If you’re a fractal artist, you might be aware of a few popular programs for fractal art, including JWildfire and the up-and-coming Chaotica. JWildfire is almost a fractal creation suite, an all-in-one package that allows you to not only make fractals, but watch them dance to music and make videos with them. Chaotica, on the other hand, can’t do much functionally but is an excellent program for rendering… assuming you can figure out how to get the thing to work for you.
At first, I didn’t take Chaotica very seriously. I could see that it had nice smooth lines, but eventually I noticed that its rendering resulted in much stronger colors that other programs (the exception being UltraFractal).
By luck, I happened to find someone willing to share their parameters from Apophysis that were rendered in Chaotica and also worked in JWildfire, so the sample is even. The image is called “Dawn”, and it’s by 2Plus2Equals on DeviantArt.
Beautiful, no? Thick, vibrant colors, smooth lines, all composing a solid, space-filling fractal.
Now let’s have a look at the JWildfire raw render. This is what you get when rendering high quality in the main editor. It is also the only way to add anti-aliasing. For this fractal, there is no X-form anti-aliasing, but you can see that the engine adds some anyways.
NOTE: You may have noticed that the bright glare on the right doesn’t show up. JWildfire is probably missing the plugin that makes that, and so the test isn’t perfectly even, but it’s absence shouldn’t effect the demonstration of my point.
In order to improve the quality of the fractal, we can raise the alpha channel values using Gimp, the free image editing software.
The curve I used:
As you can see, the alpha value change isn’t as simple as doubling the alpha value. The fractal image is enhanced because now we can see the details of the fractal without it looking weird. This is where I often stop modifying since it’s the often times the best I can get out of a JWildfire fractal.
JWildfire also has the option of allowing us to render continuously using the Interactive Renderer. The result improves on the detail, but you lose the anti-aliasing effect.
As you can see, the result is rather grainy.
To be fair, I did render this fractal using the free version of Chaotica and got a different result, which I present here. Clearly the original artist did some modification, probably using the tone map editing system of Chaotica.
Even still, the texture and lines are very smooth. It is my assumption this is because Chaotica performs anti-aliasing every render cycle.
The graininess of JWildfire shows up even in high quality renders. This may have to do with the sampling algorithm in JWildfire and other programs that use Scott Drave’s flam3 algorithm.
Notably, my comparison wasn’t even. Notice that the Chaotica render had a bright light in the right side that probably influenced everything else. However, it doesn’t diminish the fact the JWildfire, amongst other similar programs, doesn’t have as refined of a rendering system. In general, since fractal programs are becoming more distinct in the types of fractals they are used to make, it is easier to tell what program was used to make a fractal. JWildfire fractals tend to have a kind of ghostly appearance, similar to Apophysis.
Since JWildfire is open-source, I intend to improve the rendering system, but like all good intentions … we’ll see. I’m still figuring out the program at this time.
Jan 7, 2014
Okay, so Andreas Maschke was kind enough to find the missing plugin and remake it for JWildfire. I tried it out, and while it makes the image look similar with a high-definition render (shown below), I need to do another render with anti-aliasing to see if the spots go away. I doubt it – they didn’t last time.
Also, it still isn’t as vibrant in color as I’d like. To fix this requires rendering using the interactive renderer, but the results tend to be alittle grainier. I intend to try it out later and see if that fixes the problem.
Jan 18, 2014
Alright, I intended to get around to using the interactive renderer, but after tweaking with JWildfire and having little success at getting a smooth render, I turned to Gimp and discovered Selective Gaussian. It’s tricky at first, but I found settings for one size fits all.
- Create a high-definition render in JWildfire with standard anti-aliasing.
- Import into Gimp.
- Double the image and merge. This doubles the color values and alpha so we don’t have to use the Curves tool, although you can if you want.
- Apply a Selective Gaussian Blur with the settings: Radius = 10 and Max Delta = 10
I would say that the results are better than they were before – I’m getting the smoothing that I want with the details being preserved and the color being enhanced alittle. Now granted, I don’t have great anti-aliasing, but there’s some so it’s not a big deal. There is still a lack of vibrancy as in the original from Chaotica.
I did use Hue/Saturation change in Gimp to add vibrancy, but it brought out the background blue that wasn’t in the original. This, of course, has to do with the fractal and not merely the background color.
Jan 22, 2015
Recently, I sat down and took the time to match up JWildfire rendering with Chaotica and I managed to accomplish what I wanted (for the most part). Granted, there is no walking around JWildfire’s long rendering process in the Interactive Renderer, but the coloring and quality can be matched (at least fairly closely).
