in with small files and want big prints on a regular basis.

[Kirk Gittings]

Peter, when I tested this about 2.5 -3 years ago, I found GF to be superior to that technique and GF has had one major upgrade since then which improved it even more.

[Mike Anderson]

At my former work place, the post production guys preferred interpolating in 10% steps with bicubric smoother, with a very slight unsharp mask filter at each step. They preferred this to any of the task specific software that they tried, including Genuine Fractals. They did this test about 2 years ago.

They do a lot of up-rezzing, as clients come in with small files and want big prints on a regular basis.

Do you know if they automated this process (with Photoshop actions or something) or was step by step "hand" done?

...Mike

[QT Luong]

http://www.ononesoftware.com/landing/2010/05/gf/landing.php?utm_campaign=GF6_Demo_0510&utm_source=QH_GF6DemoGF&utm_medium=email

Is this a fair comparison of Genuine Fractals versus increasing file size in Photoshop?

Sandy King

My understanding is that back in the Photoshop 7 days, GF presented an advantage. However, the resizing algorithms in PS have been improved (mostly when they offered the two specific reduction and enlargement options).

I've done a couple of tests this week with the new version of GF, comparing it to PS CS3. My methodology has been to take a 35mm scan or digital file, and enlarge it to my largest printing size (30x45).

Honestly, I don't see much of a difference. The GF file might be a tad sharper, but it is also more noisy.

I was certainly not able to observe a difference as dramatic as the one shown on the page linked. I'll be *very* interested to see such an example with an independent user image.

[Mike Anderson]

Here's one recent comparison I found comparing GF6 to Photoshop's bicubic smoother, about 4.5x enlargement:

http://www.nikonians.org/resources/reviews/onone-plug-in-suite-5-review-en?p=2

To my inexperienced eye the GF6 enlargement is ever-so-slightly better. Barely noticeable. The test image was from a DSLR, and probably had very little noise. I wonder how noise or grain would be effected. GF6 could be sneaking a touch of sharpening in.

The article makes the point that fractal enlargement is less beneficial to smaller enlarging ratios.

...Mike

[8x10 user]

I just took 1000x1000 pixel image and brought it up to 15000x15000 with GF... It took over 3 hours on my old mac. I actually had to run the filter twice as it doesn't like enlargements over 10x.

It seem like the software works better with some images then other. It does wonders with straight lines and things that follow a mathematical patterns like clouds. If you are working with images with noise or grain your going to get an issue with the software enhancing the grain instead of the desired detail. It is best to do your grain removal first. If grain is still an issue use a lower detail setting like "low res JPG" and see if that helps.

[Kirk Gittings]

Here's one recent comparison I found comparing GF6 to Photoshop's bicubic smoother, about 4.5x enlargement:

http://www.nikonians.org/resources/reviews/onone-plug-in-suite-5-review-en?p=2

To my inexperienced eye the GF6 enlargement is ever-so-slightly better. Barely noticeable. The test image was from a DSLR, and probably had very little noise. I wonder how noise or grain would be effected. GF6 could be sneaking a touch of sharpening in.

The article makes the point that fractal enlargement is less beneficial to smaller enlarging ratios.

...Mike

FWIW My experience with GF is mostly on digital capture files, where grain is not an issue and noise has been fixed before hand. Yes GF uses some sharpening, but all that is individually adjustable, though I tend to just use the default.

[QT Luong]

The most thorough evaluation I have seen is by Ctein:

http://theonlinephotographer.typepad.com/the_online_photographer/2008/02/its-bigger-but.html
http://theonlinephotographer.typepad.com/the_online_photographer/2008/03/its-bigger-but.html
http://theonlinephotographer.typepad.com/the_online_photographer/2008/03/its-bigger-bu-1.html

[David Luttmann]

Kirk, do you do any capture sharpening prior to interpolation in GF? I find that GF, Alien Skin and PhotoZoom all end up with a painterly look I'm not a fan of. But the results vary depending upon capture sharpening settings prior to interpolation.

In all honestly, I find the results more natural looking by simply applying some capture sharpening, using bicubic smoother to about 25% past the desired final size, applying USM or Smart on the LAB luminosity channel, adding a slight amount of Guassian mono noise as a print dither, and downsampling via bicubic sharper to the desired print size.

In fact, sometimes just doing capture sharpening and using Bicubic Sharper to the desired size looks better on print than GF. You want the sharpening to look a bit "crunchy" to maximize detail.

Of course, this is just my opinion....some people may prefer the look of GF.

[Bill_1856]

Does it work with PS Elements, or only with the full PS?

[Greg Miller]

At my former work place, the post production guys preferred interpolating in 10% steps with bicubric smoother, with a very slight unsharp mask filter at each step. They preferred this to any of the task specific software that they tried, including Genuine Fractals. They did this test about 2 years ago.

They do a lot of up-rezzing, as clients come in with small files and want big prints on a regular basis.


Bicubic Smoother was created to prevent having to do 10% increments manually. Incrementing an image in 10% steps using Bicubic Smoother each time strikes me as a bit odd.

[Peter De Smidt]

Greg,

I made a mistake in my reporting. They used regular "bicubric" upsizing and not "smoother."

Occasionally they'd use something very similar to Mr. Luttmann's stated workflow. It all depended on the image.

