Welcome to Fractal Forums

There are many ways of getting one's fractallography printed professionally, and I suppose soon enough someone will make a topic asking what methods people use, so I thought I'd post my response first to save time.

If you have a deviantART ( http://www.deviantart.com ) account, for a small fee you can upgrade to a prints account. Then you could make money on your artwork submitted as prints.

The only limitation is that one's image must be sized to print. When rendered or modified, the DPI (dots per inch) setting must be calculated to allow a) a high level of detail, and b) the size of the work to fit the dimensions accepted by deviantART. (They accept many different sizes and aspect ratios. You can see their list here: http://help.deviantart.com/337/ )

When I upgrade to a print account (in near future), my submitted renders will be 2400x2400 PNGs with a DPI of 300. This should make an 8"x8" print.

When submitting prints to deviantART one should also decide beforehand how much to charge per print, so as not to price oneself out of the market by being too cheap or expensive.

As this is just one option, I welcome others to share their professional printing methods here.

I am lucky to have a friend with a gallery and frame shop and a big printer. She is enthusiastic about the idea of selling some of my fractals. My main problem is getting large enough images. Using Fractal Explorer's batch processing mode, I can go grocery shopping and come home to find either a 3600 to 4800 pixel wide image, or 50MB of corrupt mess, depending on the mood of my computer. No use trying for anything bigger, with what I have. I am aiming for a one to two foot size range in prints, for wall decor. At 300dpi, that's a lot of pixels :(

It is possible to enlarge them to a certain extent. I use IrfanView for that, because it works best with available computer resources. I must be patient, but it never gives me out-of-memory type errors, and always comes through.

So far I have found that making a very small enlargement - only about 50 pixels - using the B-spline resampling filter first, and then cranking it up all the way with the Lanczos filter, gives the best quality enlargement. The first incremental change serves to anti-alias all the hard edges. (Yes, I could do that when generating the image, but it would take three times as long, and the result would be no more certain. I might wait seven or eight hours for another failure rather than two or three.) IrfanView sets the DPI too. I save them in TIF format (best for my friend's Mac) and burn them onto a CD.

The enlarged images are not sharp at full size in the monitor, but when reduced to printed size they should appear as clear as the original unless viewed with a magnifying glass.

We're still in an experimental stage, nothing has been printed yet.

 I have found that the DeviantART print account is a great way to get printing done. I have to add though that it requires a bit of trial and error. I order small prints of my artwork to check out the quality and found that some of my images were printing too dark and needed to be lightened and resubmitted. I have to also consider whether it's worth rendering at say for example 3000 x 6000 at 400 dpi for printing. I've had renderings take up to 106 hours for the purposes creating a print file. Has anyone used Kinko's? I talked to a rep at Kinkos and he told me that my file sizes would lock up his system. Of course they require files to be in Photoshop formats which may make a hell of a difference.

Living in a major metropolitan area, I am lucky enough to have my choice of various professional printing services (some within walking distance of where I reside).  And this gives me the option to work directly with the person that will be doing the printing of images I supply.  We can work out any problems very quickly and I can do "test prints" right then and there to check for possible color and/or brightness problems.

I normally go for various poster sizes, starting at a typical size of 24-inches by 18-inches at 300-DPI, and then increasing in size from there.  They are produced with a high-end commercial ink jet printer, using oil based, UV resistant inks, on heavy, high quality paper.  And something in this size and format costs me around $35 to $45 US dollars.

Having so many choices locally keeps prices at a fairly competitive rate, but it is always wise to shop around for the best quality and price.

w00t I saw my first big prints today. Just have to celebrate. w00t

drat, no beer.

Edited to add...
Sold one. Maybe I'll get that beer some day.

i've had about a dozen or so posters printed..

usually i render around 4800 x 3200 pixels from fractal explorer (ends up well over 50mb)
and have them printed on 24" x 18" at 300 dpi

this usually costs me around $30 cdn

this is by far the best way to view fractal art in my opinion..

the best results i've had for printing, is render several fractals with slightyly different filters, colours and parameters.
then layer in photoshop to give a more detailed effect

Some of you may have noticed that I offer "ordinary" unlimited edition prints as well as limited edition ones.
The unlimited edition are inkjet or colour laser (which has improved vastly recently).
The limited edition are a choice of either Giclee on canvas or Durst Lambda prints.

At one point I started rendering all my work at 8000*6000 with an eye to having them done at 20"*15" at 400 dpi (the 'correct' dpi for Durst Lambda).