To do it requires working with the Coloring Tab in JWildfire. I found that, in this, the ideal Gamma is between 2 and 2.2 (usually the latter). The Gamma Threshold doesn’t usually matter, but anything between 0.0025 and 0.010 will work. The brightness needs to be adjusted on a per-fractal basis. For “Dawn”, it turned out to be as high as 10, but in most of my fractals, it tends to be between 0.9 to 1.5. Also, I now set my anti-alias effect to 0.3 instead of the default 0.75. This creates more defined edges but they are alittle jagged.
Sorry, no test render for “Dawn” using this, although from what I saw in the quick rendering, I got it to look identical in color strength. Since JWildfire’s point-selection is random, it tends to take longer than Chaotica to smooth out its rendering, but if you leave your fractals rendering for a few hours, the image should turn out fine.
Below is a different example using JWildfire. It is only a low-quality / partial render at the moment:
That said, every fractal flame is unique, and you will probably need to toy with your fractals to get the best appearance. Some of my fractal flames look like garbage at a Gamma of 2.2 or lower, and some require a brightness of 10 or higher. While these may be rare, it all depends on the composition of the fractal.
Quick tip: Work at the default settings in JWildfire and you will be able to see in the quick view how things look. Only once you are done should you change the gamma and brightness settings. If you don’t wait, you’ll find yourself working in the dark quite a bit.
Jan 19, 2015
Interesting, I seem to update this post every January. hm…
Not too long ago, I did some testing again with JWildfire vs. Chaotica. If you’ve been keeping up, you’re probably aware that the brightness settings of JWildfire have been “fixed”, making brightness relatively independent of gamma. That’s great news! … except that fractals with old gamma settings don’t look the same, and thus more tweaking is required.
While the standard gamma setting of 2.2 is still the “ideal” setting, there are other ways gamma settings can (and sometimes should) be adjusted in JWildfire. On the right side of the main editor interface is the transforms area. Of the four tabs, you want the one that says “Gamma”. Within this tab, you can edit the gamma settings for all points that pass through that specific x-form. The top setting adjusts the actual gamma (and should just say “Gamma”) and the second setting should say “Gamma Speed” or “Gamma …”. The gamma speed is an indicator, like the Color speed, that says how much the points are to be like other points (or essentially, how much this X-form is allowed to modify their gamma/color). At its maximum value, the points will inherit the same gamma as other transforms. At minimum value, they will have the unique gamma, as given by the gamma setting on this tab. And you can probably imagine what happens if the value is in the middle. Use this tab to enhance the gamma (more precisely, lower the gamma) of transforms that dominate the pattern. For example, in the case of Julians, its the Julia/Julian transform whose gamma should be enhance. Don’t set the gamma speed to -1 though – In most cases, this makes things too bright.
Phew! That said, it is possible to make fine tune adjustments. These settings in JWildfire have been available for some time now, and I have them in JWF 2.57 and I believe 2.34.
Anyways… I performed a comparison between JWildfire and Chaotica using different fractals. It is interesting to note that, in a number of cases, the results are roughly even, in some cases, my tinkering in JWF gets better results than the default in Chaotica, and then in a number of other tests, Chaotica has unmatchable results.
Perhaps why Chaotica can get unmatchable results can be demonstrated with the usage of two particular variations (your choice): radial blur and rays. Create a structured fractal (one that does not overlap itself in theory) and try using one of these two variations in it. The results in JWF can be messy, especially in the case of a “spherical”-variation final shape (and if I remember, I may upload an image for that soon). First, in JWF, the fractal starts to overlap itself, so you end up seeing a kind of cloud over parts. In Chaotica, there is no such cloud. Second, observe the smoothness in quality of the radial blur and rays. In Chaotica, these are crisp and clean, and the radial blur is even “extended”, as it were, into areas it might otherwise not reach under normal processing in Apophysis or JWF. To get a crisp result in JWF requires sacrificing the anti-aliasing (which is technically a randomized blurring) and increasing the spacial filter radius (for spacial oversampling) on the Anti-Aliasing tab in JWF. That does come with a cost to rendering, but perhaps you can get better results. A radius of >= 1 (the default is 0.75) seemed to get decent results with not AA. However, at the current time, it isn’t possible to mimic the crisp look of the rays without recoding the variation itself, which I may do in my spare time. All of the current rays variations, which the exception of “rays”, can be tricky to work with, much less get something cool on a single pass. Furthermore, getting the “extending” of the radial blur may not be as easy as the rays.
My guess is that Chaotica probably does some behind-the-scenes modifying to the overall look of the fractal without having to change the variations quite as much. While in many cases, the results are superb, from time to time, you’ll have to play with the settings to get what you want. The default seems to be a moderately sharp gamma, many times (but definitely not always) similar to the 2.2 value of gamma in JWF, but as experience will tell you, not all fractals look good with bright edges, though most look good in basic black. 😄