I'm interested to know about others' views about post-processing their fractal imagery in graphics programs, to try and achieve the artistic effects they want. (By post-processing, I'm referring to the process that takes place after an image has been created in a single fractal-generating program, such as tierazon, ultra fractal etc, and not to the more controversial topic of what constitutes pure / impure fractal imagery blah blah blah.) A few things have put me off it in the past: 1) I had no clue either what it was or how to do it in my early days of experimenting with fractal software! 2) Somehow there was always this huge bone of contention too on the newsgroups and mailing lists about p-p, and whether or not it was something "acceptable". I was scared at that newbie stage of stepping out of line! 3) One of the things I really like about creating fractal images is the reproducability of it; ie an image can be created in UF (for example) at 640x480 on screen, rendered are 800x600 suitable for displaying on the web, saved as a simple text file, and at a later date exactly reproduced at, say 80cm x 60cm, burned onto CD, printed, sold, framed and hung on the wall. Is it possible to achieve the same reproducability with a small image that has been p-p'd in paint shop pro or photoshop? Or that is a combination of a fractal and a photo? I have experimented a bit recently with post-processing smaller images using filters in PSPro, but would feel very daunted by the prospect of dong the same thing on image that was 8000 x 6000 pixels - in fact I think my computer would be in danger of exploding if I tried! Does anyone else have these same dilemmas, and are there techniques that others have found to circumvent them?

i am a Post Processing fan. for me... the tool and medium is totally irrelevant. the finished product is what i care about. as far as 'reproducability in various sizes' it depends on what you do in post processing. if you do purely procedural postwork, then it is duplicatable. if you do hand painting and cut/copy/paste then probably not. i know that my work makes most purists cringe... i make the 3d render folks crazy and the fractal-ites... since both groups tend to think it's 'unholy' or somehow cheating, to mess with what the computer creates in it's initial 'render'. hehe, me, i kinda like making folks cringe

So how do you go about the p-p work if you want your finished product to be a poster sized print? Do you do it small, and save it with lots of technical notes to yourself, or do you do it big in the first instance?

i work big from the start. in digital, you really can't get a good image if you need to blow it up too much as there just isn't any data to work with. you can always go smaller. consequently, a complex image will take several 'versions' with various layers exploded... as each section or 'piece' gets finished, it is compressed onto a single layer and the next section gets worked on. one of the real PIA of working for 'hard copy' output. but, ultimately worth it i think.

Whhat kind of computer hardware do you have / need for doing this?

I have difficulty even opening a large rendered file in paint shop pro, and applying a filter or copying it and pasting as new layer will invariably crash the computer, even with regular maintenance, defragging etc. I have a PIII500 processor and 256 ram, running win98se.

By large I mean a bitmap or tif file, 9000x6750 or thereabouts, approx 174MB.

You could definitely use more memory, and upgrading to Win2k wouldn't hurt. (Win98 has an allocation limit of 256m per app. You can have a gig of ram, and it will only let each app use 256m. I switched to 2k for that reason- I find it to be the most stable.) I suppose I really post-process- I create a 3d fractal in Quaternion, then export it to my modeling program to play with. Of course, technically the modeler and raytracer don't count as post processing- or do they? By themselves they don't. There's a question!

Thanks Michele - I didn't know that about the ram allocation in win 98. I'll look into whether it's going to be worth my while upgrading to 2k, but I definitely have some more ram slots I can use on my motherboard.

Tina, I used to p-p when I was fractalizing with Fractint or Stephen Ferguson's programs and I just started to do it again recently with Painter. And most of it is copy/paste/selections... My goal is not fractal purity but aesthetic. And what is fractal purity after all? Most of the computer graphics bitmap programs use fractal algorithms for their filters, brushes, etc. Anything that must look natural use fractal algorithms. As Audre says, you have to start big if you use bitmap images. You must foresee your biggest format when you start your image as you can reduce but you cannot enlarge. If you enlarge, some debasing artefacts like jaggies or white dots will be created in your image. So, if your foresee an image printed 20"x20" at 300DPI, you can calculate the number of pixels you will get for a layer in Painter, Photoshop, PSP, PhotoPaint, etc.. Estimate your calculation on a count of 4 bytes per pixel: three for each colour in RGB + one byte for alphachannels. However, many bitmap graphics formats allow compression or have compression schemes you'll be anaware of. So, images won't be so big. But, here I cannot give you useful information. Only vector/fractal generated graphics can be enlarged without loosing or debasing information as the images are directly based on mathematical formulas that are recalculated each time you change the dimension and resolution. Exemples of vector graphics programs would be Illustrator or Corel Draw. For fractal generator, well you know as me. Miche

