Jock,

My network rendering for large images shows the expected speed increase when considering the relative performance of the computers in the network. I don't have the numbers for you right this minute, but I do remember specifically benchmarking the network rendering in order to verify that it was working. If you watch it, though, it sure does seem like it's erratic, and the pixels/sec numbers don't seem to correspond to the overall speed.

Of course, network rendering does not help flames.

--Damien

Mindy,

I once rendered an image that took 900 hours to render, just for a 20x15 print. Between upgrading my main workstation (3x faster) and adding a render farm (another 3x faster) I can now render the same image in just 100 hours.

Yes, my computers sputter and spasm when asked to render certain images. I have a theory about this; we explore to the limits of our patience. When we get faster computers, we never just enjoy the faster speed by doing the same kind of artwork faster; instead, we explore new things that we just didn't have the patience to deal with before. Dealing with five PCs to get one render done is not exactly my idea of convenient, but with even that render on tonight's image threatening to take thirty hours, I'm glad I have the resources available. No matter what, this art form is capable of humbling even the most bleeding-edge computers.

At least I'm not trying to render 8000x6000 flames. (Yet.)

--Damien

Rick,

In fairness to Photoshop, the reason it is so "difficult" to work with layers, doing basic things like moving, etc. is because Photoshop allows you to do a lot more directly over the image surface than UF does. UF doesn't have things like text, selection areas, and so on to worry about; every click inside the image can be treated directly as location control. PS has to have an interface with different "modes" where your clicks in the image are treated differently depending on what tool you have selected.

In fact this is why Frederik has resisted adding too much Photoshop-like stuff to UF (although I may be stuffing words into his mouth here). There's a point at which adding new capability means diluting what is already there. Sort of like how you can now buy a cell phone that has a built-in camera, MP3 player, movie player, video games, etc. and yet digital cameras, iPods, portable DVD players, Nintendo DS, etc. still sell amazingly well. They excel at their one thing; they focus. Strictly speaking, that cell phone/PDA may do all of those things "better", but in so doing they have to clutter their interface and that makes them harder to use.

Photoshop is incredibly powerful but that very power is part of what handicaps its interface. But it didn't start out that way; its interface and capabilities have grown over the years, and its users have grown with it. Back in the day, most of the funky effects you can now do so easily had to be done step-by-step using chops (channel operations). You had to be pretty smart to figure out how to get all that stuff done, and sometimes reproducing it was very difficult. (I did the coolest thing in a graphics program once, and then could never do it again. I'm so glad for near-infinite undo.)

I think it should be possible to duplicate the effect in FP's "Flood" using a UF formula, it's just that nobody has yet. But this illustrates another problem, that is shared by both PS and UF: some effects are so "heavy" that they dominate whatever image you put them into, and it's really distracting to see them. They also tend to become trite very quickly. (How many KPT Page Curls do you see nowadays? Thought so.) I like the occasional bit of water, but I wouldn't want to look at a gallery full of them.

--Damien

Mindy,

You are right in that UF doesn't have the same kinds of color controls that Photoshop has. I was actually thinking of that quite specifically when I said there were things PS could do that UF can't, and things UF can do that PS can't. And that's because the two programs are completely different.

With PS, what you're looking at is just a collection of pixel data. PS lets you select color ranges, boost contrast, adjust levels, etc. by working directly with that pixel data. The same thing goes for all of those filters; PS has the pixel data handy and it just manipulates it in whatever fashion you describe.

UF never treats your image as a collection of pixel data. It always treats your image as something generated mathematically. When you adjust the gradient in UF, you're tweaking the numbers that go into producing the image, and UF then recomputes the image. If you have a gradient with two separate blue ranges, you can tweak each independently in UF; in PS, this would be extremely difficult, as selecting "blue" pixels over the entire image would get areas covered by both ranges. On the other hand, editing color in a localized area in UF is done differently than in PS; for PS, you just select the area and edit away, but in UF, you have to create and use a mask for the area (not always easy).

If UF were to support all of PS's color control options, it would require a very confusing interface. The whole philosophy of working with UF is that you look at the final result, but UF retains separately all the pieces that went into assembling it. So all of those lovely color tweaks would have to appear in a list somewhere, for you to create, edit, and manage. There's never a concept of "just do this one action", because UF has to be able to re-create every action you do to reproduce an image. It is a difference in approach. You have more control over some things than you would in PS, but it's not a direct translation between the two programs.

Using PS's filters is problematic unless you're working directly at print size. A Gaussian blur with a radius of 10 pixels on a 1600x1200 has to be scaled to 50 pixels on an 8000x6000, as you know. But some effects flat-out won't scale properly. (It's possible to use effects like this in UF, too; I'm careful not to.)

I just this evening finished a new piece my wife really likes (it's in her favorite colors) and I've been sweating over certain effects for two days. In PS they would have been fairly straightforward, except for that pesky print-size issue. In UF they required me to write a custom formula for this image, which I hate to do generally, but now that it's done I know I can render at whatever size I want.