So far the largest I've actually had printed is an inkjet at 40" by 30" which used a source 12000*9000 i.e. 300 dpi.
I also had to render "Starfruit" at 12000*9000 for the ICM 2006 exhibition.
The longest render time was two solid months for "Aquamarine" for which I rather overdid the UF oversampling option :-)

I'm still investigating printing possibilities, though I don't think I'll be using any online ones, I'm far to picky about colour matching - if I could afford it I'd be using the professional matching method with colorimeter hardware etc.

i really wish i had access to good printing services, most of my fullsize images are 13k by 8k and upwards and my walls are bare :(

Aaron H. "ansr23" or "joey jojo jr shabadu" wrote from Toronto:
>
>    i've had about a dozen or so posters printed..
>    usually i render around 4800 x 3200 pixels... (ends up well over 50mb)
>    and have them printed on 24" x 18" at 300 dpi
>

Actually, you would need to generate an image at 7200 x 5400 to get a 300 DPI final image of 24" x 18".

Based upon the resolution size you list above, along with the 300 DPI, the final image would only be 16" x 10.66" in size.  But if you used a 4800 x 3200 pixel image for printing a 24" x 18" poster, more than likely it was printed at 200 DPI.

i usually expand my images to the desired size in photoshop, then apply filters, work on the colours and whatnot. I don't usually have the final print size in mind when rendering the image. I find that 4800 x 3600 is definitely high enough detail for that size. Usually, i don't really know how good they're going to look untill they're finished rendering, and end up geting rid of most of them.

Aaron H. "ansr23" or "joey jojo jr shabadu" wrote from Toronto:
>
>    i usually expand my images to the desired size in photoshop,
>    then apply filters, work on the colours and whatnot.  I don't
>    usually have the final print size in mind when rendering the image.

The problem with that is, your original rendering only has so many pixels.  By expanding the image, it becomes distorted, as in "blocky".  The larger the expansion from the original the more the distortion, causing visible "blockiness" in the image.  This is what happens with raster/bitmap images.  Only vector images may be expanded without causing distortions in the original.

It is usually always best to render much larger than what you intend on printing at (most people say no less that twice the final version).  You can always reduce and image by resizing/resampling, but increasing is another matter.

i find the quality good for what i'm doing on a poster that size
300dpi is extremely detailed

the reason i generally don't go any larger is because i work with multiple layers so
any larger file size and i'll crash my current machine. there's also the rendering time

I also favour more fractal "textures" than harder edge looking stuff, so any 'blockiness'
is not really visible, and photoshop seems to do a pretty good job expanding images, especially
when using some filters after expanding.

thanks for your advice, i'll definitely consider final print size when rendering my next one.
are there any file size limits for software like fractal explorer or ultrafractal?

How do "dots per inch" translate into "pixels per inch"? I think a dot in CMYK is either plain white, cyan, magenta, yellow or black; whereas a pixel can have many more values. So is it really useful to go on a 1:1 basis?

that's why on a screen we can get away with 72dpi, whereas in print you want 300-600 and sometimes even more for photographic quality.

Bernard Helmstetter wrote:
>
>    How do "dots per inch" translate into "pixels per inch"?
>    I think a dot in CMYK is either plain white, cyan, magenta,
>    yellow or black; whereas a pixel can have many more values.
>    So is it really useful to go on a 1:1 basis?

Rarely does DPI ever translate into PPI (unless one is scanning, or digitally photographing, a printed image).  And then it does not mean much anyway.

In most cases, the digital image size, represented by the number of pixels (width x height), determines the size of the output, based upon what DPI will be used.  For example, if the original rendering was 1600 x 1200, then printing at different DPI values would produce the following:

    DPI         Printed Image
-------------------------------------------
   100           16"  x 12"
   200            8"   x   6"
   300          5.3"  x   4"
   400            4"   x   3"
   600          2.6"  x   2"

Aaron H. "ansr23" wrote:
>
>    the reason i generally don't go any larger is because i work with
>    multiple layers so any larger file size and i'll crash my current machine.
>    there's also the rendering time        ...................     are there any
>    file size limits for software like fractal explorer or ultrafractal?

One way or another, there will always be file size limitations, regardless of the software application used.  There are several factors that should be considered when rendering large images:

  1.   The CPU/s and it/their capabilities.
  2.   Available Hard Drive space.
  3.   Amount of RAM installed and usable.
  4.   Virtual Memory allocation.
  5.   Whether rendering from the application window or using a "Render To Disk" option.

By choosing the Render To Disk approach, an image can usually be generated to create a high-resolution image with better quality than is possible from the application's window.  Which means larger images beyond the capacity of your video adapter.

