up-sizing paintings?

[Mu]

Hi,

I have inquired to a digital print shop about the print size and resolution and found out that anything over A2 format was going to be printed not with 300dpi but with 100dpi, because they use a whole different brand of printers for the bigger jobs.

So, you might inquire wether you have reached those measure already with your local pro-printer.

As for actual upsizing: I guess you're just stuck with what you have...:shrug:

[BaronImpossible]

I agree, 300dpi seems excessive for a portrait of that size. 150dpi or even 100dpi will do the job fine (that's around 2400 - 3600px on the longest axis)

To answer your question about upsizing; you certainly won't lose any detail by upsizing, your image will just become pixellated. So I guess you mean is it possible to still have a sharp image after upsizing, and the answer is no. There is some software out there that helps minimise the effect but of course it's not perfect (sorry, not sure what the name of the s/w is, perhaps someone else can help?)

[ThePhotographer]

I have an Epson 7600 Pro tracer and I must admit that the results are sometimes much better than expected with smaller resolutions. A couple of weeks ago, I printed a digital photo from a customer at 60 x 80 cm and the filesize was 1,5 megs. I was honestly very surprised by the result - it actually looked quite good !

That said, with paintings, I always go for 300 ppi from the beginning. That doesn't mean that it wouldn't look good at 150 ppi though - just that I don't do that with paintings .... you know .... you always want the best result possible ....

[dougtoney]

After some noodling around with it I think I came up with a compromise of your suggestions. I took a painting that I had done previously that was 11 x 15" at 250 dpi. I then changed the image size in Photoshop so that the smaller edge was 16" and the larger edge ended up being more than 24" but still leaving the image size at 250. Surprisingly it didn't pixelate. It didn't even really soften the picture. I was pleased with the small portion that I printed out so I believe that I'll do the image about 11 x 16 (2/3 of 16 x 24) at 250 and then scale it up when I go to print it. Thanks again for the responses.

[acmepixel]

I'm looking for information on up-sizing a painting. I'm doing a portrait for my friend and she would like it to be 16 x 24". At 300 dpi this is very taxing on my powerbook's memory. Does anyone have a good process to do the image at a smaller size and then upsize it without loss of detail? Or even a good enough ratio from small to big? Any help would be appreciated. Thanks.

I usually work at about half to two thirds the final artwork size. Unless it's a billboard or really big mural, then I sometimes work at on quarter size. The secret is to then convert the bitmap file to math using a plugin like Genuine Fractals. Once converted, it is no longer pixels and becomes resoluton independent. It can be enlarged with virtually no loss of sharpness, unlike a pixel resampling algorithm that Photoshop or Painter uses. Such plugins are a must if you need to output 40x60 inch art posters from programs like Zbrush that have a 4096 pixel output limit.

With Genuine Fractals you can go from 8x10 @ 300dpi to twenty feet or so, with no loss of detail.

check it out
http://www.ononesoftware.com/
http://www.digitalgraphicsresources.com/index.asp?PageAction=VIEWPROD&ProdID=1136&gclid=CNrsq8700IYCFTt1FQodr0bp6Q
http://www.hallogram.com/genfractals/?source=google&gclid=CNjf_s_00IYCFQI6FQodykBPBA

[Mu]

Okay, this is a very interesting tunr in this discussion, because I was wondering about something like this earlier on:

I don't know if any of you know "Inkscape" which is a free vector graphics program (the commercial equivalent with a mountain of features would be Adobe Illustrator...). This software has a feature which lets you take a pixelbased image and convert it into vecotr based shapes which then scale to any desired size without getting pixelated.

Methods like this would of course work, too, once the image is done painting.

Has anyone ever used a feature like this and, acmepixel, is it comparable to what you suggested with GenuineFractals?

[acmepixel]

Has anyone ever used a feature like this and, acmepixel, is it comparable to what you suggested with GenuineFractals?

Hi Murat. I've used Adobe's product; Adobe Streamline. It is a bitmap tracing algorithm for producing vector art from images. But no, this technology is an entirely different animal than fractal image conversion.

[Mu]

what does fractal image conversion actually do, from a technical point of view? I am not asking for mathematical formulae (seeing that I am more or less an algebra idiot...:D ), but rather what the differences are to converting the pixels into, let's say shapes the boundaries of which are defined by Bezier curves or whatever vectorising exactly does.


That's a totally new piece of information for me and I am sorta thrilled...:D

[Jinbrown]

Hi,

Images can be "up-sized" by simply increasing their dimensions or they can be interpolated to make the dimensions larger while adding new pixels to attempt to retain the best possible visual quality.