There is no one-size-fits-all. It's funny this discussion should pop up, since someone recently asked on the Apophysis list why anyone would want to put a flame into UF when Apo can edit the flame so much more directly. It's the same sort of thing: if your creative process flows down a particular path, using a particular set of tools, to have that option closed off is not acceptable. I would never encourage you to ditch PS, because your workflow doesn't support it. That's just not the way you create your art. The good news is, there are options. You may find in some circumstances you want to use layering in UF and export a PSD to PS for further tweaking. Or you may just want to treat PS alone as your canvas, and work from there. There's no "right" way, there's just "your" way. Sometimes there's more than one way to do the same thing, in which case one way might be more efficient than the other, but there's usually more to worry about than just efficiency.

Sorry for the rambling nature of this reply. I had to run out halfway through and cook dinner, so it might not be totally coherent. ;-)

--Damien

Mindy,

As someone who was layering fractals before Photoshop came along, I think I can answer this question. Primarily, it comes down to interactivity. Secondarily, it comes down to printability.

When I started layering I was using FractInt. I created images, saved GIFs, and then brought them into a graphics program to do layering. (It doesn't have to be Photoshop. Lots of programs support layering. For convenience, we will assume I was using Photoshop.) The problem with doing this is that if there was any error in framing, I had to go back to FractInt, re-render the image, save it, and re-import it into my composite in PS. I found the back-and-forth to be more than just annoying, it was interfering with the creative flow and discouraging me from trying certain things.

Being able to do layering directly within one's fractal application is as much a benefit as being able to do it in Photoshop is. Imagine you are making a collage of photos using Photoshop, but it doesn't support it natively--instead, you can edit a single image to your heart's content, but you have to export all your cropped, trimmed, positioned images and use a separate program to composite them. You would think this is ridiculous. Well, that's sort of how I feel about not doing the layering directly in UF. You could, but it sure can cramp your style.

The second reason for layering directly within UF is printability. Now, lots of people who explore fractals don't care one whit about print quality, and that is fine. But for me, being able to reproduce a fractal image that I created at 1600x1200 as a large 8000x6000 print is pretty darned important. If I do my layering inside UF, I can do it on the manageable 1600x1200 image and UF will automatically adjust when it renders the larger size, and it will do the compositing for me on the final render. If I use Photoshop for my compositing, then I get all of Photoshop's wonderful tools, but if I want a print version, I have to be prepared to import many 8000x6000 images into Photoshop and assemble the composite manually. Each of those layers is 137M. I have one image I actually printed that had 52 layers, which would have required over 7G of swap space in Photoshop to handle. Imagine how long that would have taken. I can imagine, and I can promise that if UF didn't handle layering internally, I would never have produced that image, simply because it would have been prohibitive to try printing it. That avenue of creativity would have been closed, by virtue of my tool selection. I can see you're actually running into this with PS already.

This is not to say that Photoshop is not valuable tool. The only people who suggest that PS is useless or has no place are either substantially unaware of what PS can do or are deliberately deluding themselves. But the advantage of being able to layer directly in your primary creative environment should not be ignored. If your primary creative environment is Photoshop, rather than UF, then it's completely reasonable that this might not be obvious.

As to more specific things, what you can do in UF that you can't do in Photoshop: you can control gradients, colors, blending in ways that you can't in PS. You have a couple of merge modes that PS doesn't (particularly HSL Addition, one of my favorites, which produces some very interesting texture effects). The reverse is also true, PS has some that UF doesn't. Moving the entire fractal and re-framing can't be done in PS, you have to go back to the fractal app.

If you're going to build a serious Photoshop rig, you want (1) lots of RAM. See if you can find a system that will take 4G of RAM, and put 4G in it. (That way, your other programs will not be competing with Photoshop for memory.) (2) You want really fast hard drives. When Photoshop runs out of RAM, it uses swap space on disk, and for that you want fast. Find a SATA (Serial ATA) system that supports striped drives (RAID 0). (3) You want the fastest CPU you can afford. Filters eat CPU for lunch. Dual-core may not help you much (it depends on the filter). Dual-core does help with UF, though, as do dual processors and render farm PCs.

Rick,

Celerons should be fine, your bottleneck with UF rendering is rarely memory throughput and almost always CPU/floating-point speed, and Celerons have that.

--Damien

Sure, I can help you. I'll follow up by email.

You can import images into XD as backgrounds, and apply various post-processing effects. Some of them are very interesting. I can't give you the exact steps, since I don't have XD installed at the moment.

I'd be pretty surprised if FB was PNL. Not only does the writing style not match, but the personality doesn't, either. It's still possible, as PNL has assumed other online identities before, but those identities always seem to grind PNL's axes, and I don't see FB doing that. So if it's PNL, I'd almost be pleased, because it would show Paul is maturing. In any case, FB must come by here from time to time, so I expect they would be highly amused by this thread.