[QT Luong]

I had to use extreme enlargement sizes (that I am not likely to use) to see a significant difference. Here is an example:

original image http://www.terragalleria.com/large-format/np-image-lf.deva20807.html
5x7 scanned to 300MB, downsized to 50MB, resulting in a extremely high quality file.

Upsized to 800MB with GF and PS (a 4x linear enlargement)

GF file is more "clean" and has definitively sharper edges, but texture is lost and replaced by a polygonal mesh.

Overall, my conclusions are in agreement with Ctein's.

[Greg Miller]

Greg,

I made a mistake in my reporting. They used regular "bicubric" upsizing and not "smoother."

Occasionally they'd use something very similar to Mr. Luttmann's stated workflow. It all depended on the image.

Thanks for the clarification. That makes more sense. Although Bicubic Smoother was really intended to automate the 10% stepping technique, thus yielding very similar results but in a one step process.

[Jeffrey Sipress]

And, what about image size reduction? Do I have to worry about any of this and just use the image size window in PS?

[Sevo]

Size reduction is trivial, as the software there has an excess of information to work upon - depending on the target, at least one of the default methods will be near optimal, there is no need to buy extra software for that.

[Peter De Smidt]

For reduction, I generally sharpen a bit at full size, usually with smart sharpen, and then downsize with regular bicubric interpolation. Unlike 'bicubric sharper', this lets me adjust the sharpening applied.

[Tyler Boley]

I generally prefer BlowUp when necessary, but the settings options in the newest version make this sort of direct comparison impossible. Not only does the nature of the original image dictate results and optimal settings, but how it looks on paper. GF has gone through some major evolution in recnet years, so older reviews are no longer relevant, but as of mid 2009 I still preferred BlowUp.
If I were on a PC and had to do this more often, I'd be playing with Qimage as well.

[Lei Meng]

What about LightRoom? did anyone compare Lightroom vs. GF?

[Mike Anderson]

What about LightRoom? did anyone compare Lightroom vs. GF?

I can't imagine Adobe putting better scaling technology in Lightroom than in Photoshop. But I don't have Lightroom, so I'm just guessing.

...Mike

[D. Bryant]

http://www.ononesoftware.com/landing/2010/05/gf/landing.php?utm_campaign=GF6_Demo_0510&utm_source=QH_GF6DemoGF&utm_medium=email

Is this a fair comparison of Genuine Fractals versus increasing file size in Photoshop?

Sandy King

Take a look at Alien Skin BlowUp. I believe it holds the edge these days for products of this type.

http://www.alienskin.com/blowup/index.aspx


Don

[Greg Miller]

What about LightRoom? did anyone compare Lightroom vs. GF?

Lightroom would use the scaling engine in Adobe Camera Raw (so would be the same as RAW conversion form Bridge). Some people (such as Jeff Schewe) believe that scaling in RAW is much better than scaling after the the RAW conversion.

I think much of this, like so many things photographic, comes down to personal taste. There are many good options out there, many with free trials, so people need to take a look and choose the one that fits their aesthetics best.

[Mike Anderson]

Lightroom would use the scaling engine in Adobe Camera Raw (so would be the same as RAW conversion form Bridge). Some people (such as Jeff Schewe) believe that scaling in RAW is much better than scaling after the the RAW conversion. ...

I've always thought, theoretically, scaling the raw data should have an advantage over scaling after the interpolation. It seems to me (and I haven't thought this out completely) that there is more information available in the raw data. I prefer scaling a raw file with DXO Optics Pro, which does a lot of the image processing during raw interpolation.

...Mike

[Kirk Gittings]

It would appear from all the info here that a fresh comparison is in order. All my tests, including a test with up scaling Canon 5D files in ACR were done awhile ago, like with CS3. In those tests ACR was the worst upscaling method between 10% BiCubicS, ACR and GF. Also IME there is a big difference between upscaling on your screen vs. on a print where to me image softness is a greater evil than the odd minute artifact (and there is a large perceived difference between mat and smooth finish papers). While I am a sharpening minimalist, a certain amount of sharpening is crucial and for my money when I did the tests 2-3 years ago GF gave me the best looking prints.

[D. Bryant]

I've always thought, theoretically, scaling the raw data should have an advantage over scaling after the interpolation. It seems to me (and I haven't thought this out completely) that there is more information available in the raw data. I prefer scaling a raw file with DXO Optics Pro, which does a lot of the image processing during raw interpolation.

...Mike

I have had much better results up scaling my RAW image files using Capture One Pro than from ACR or DXO Optics Pro, but I'm not making wall sized prints either. However, that was pre ACR 6.x and LR3 beta. Adobe may have woke up and smelled the coffee, lets hope so.

I think someone on the net found that C1P really did produce the best output from files with exposures captured with digital sensors. Film scans may pose different problems. For film scans I've used PS in the past. I used the original version of GF but the product was pretty slow and PS improved over the generations. I have no doubt that Sandy will start with an optimum scan so PS may work just fine.

At any rate most if not all of these products do a really good job and the final results depend on the original image just as much as the final output size. Ctein's article last year was a pretty good primer for these kinds of software.

Don Bryant

[8x10 user]

Gee I wish I had the images that I worked on before. There were mostly all from the Hubble telescope. Genuine Fractals was really able to add new details to the fine wisps of the "space clouds".