Simply increasing the image's dimensions without changing the Resolution (ppi number) to add new pixels enlarges the existing pixels and the result can be pixelation (jagged edges along edges and curves).

Interpolation is usually the better option and depending on the software used and how much the image dimensions are increased, the kind of image it is, and how/where it will be viewed, the results can be from good to at least acceptable.

Here's a partial quote of a message I posted in another forum several months ago. It contains a link to a site that should help explain all this better than I can:


Interpolation, basically, means to insert something new between other things already existing.

In the context of digital images, it means adding more pixels between the existing pixels when resizing an image to make it larger. When that's done, the software has to "guess" what colors those extra pixels should be based on colors of the surrounding pixels. Some software does it better than others, the amount of increase in size can affect the outcome, and some kinds of images are better candidates for a successful result.


I found a link to what looks to be a very good tutorial at Wikipedia.org in their article on Interpolation. I'm going to bookmark it because it's on an external site and since it looks like there's a lot of other very good information in it that will be useful every now and then. You might want to do the same.

Digital Image Interpolation (http://www.cambridgeincolour.com/tutorials/image-interpolation.htm)

If you scroll down the page a short distance, you'll see a very pertinent demonstration. It's the one under the heading Image Resize Example and it shows very clearly what happens when an image is resized larger, first before interpolation, then after interpolation, and when no interpolation is done.

Other parts of the tutorial cover things we run into as digital artists too. I haven't had the time to read it all so if you do, I'd love to hear what you think.

In any case, this is stuff any serious digital artist should learn, especially if they ever plan to print their work or have someone else print their work. It's best to understand what can affect the finished image's print quality even before starting the image as when it's finished, it can sometimes be too late.


If you're going to have your image printed at a print shop, the best way to work is to talk to the print shop expert before starting your image, telling him/her:

What kind of image it will be (sharp detail, text, etc. or soft like a watercolor painting, for instance)
How large you want it printed.
How it will be viewed (up close, or from a distance)
What kind of material you want it printed on (smooth paper, canvas, etc.)

Then your print shop expert can tell you what Resolution (ppi number) to use that will work with their print machines to give you the best result.

[acmepixel]

what does fractal image conversion actually do, from a technical point of view? I am not asking for mathematical formulae (seeing that I am more or less an algebra idiot...:D ), but rather what the differences are to converting the pixels into, let's say shapes the boundaries of which are defined by Bezier curves or whatever vectorising exactly does.

That's a totally new piece of information for me and I am sorta thrilled...:D

Hi Murat, fractal technology has been around for a while. As a matter of fact, the JPEG format is one of the earliest uses. Then came the MPEG format. Genuine Fractals takes the concept a bit further with more complex algorithms that convert each pixel to math, with special emphasis on the high frequency areas of the image or edge transitions. There were a couple of programs in the early 90's that first used this technology, Live Picture, and later on, XRes by Macromind. But faster chips and cheaper RAM basically put them out of business. Unlike vectors, fractal technology does not convert shapes but each an every pixel in the image is represented mathematically. It's easier to think of it as being like vector technology, even though it is not. So, unlike vector conversion, which cannot convert the fine detail and textures in an image, fractals convert every pixel and thus every detail in the image. Vector graphics are actually small programs or scripts. Specifically, PostScript, which can be very cumbersome with detailed images with lots of blends and overlapping shapes.

So, as long as you paint an image with enough pixel density, say, 9 inches by 12 inches at 300 dpi, plugins like Genuine Fractals can create a resolution independent mathematical conversion that is saved in a proprietary file format. This file can then be opened at any resolution, up or down, without any perceptible degradation from the original.

[Mu]

acmepixel,

thanks for the reply! I think I will have to try it...:D

so many things to learn...:sigh:

[Cris-Palomino]

Genuine Fractals is a great program. They came to show this program to us at Disney when it first came out. The company was based near us in Burbank.

Digital Cameras at the time had very little resolution (compared to now) and they were able to take those small images and blow them up to billboard size.

Photoshop has gotten much better at interpolation over the years and so enlarging is much more accurate. Rule of thumb still remains that you have more room to increase size the larger the image is to begin with and that it is in a lossless (tiff, psd) as opposed to lossy (jpg, gif) format.

Cris

[Nuages]

I use genuine fractals too! Probably the one of the best tools to increase your fotos! There is another tool that might working great as well: S-Spline (Photozoom). A tutor at my university told me, that genuine fractals is better for architecture and landscapes, photozoom is better for portraits. He is a professional photographer and should know that... for me I can´t see any differences!

If you want to print your painting with inkjet, you do not need 300 dpi. 150 dpi are enough. You need 300 dpi for offset or digital-print!

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.