Most of the time, the parameter files are specific to each program, and highly dependent on the features of that program. Theoretically, any fractal program with both a formula parser and the ability to control color values directly can render the same fractals as any other program, as long as the necessary formulas are available. In practice, this turns out to be very complex. Ultra Fractal, for example, can import many FractInt fractals--but not all of them, because it doesn't do everything that FractInt does. Going the other way, I've managed to render an Ultra Fractal image in FractInt, but to do so I had to write a true color simulator into the FractInt formula, and the formula was specific to the multi-layered image I was rendering. Chaos Pro can open many UF parameter and formula files, too, but this required careful programming on the part of its author. So while it is possible, it's not often done, because it's very complicated and far from perfect. For the fractal artist working with their own images, wanting to import them into new software and tweak them further, these are useful tools. For people wishing to share images, they're less useful, because unless you send an image too, you can never be sure the parameters are being interpreted correctly!

Many new video cards don't support the old DOS VESA interface that FractInt relies on to get good video modes. The 360x480 is an old "Mode X" display hack that actually works on just about every video card, because it doesn't rely on the VESA interface that's no longer supported. The FractInt discussion list had some talk about this not too long ago; the practical suggestion was to dual-boot into DOS, but that won't help if your video card isn't providing the support FractInt needs. My personal opinion is that UF is worth every penny of that $49, but I'm biased, so don't take my word for it. (smile) It's free to try for a while. --Damien

Ultra Fractal 3Apophysis

Flames are a variant of IFS fractals. Apophysis is a tool for designing flame fractals, which you can then paste into UF (since flame formulas have been written for UF). The advantage to pasting them into UF is you can layer them in UF along with other fractal formulas, and in theory render them at larger sizes.

The problem is that because of the way flame fractals are generated, the computer needs to have enough memory for the entire image. With other types of fractals, you can render an image bigger than you have memory for, because only a part of the fractal needs to fit into your computer's memory at any one time. But the flame technique (just because of how flames work) needs to fit it all in memory at once, so if you try to render a really big flame, you need lots of memory. If you have lots of flame layers, this becomes even more difficult.

This is why people say you can't render flames large enough for print. There are some other issues, such as flames looking slightly different on the large render (again, because of how flame fractals are generated), so flame prints aren't easy.

--Damien

This looks like a fun discussion.

I've noticed that my particular style has evolved over the years. What I'm looking for in a fractal now isn't the same thing I was looking for seven years ago. For example, this page contains fractals I made in 1995 and 1996. Old, old images. Back then, I was more content with basic spirals and minibrots. I don't think they're ugly now, I just don't find them quite as interesting.

But now as I look at these old images I see hints of style that I think I've developed more in recent images. Spirals deliberately placed off-center. Looking for unusual shapes and framing. A lot of these are butt-ugly, but they're interesting.

After this pleasant stroll down memory lane, I took a quick look at the artwork I did after that. I was looking for more texture in my fractal images, so I experimented with post-processing. Some of it worked, some of it didn't. I started looking at new formulas. I found more interesting stuff, and I was getting better at things that didn't look like the same old stuff, but I was still sort of restless.

Then I hit what seemed to be the jackpot: new ways to color existing fractal formulas. Oh I had so much fun; everything was new, different, interesting. But I was, in a way, back in my rut: spirals and minibrots again.

It took me a while to get out of that rut. As in, years. It's oh-so-easy to find spirals once you know where to look. If I'm not paying attention, I'll end up with a spiral. Or a minibrot cluster. Especially when I make a new tool--I want to use it, so I'm in a hurry and I make a spiral. I end up with something rather ho-hum.

I still saw glimpses of interesting things, things that don't seem to fit the normal fractal mold. Things like "Liquid" or "Fear". Things to look at that don't immediately scream "fractal" but are produced with the same tools as more obvious fractal art.

Over the past few years, images that are less obviously fractal are among my favorites. It's not that I don't like classical fractals: I still do, very much. But would you rather put "Banded Clouds" on your wall, or "Thunderhead"? I chose Thunderhead, and a large framed print hangs in my living room. Most of my visitors don't even realize it was produced with mathematics.

I wish I could show you some of the other new things I've done but my new web site isn't quite ready yet. Only a few of the images I've made since 2000 have been posted in various forums. I did post one of the most interesting here at Renderosity, called "Abandoned". That's about as unconventional a fractal image as one can get. (smile)

I guess to make a long story short: yes, I sometimes get tired of the classical shapes, especially if I've just had to look through forty variations of the same spiral. Fresh is good. Unconventional images keep our imagination active. It's easy to get into a rut and hard to get out of it, but--if we consider ourselves artists--we owe it to ourselves to pry ourselves out of our ruts and do new, original work. We shouldn't stick with derivatives and variants on our old work.

--Damien