I persoanly dislike post processing because I dislike having to use multiple programs to achieve an effect so I tend to try to force the one program i'm using to do what I want. In my case it's got nothing to do with trying to stay true to the fractal math, it's just lazyness. My feeling though is that if you are using fractals to illustrate exactly where something is being plotted and how, then you want it simple, you dont want to do anything to it after the computer finishes because if you do, you can't very well tell what the original formula was doing. But if you want to create art, then for heaven sake go create art and dont worry about what anyone says. About your choice of tools or your methods of using them. Or even the final product (unless you're making it for someone specificaly).

My vote: do that which achieves the result you desire.
I have done, and will continue to do, both, depending on my needs for a particular image.

If you do decide to do post work, yes, working large to start with is always a good idea, computer memory and processing capabilities allowing. Also, if you have Painter and use it for post-p, there is a way to record a low-res work session and have the program do the exact same thing in a higher res format. Check around on the web for info on print issues (hint: there's a print forum here at R'osity--you can ask print-specific questions there), and there are many many places that discuss other graphics apps (hint: R'osity forums...)
Hope this discussion topic helps...

Rosemary - I'm interested to know how to do the thing you mentioned in Painter, with recording the processing done on a small image, and at a later date apply the same tecnique to a larger image.

Ideally I'd like to be able to the same in pspro too - is that possible?

When I asked about applying p-p to samller images, I wasn't suggesting that I would then interpolate them to the larger size. Rather that I would then apply the same p-p to a newly rendered larger image, having already decided how by working on a small image first, and having saved the info about how to do it - if that makes sense!

Kelly - as I said in the initial post on this thread, I am not in the least bit interested in the pure / impure fractal argument, only in ending up with the image I want, in a reproducible form. So far I've only been able to achieve that reproducability by sticking to getting the finish I want in UF itself.

Attached Link: http://www.dreampaint.com

Hi Tina Your question is very intersting. i create poster 60x80cm with Ultra Fractal and witha lot post processing. There is no probleem if you havea lot ram memorie (256 is ok). I can make 10 layers without probleem (and with little patient) with a P3 double copro NT windows and same result with p4 win98. Becarefull if you use some filters in photoshop or paint shop pro, some result are different beetween a 800x600 pict and 9450x7090. For exemple Frax flame can't make a good result i can use only 30x40cm with difficulty. In a CD i can put 3picture in PSD format, and after for the printer i use JPG file i high quality without compression and i can put 25 pictsper CD. Hope you have a good harddisk :) have a nice day all. Laurent

Thanks Laurent - maybe I need to just try a bit harder to get my system efficiently before having a go at this. I'm encouraged to hear that you've been able to do this. Tina

aartika:
The recording is done by a function called "Scripting", which is found under the Objects palette in Painter--(at least from version 5 on, maybe in v4 also...). Essentially, it records every action you take and can be replayed, and even edited on a step-by-step basis. There is a full section in the Help on how to do it. You can do a pic in a small size/low res, recording as you go, then start with a large size/higher res pic and replay. But as dreampaint said, sometimes filters and such don't always translate up quite right, so experimentation is necessary.
As for PSPro, I don't have that prog, so I couldn't tell you--but my gut feeling is no, PSPro doesn't do it.

Does it? Lemme check...brb.... ...click click click.... ...hmm. It looks like PSP has something called a "command history". I think it only has linear undo, but I'll have to play with it.

psp has a history list but you cant' selectively delete things. if you want to delete something 5 moves back, it deletes that one and all others since.

Attached Link: http://www.p-gallery.net

Hi,

Most of the post processing process are scale dependent. Suppose I apply a mere "sharpen" filter. What it usually does for each pixel is to substract a fraction of the surrounding ones (at least I think). When doing this, the contrast is increased.
The trouble is that this process is scale dependent, ie the surrounding pixels at 640x480 won't be at the same place in a 64000x48000 render (there will be 100 pixels between them). So the sharpen mask WON'T act the same on images of different size. An if this effect will probably be hardly noticeable with a sharpen mask, it can be spectacular for other postprocessings.
What would be interesting would be to know if there is a way to avoid this, ie to be sure that your high resolution image will look the same that the one you post on a website.

