Genuinefractals vs Smart Scale 1

[Servicecard]

Hi

There a couple of software products/plug ins in the market (there are probably others) for scaling up images without supposed loss of quality - Genuinefractals (Of which I think there is a LE version as well as the full) and Extensis has recently launched Smart Scale.
Has anyone used them?
Do they do what they say they do?
How do they compare to Photoshop's resampling?
And is one better than the other?

Many Thanks

 

[thombathe]

I have never used Genuinefractals but I seen it done on a few posters.
They took an 8X10 high res image and turned it into a 5 foot standup poster. It worked very well, it took a long time to rip but what the heck. I have never heard of Smart Scale, I'm sure the price is about the same but if you do a lot of that its worth it.

Thom

The longest jounney starts with the first step.

http://www.thombathe.com

 

[SPYDERIX]

Hi,

First of all let me say that nothing can increase raster images with NO LOSS of quality. To believe that is like believing the sky is green, the grass is blue and all our water supplies are actually pure arsenic.

Don't but into these scemes. Products like genuine fractals work to a certain degree but never give you a perfect image. I have a friend who got a movie size poster out of a 1600 x 1200 wallpaper image, but it was not in focus, blurry and didn't look that great up close. But when you look at the poster from a distance it had more clarity b/c the physical area appeared smaller in your field of vission.

Hope this helps!

Nate

http://www.mainframe-webdesign.com

 

[Servicecard]

Thanks for the messages - I'm going to try and get a demo of each to see quality.

I am not certain how Genuinefractals works , but reading the Smart Scale blurb the way Extensis explain it is that the plug in takes the original size rasterized image, converts it to vector 'outlines?' (I think this is what they mean) then increase to whatever size specified and then re-rasterize the image. It sounds plausible and I can sort of see the logic, but the proof is in the pudding.

I wouldn't use it for poster size, but as I have to manipulate images for customers, very often the customer wants their 72ppi image they have grabbed from their website for a printed advert or brochure (so 300ppi needed) - and they can't understand why the image falls apart at the seams because - "It looks alright on screen" - sounds familar to anyone?

 

[duncdude]

Bear in mind that to get approx 300 dpi you will need to blow the image up 400%. This means the image will be need to become 16 x area of original (4x4). That's alot of data and my guess is it won't be good.

You'll probably find that if you blow it up to 150 dpi in Photoshop that it will look every bit as good. The additional resolution won't do much.

I own a drum-scanning business and own a ICG 355i, true 8,000 x 8,000 dpi, vertical drum scanner. It cost £50k! It doesn't get used much now as 72 dpi images sourced from the web, digital cameras and low-quality flatbed scanners are ruining final print quality.

So webmasters now are providing images to be used in publicatons, digital cameras make everyone a professional photographers (of course) and muppets with no reprographic knowledge are doing the scanning.

Great!

Duncan

 

[blueark]

It's easy to enlarge any image with no loss of quality:

1: Change the interpolation mode (in preferences) to Nearest Neighbor
2: Enlarge your image in multiples of 2 (eg, 200%, 400%, 800%...)

That's it. Ok, so your pixels are the size of bricks, but theoretically, there's no loss in quality because the original pixel data is still there.

My point is, marketing-speak is usually misleading. When software says it can scale without loss of quality, it's probably telling the truth, but people have different expectations.

When you view an image at, say, 25% enlargement, and zoom in to 100%, you don't see better quality, but more detail. If you open an image that's 4 times smaller, and enlarge it 400%, there's no more detail to work with, so the results won't be anywhere near as good.

The software packages you mention are simply alternatives to Photoshop's interpolation, and in some instances may produce better results. Personally, I don't think they're worth the money: Smart Scale is similar to enlarging in Photoshop and applying a Median or Watercolor filter, and Genuine Fractals, well, I don't really see any advantage at all. Try out the demos, but don't expect to be blown away.

 

[SPYDERIX]

Ok, I need to clarify something.

Blueark is correct that there is no loss in quality with his method b/c all the pixel info is still there. However, that technique would only apply to billboards (or images you view from a distance) because from a distance the pixels will appear to be closer together and more compact giving a good quality image. When enlarging images to be used on billboards and posters etc often printers will use a technique known as halftoning to simulate the appearance of continous color by placing a series of dots close together, but not right next to each other like your household printer would do.

Take the example of a city bus in downtown Vancouver with the giant McDonalds big mac and fries. That image has been half-toned. When you are right next to the bus you can see the small dots that make up the image, but they have taken into consideration that you aren't going to be up close and personal with the bus looking at the image rather you will be 5-10 feet away or more. That extra distance is enough to fool the eye into thinking the image is continuous color, and that the dots are close enough together to yield a hi-resolution image.

But...

What I said previously refers to enlarging an image while still looking at it up close. There is no way to get a raster image to enlarge perfectly with NO LOSS of quality while still looking at it up close due to pixelation.

When you are working with enlarging images for use in billboards or large print take into consideration the size the image will take up in someone's field of vision. If it's a poster that someone will be looking at up close then that image will take up most of the person's field of vision and the pixels will have to be small and close together, however when you view a billboard on the side of the road as you drive by it takes up a small portion of your field of vision and therefore the pixels that make up the image don't have to be as small or as close together.

Hope this helps!

Nate

http://www.mainframe-webdesign.com

 

[dreadlord]

I tried genuinefractals and it did not work well for me at all. The best way to enlarge images perfectly is vector, that's what I'm sticking to.

 

[duncdude]

"Small & close together..."?

The distance between the centres of two adjacent 'pixels' (picture elements) in a 'Amplitude Modulated' screen pattern (one that is fixed rather than random in the case of 'Frequency Modulation' for example) is simply the resolution. The dots vary in size depending on the percentage fill in that particular area of the picture. The distance between the centres of the dots stays the same - RESOLUTION. A picture printed in a magazine may be 300 dpi and on a large poster may be 50 dpi. The halftone process is the same.

Duncan

 

[SPYDERIX]

When I said "closer together", I meant more pixels per inch (PPI).

Nate

http://www.mainframe-webdesign.com

 

[blueark]

Spyderix,

Actually, I was agreeing with you - my tongue was firmly in cheek! My point was that you can say there's no loss in quality and theoretically be telling the truth because none of the original image data was lost. I was giving an exaggerated example of how that could still hold true.

The problem is that the extra data that these plug-ins pad your image out with is not what we're led to expect. After all, the software can only guess what detail is missing. Sorry if I was unclear - I did work in large format printing for years, and I hope people don't think that the example I gave is a recommended way of working!

 

[SPYDERIX]

When you posted that you made me stop in my tracks for a second, because you are correct, but like you said not a recommended way of working.

Nate

http://www.mainframe-webdesign.com