Perhaps my question was not clear. I was speaking in terms of pixels per inch of paper, not per inch of screen. You seem to claim that we need as many pixels as we can print dots, but in my opinion it is not so useful.

Bernard Helmstetter wrote:
>
>    I was speaking in terms of pixels per inch of paper,
>    not per inch of screen.  You seem to claim that we
>    need as many pixels as we can print dots, but in my
>    opinion it is not so useful.

Printed images never do have pixels, they use "dots" of colors.  To my knowledge, nobody refers to the "dots" on printed output (paper or otherwise) as PIXELS.  Which is why there is a difference in terminology:   DPI for printing   and   PPI for rendered images.  The term  DPI is often used interchangeably with PPI, causing a lot of confusion, however, DPI refers to the resolution of a printing device.  PPI is a measurement of image resolution that defines the size of an image.

There are several thousand web pages that can help in your understanding of these differences.  Try a quick search with Google, using the following two words:   DPI  PPI

By rendering an image with more pixels than you think you may need, it will give more flexibility in the printed output.  It will allow you to resize/resample the image if necessary to accomplish anti-aliasing, and to print with a finer DPI value for a higher quality of output, without getting a really small printed image.

I know all of this! I have never called a dot a pixel. But obviously, values of dots come from values of pixels, and in average we have x pixels per dot. The question is: is it really necessary to have x=1 to get optimal printing results, or we can we do with less?

Next time I'm writing in chinese, we may understand ourselves better  :P

easy now, not everyone on these forums makes the distinction and i don't think that was directed at you :)

edit: to answer your question, no, they are not directly comparable - for example the change of colour space, as you mentioned, makes a direct resolution comparison impossible. as a rule of thumb a dot pitch (as it's usually given) of .28 on a monitor looks reasonably smooth, and 300dpi on a printer (varies from printer to printer of course) looks good.

yes, but there needs to be a destinction between inkjet and commercial printing

commercial printing presses use halftone dots for shading, the level of detail is given as lines per inch (lpi)
inkjets however use colours for shading and randomly disperse the ink to get the overall desired colour (stochastic colour)

from what i've read, for inkjets, a file output of 1/3 the printer's resolution is recommended. so... 300 dpi = 100 ppi,
and 300ppi = 900dpi!!
if the file is larger that's fine, but it will only slow down printing, and may cause unwanted blurriness

If the original rendering was 1600 x 1200, then printing at different DPI values would produce the following:

    DPI         Printed Image
-------------------------------------------
   100           16"  x 12"
   200            8"   x   6"
   300          5.3"  x   4"
   400            4"   x   3"
   600          2.6"  x   2"

Pixels per inch (PPI) or pixel density is a measurement of the resolution of a computer monitor's display, related to the size of the display in inches and the total number of pixels in the horizontal and vertical directions.  PPI may also be used to describe the resolution of an image scanner or digital camera; in this context, it is synonymous with samples per inch.

The measure of pixel density is useful for calibrating a monitor with a printer; software can use the PPI measurement to display a document at "actual size" on the screen.

The apparent PPI of a monitor depends upon the screen resolution (that is, number of pixels) in use; a monitor in 800 by 600 mode has a lower PPI than the same monitor at 1024 by 768 mode.

   http://www.michaelfurtman.com/pixels.htm (http://www.michaelfurtman.com/pixels.htm)

   http://www.tildefrugal.net/photo/dpi.php (http://www.tildefrugal.net/photo/dpi.php)

   http://www.shortcourses.com/pixels/printed.htm (http://www.shortcourses.com/pixels/printed.htm)

I don't know about Fractal Explorer but Ultrafractal has a "render to disk" option which effectively makes the possible render size unlimited - actually it makes it more subject to the file size limits of the system you're working on and even that's not that relevant since you can render direct to jpg.
Having said that however in UF there are still issues with rendering IFS or flames using the "normal" methods (or other fractals calculated in the "global" section - like some 3D fractals or strange attractors) which make the image size you can manage dependant on how much memory is available.

Bernard Helmstetter wrote:
>
>    .....values of dots come from values of pixels, and in average we
>    have x pixels per dot.  The question is:  is it really necessary to have
>    x=1 to get optimal printing results, or we can we do with less?

You can do anything you choose to do, even if it is "with less".  And it is not "necessary to have x=1".  As they say, anything is possible.  But if you want QUALITY, then certain values should be followed.

A single pixel can be printed to paper and take up the space of 3-inches by 3-inches, though it may end up looking like an Andy Warhol image.  That same single pixel can be printed to paper and be just a portion of a single "dot" on the paper.  Pixels and Dots are not really comparable in the way you seem to be expressing.

