Hi all,
It's been a while since I posted a development update and since several
of you have joined the list after the last ever post in May I thought
I'd let you all know what I've been busy with. What follows is a list of
features that have been implemented so far. A discussion may ensue,
bearing in mind that suggestions to enhance or alter these features will
be politely ignored for now. ;-)
-----------------------------------------------------------------------
- Quark equation was added
This is a 3D "adaptation" of Dave Bollinger's magnattractor which you
read more about here:
http://www.davebollinger.com/works/magnattractor/
I named it Quark because I love Quantum Mechanics, however Magnets is
technically much more accurate. :-)
- Chua Circuit equation was added
Will not yield the best attractors you've ever seen but worth playing
with nonetheless. (Google Chua Circuit!)
- New Saturation parameter for all modes
Renders with high Gamma tend to lose their saturation. Saturation will
help alleviate this side effect.
- Anti-Alias optional for all modes
Anti-Alias is currently "always on" in Chaoscope 0.3.1. Quality will
decrease when it is switched off (particularly with Solid mode) but
render speed will increase significantly.
- Variable opacity optional for Solid Mode
Another "always on" feature in Chaoscope 0.3.1. Switching variable
Opacity (or transparency) off speeds up the render. When Anti-Alias is
on and variable Opacity is off the alpha buffer is used for perfect
anti-aliasing of attractor edges. Note that superimposition of attractor
parts or edges will not be smoothed by anti-aliasing. This is a job for
Supersampling!
- Supersampling (up to 66 samples per pixel)
This feature highly increases both the render quality and the rendering
duration. It literally multiplies the render duration by the amount of
samples you choose. It will give you "almost" perfect anti-aliasing for
every pixel. Setting it to 66 samples per pixel is equivalent to
rendering a 8000x8000 image and shrinking it back to 1000x1000. Memory
allocation stays the same regardless of the number of samples you set
the sumpersampling to which makes very large supersampled renders a
possibility. Possible values are 2, 3, 4, 8, 15, 24 and 66.
- Second gradient optional for Light and Plasma
A third "always on" feature in Chaoscope 0.3.1. Again, switching the
second gradient off speeds up the render.
- Gradients can be swapped
- New coloring methods: Matrix for IFS and Element for Quark
"Matrix" is almost identical to the standard Flame method of coloring.
- Caching of rendered attractors in RAM
This is not completely finalized but it brings a huge performance boost
for subsequent renders of the same attractor. Is also used during batch
renderings.
- Background color for Light and Plasma modes
Long standing request!
- Render now runs on two cores
Just two for now and this doesn't always mean a 50% decrease in
rendering time. We'll do benchmark to find out!
- Perspective
Still rather temperamental. Doesn't always work as expected but it will
have to before 0.4 is released.
- Focal Blur
Works alongside Perspective. Here's something I rendered a while ago:
http://www.chaoscope.org/images/focal_blur.jpg
Only compatible with Gas and Light mode. It doesn't like z-buffers. It's
not consistent yet (the whole attractor might be blurry, not helpful!).
- Color Editor
RGB or HSV, resizable. Can be left open while using the rest of the
interface.
-----------------------------------------------------------------------
What's left to do:
- Color perspective (or haze)
- Gradient editor
- Gradient browser
- Timed rendering
- Continuous rendering (only stopped pressing Escape)
- Rendering stopped when specific amount of frustum hits is reached
- Light editor (interface to position lights in Solid mode)
- Export attractor to ASCII
Will explain those in more detail when I get round to implement them.
I'm currently working on a "recovery" feature which will let users
restarting a render after a crash or a power failure. This is necessary
because of the continuous rendering: you don't want to lose your 8 days
worth of calculation, so the entire in-memory data will be saved every x
minutes.
Apologies for taking so long to put all this together. Thank you very
much to all of you for your patience!
I will post a more detailed development update on the forum.
Regards,
Nicolas Desprez
