Welles

Hi Lee,

Here's another bit 'o visual info for you. I've added a little bit of the original .bmp image with the red box approximately around the area I used for my little study. Next to that is a zoomed in verson of the test area so it is approximately as big as the final test patches in the image above. So the zoomed version is the original pixel data prior to uprez.

I like the PhotoZoom much better because, by comparison to PS, the edges are sharp and clean, the highlights aren't pixilated but appear nice and consistent. To my eye, the Genuine Fractals uprez has added a lot of artifacts and random pixels, not present in the original, which look suspiciously like poor quality sharpening. While my images here don't show it, the red fibrous background in the larger original did poorly in the GF version. I prefer the PhotoZoom which has just been enlarged, no additional sharpening.

There's an additional issue with PhotoZoom that I haven't adequately studied. They have another six or eight algorithms for enlarging. Several of the names I'd never even heard before. I wonder what they will do. You know, with your Photographers eye, I'd love to have you try the demo of PZ and give us your assessment.

Lee

Thanks for the additional images, Welles. Even with the new information, however, I still like the GF better. I don't see the artifacts and random pixels that you see. That does not make either of us wrong or right, just a difference in how we each see things. I will try both those as you suggest and let you know.

Blinky2004

Lee this isn't for sharpening, it's for photo/image enlarging.

As seen by Welles' example GF tries to resharpen the enlarged photo to the point of adding pixel artifacts and a general poor quality replication. Look at the high contrast of the object edges in the photo Lee. Plus... GF also add discolouration to -- generally a similar result to Jpg compression.

Yes, the PhotoZoom method looks smooth (less or no resharpening), but if you'll notice too... it DOES reclarify the edges of objects within the photo. PS does this also but to a lesser degree.

Actually it appears as though PS's rendition falls somewhere between the 2 plugins' results.

I think i need to try some print tests now to see the dif at hi-res.

Gametamer

I downloaded a trial version of GF 4, and it appears not to install in PS CS2. I wanted to do a straight comparison between PhotoZoom and GF, but alas...

GF does have some of the same features as both PS's native Bicubic Smoother and PhotoZoom; I suggest that a fair comparison would be to use s-splines with GF and PZ---and Photoshop doesn't offer this setting.

My Best,

Gare

Lee

Thanks, Mark. I guess I need to read more carefully. I knew it was for enlarging but, without reading more closely, though it accomplished that by superior sharpening. Back to the drawing boards on sharpening. From everything I've read, the reviews say the new CS2 Smart Sharpen is much better then the old USM [he says on his knees with a short prayer]. Hopefully they're right.

Silky

Hey, guys, check this out!
1 pic I enlarged in CS2 400%, and the other I enlarged in CS2 and applied "noise reduction" and "smart sharpen".

Welles

Wow! p?tr??k that's a great result. Have you thought about trying that on a JPEG with obvious artifacts?

Welles

OK I just did my final test of PhotoZoom, PS Bicubic Sharpen and GenuineFractals 4. My conclusion is PhotoZoom is the winner from a number of perspectives. GenuineFractals actually does a good job but it only has one option, that of using their proprietary wavelet algorithms as opposed to PhotoZoom which provided about 10.

Additionally PhotoZoom has at least one set of controls to adjust the amount of sharpening in each algorithm and the capacity to compare the potential results of the various options before committing to processing which can take a while. A 600% uprez starting with 1024 X 768 took nearly ten minutes to process on my machine. I had increased the sharpening considerably more than the original preset.

I did give these a print test by taking identical selections from each of the enlarged image files, I put three on a page and printed them at the highest resolution my printers would produce. On the letter size print the PS file was 600dpi and the print was excellent from all three enlarged files. I was partial to the PhotoZoom slightly more than the GenuineFractals 4 because of my ability to add sharpening. Both of those were slightly better than the Photoshop Bicubic-Sharpen result, but, honestly the average person wouldn't have seen any difference. There was a noticeable difference with a 10x magnifying glass.

Next I took the same data I had just printed on an Epson C86, dropped the resolution to 300dpi while keeping the pixel size the same and shot the file to my Epson 3000 where I printed again. At 300dpi, or 4 times as large a printed surface ares, my preference for the PhotoZoom increased over Genuine Fractals and both were significantly better than PS. Once more, though I should add that someone without a critical eye wouldn't really care about the difference. I did, though...

And then I saw what p?tr??k did a couple of posts above this one and determined to purchase a CS2 upgrade and try his technique before I buy PhotoZoom. My head is spinning. 8D

Silky

I saw your question this morning Welles, and wanted to go experament right away. But, we finnaly got a school break and I had to go work. [oops] My mind was left at home on my computer for the day untill I got home. So, anyways, heres a cruddy jpeg, then another with "reduce noise" filter. This abviously doesn't get rid of everything, but it does a great job getting rid of unwanted artifacts and noise.
Hope you like it, and I'm off to go see Star Wars [confused] ......hope it's better than the last 2.

Welles

Thanks, p?tr??k! I'd say that's a pretty extraordinary filter. It seems to have done a better job than most JPEG 'cleaners.' [righton]