You may be interested in reading this:
    http://aic.stanford.edu/sg/bpg/annual/v15/bp15-08.html (http://aic.stanford.edu/sg/bpg/annual/v15/bp15-08.html)


Bernard Helmstetter also wrote:
>
>    Next time I'm writing in chinese, we may understand ourselves better  :P

That is fine with me, I will have no problem, you may also choose any of the dialects that works best for you to communicate with.  I can also communicate in several other languages as well, if that better suits you.

I've written about optimal printing results, so of course I want quality, even if I'm not screaming it. But I would also rather not send one of my 8000x8000 130MB PNG images on the internet if it's not necessary.


You are the one who has repeatedly been claiming that we need as many pixels as dots. Did I miss something? When I write that there is less information in a dot than in a pixel, do you understand me? Do you agree?

That's a good link. Thank to all who answered me btw, even if I'm still a bit lost.


I could enjoy exercising my humble knowledge of chinese by discussing fractals with you, if you master the language; but you must have noticed the smiley. I had better choose easier topics.

Bernard Helmstetter wrote:
>
>    You are the one who has repeatedly been claiming that we
>    need as many pixels as dots.   Did I miss something?

I do not believe I stated what you claim I said.  I have been saying that pixels and dots are different, and that the number of pixels will take up a certain amount of space on the printed paper based upon the printer's DPI setting.  It is nothing more than a simple formula to follow when one knows what size they wanted the image to be printed at and the DPI value.

>
>    When I write that there is less information in a dot than in
>    a pixel, do you understand me?   Do you agree?

Well that depends on the context of how each word is used.

Literally, there is no information in a printed DOT, it is just a splash of color on the paper.

And the PIXEL (the picture element) is very context sensitive when it comes to a definition and it's use.  When we talk about a graphic image format, it usually means the smallest complete sample of an image.  And that usually takes about three bytes of storage for a single color out of the number of distinct colors that can be represented.  But it could be as little as one byte of storage, or as much as six bytes of storage, depending on the number of bits per pixel (bpp).

But a single printed DOT could have been a combination of several PIXELS in the final output, and therefore required more bytes of data to make up that DOT.

If you upload to DeviantArt, do so in JPG. While I agree that JPG is sacrilege, and PNG is better, the fine print on their site reveals that they actually convert your PNG to JPG on their end   :-\. I'd rather have direct control of any compression going on, if there has to be compression at all.

You can do better with an old DOS program called Fractint.  Version 18 or better allows you to edit the screen resolutions all the way up to 20,480 pixels squared.  You first have to get used to the DOS interface which is primarily keyboard driven.  The mouse is only used to create selection boxes that can then be used for in and out zooms, rotations, and skewed images.  You can also switch on or off the co-processor (aka FPU) which can result in longer computing cycles, but with seriously funky results.  At resolutions larger than what your video card supports the program also lets you work with billboard sized graphics by switching to a text mode rendering stage.  I cannot recall what it is called inside the program.  What is basically does is let your computer create marvelously huge fractals without ever seeing what it is you are actually working on.  And it does it all in just 640 KB of memory :0)

When I want to make a huge hunkin' fractal what I'll do is make a preivew size at a screen resolution like 1024 x 768 or 1280 x 1024.  Then working with that I will make all of my selections / tweaks and save my results.  Then I will re-load that same fractal and pressing the "Delete" key I can then scroll through to my ridiculous resolutions (using the arrow keys) and choose the video mode I want.  Once that new mode is entered you will see a screen of text showing what pixels are being computed and some form of ETA for the completion of your fractal.  Of all the fractal program I have and do use, I still prefer to use DOS Fractint.

Now, as far as printing goes, I am sold on the LightJets.  There is a company not too far away that I use to make my Limited Edition prints.  The cool thing is I get precisely back what I supply.  Pixel perfect results with zero ink used.  Instead a RGB laser is used to develop Fuji Crystal papers.  For a 16 x 20 I think I paid $18.00.  The image was flat and gorgeous.  I didn't have to worry about paper curl like I would if I had gone to a place like Wolf Camera to have the same fractal printed on one of their Epson inkjets.  The life span of the LightJet is estimated at around 100 years before fading can be seen by the naked eye.  That is the same life estimate I got when I used the local Wolf Camera / Epson print service.  I believe their LightJet has a 200 DPI output and can handle up to 10 feet wide by any length (or until they run out of paper).  Another service I am contemplating is www.wallmurals.com.  They have a minimum print size of 4 x 6 feet.  The samples that I got from them were most impressive.  You might also look into grand format printing.