This problem is imho a major downside of (heavy) post-processing.

Regards

Samuel

well...you could post process the large image, get it the way you want it, then shrink it. then modify the shrunken image untill it at least approximates the large one fairly close. maybe not the fastest solution but certainly doable.

I work on the principle that you do what you can to achieve the final result that you are after. And if its Post pro required then do so :) Easy! Stacey

I sometimes work with PP but on a very beginner's level. And since I often foret how to achieve a certain effect again after having layered some PP'ed versions of the same fractal I always get depressed that my effects cannot be repeated (by me).
My greatest love is to produce several effects on my own, therefore I spend most of my fractal time with creating code, watching its effect on some standard Julia spirals or the primitive Pixel formula (a non-fractal fractal) and then delete these experiments.
What I like when working with fractals is - I can reproduce the image at any scale, zoom in and discover new "images", work on them, zoom in again, ...
As soon as I have PP'ed a fractal, this possibility is gone.
And my non-fractals made in a fractal generator - I stop when I am content with them, therefore PP is not necessary, or I am too lazy to do it...
Thanx for this interesting thread!

I did look into using post-processing using either PSP or Photoshop and found the problem that Laurent mentioned - ie. many of the interesting effects work OK up to around 2000 pixels but beyond that you can't up-scale the effect to say an 8000*6000 image (even if you do have a system that could manage it). Obviously the simple overall colour/brightness/contrast changes don't suffer from this problem and I'd hesitate to call using those post-processing at all as they work on a per-image rather than per-pixel basis. I'd suggest we petition the programmer/s (Frederik) to add image processing to the facilities in the fractal program on the same basis as the formulas ie. so people can write their own post-processing algorithms (I think I see a flying pig).

You might try using dogwaffle to do post processing in. So far it's leaving photoshop and paintshop way behind. and it's freeware.

Hey I heard about Dogwaffle a while ago, and it looked good, but I dont like just downloading everything I see that looks good. I prefer to find out what people who use it think. Could you email me off list with details? Im a PSP user

Are we still focused on what Pre-Post/Production in computer art/graphics/IT are? Or have we slid off in a focused site-seeing trip??? Give me your feed bak'..

OK im not really sure how you pre process, unless its something like take a UF image and chuck it into Bryce or XD? Post processing to me is one of two things: Tweaking the image to improve it without radical changes: Sharpen, blur, rezise, crop, frame, add sig etc Further Processing the image: use of filters to give diff effects like watercolour, painted etc. The first is a necessity, no one ever creates a perfect image, the second is a matter of choice. And FWIW I did ask to take my Dogwaffle discussion off list. Stacey

BlueRose: apparently the consensis is that, for the most part, post-processing = your 'second choice'. But as you said, and most seem to agree, it's a matter of choice based on what each artist believes is best for a particular image -in that artist's mind-. Luckily, most folks in this forum, just want to see good art, and aren't worried overmuch about methods. But, the discussions sure are fun! :D

Attached Link: http://www.aartika.co.uk

This has turned out to be an interesting discussion - thank you for all your replies.

Dave - I was thinking about what you said about incorporating more image processing tools into UF - I think the reality is that they are probably already there in form or another, if you know where to find them. Pigs are already flying ;-)

There are the obvious one like layering, masking, adjusting gradients etc, but also many of the existing algorithms in the database are very similar to the filters etc found in PSpro.

For example many transforms such as ripple, waves, turbulence etc mimic some of the effects and deformations - in fact there are almost certainly many more available, with many more variables and in a reproducible form, than in a program like paint shop pro; many ucls include all sorts of different texturing effects; it's possible to make shapes, shadows, frames etc with a little effort.

The one major thing that seems to be missing is the ability to paste in an imported image layer, such as a photo or other artwork - now that would be a very cool tool to have :-)

I've also found it very difficult to make embossing effects in UF - but Kerry's images are certainly a testament to the fact that they are possible.

Overall, for me personally, like Andreas, the advantages of being able to exactly reproduce at any scale, rework an image, and save it all as a very small text file outweighs most of the advantages of using p-p to radically change an image, unless I want to combine, say, a fractal and a photo.

It'll be interesting to see what developments there are in UF v3.

Tina