Had to change hosts due to bandwidth limitations. I guess all the PNG images on the site were really chewing through my quota! If you find a link that doesn't work because of the changeover, you can send me an e-mail using the comments form in the feedback section. Thanks! :-)
Okay, so it's been a couple of months since my last update and although this period certainly hasn't been unproductive, I don't have any new images I can post yet. Firstly, my appologies to a couple of people who have e-mailed asking for release dates. The lack of news doesn't mean things aren't happening, it just means I'm too busy to work on cleaning up the code ready for release. Most of my time has been spent on research which, if all goes to plan, should begin to bear fruit quite soon.
I drafted a format for the XML file documentation. You can see two examples for the renderer and Lambertian BSDF shader. Any comments are more than welcome!
Completely re-designed the site to make it easier to maintain. Hopefully this should mean more frequent updates. Trying to push forward towards releasing the binaries for alpha testing (you may have noticed the new documentation page). Check out the Features/FAQ page for information about possible release dates, source code, testing, feedback, proposed features, and pretty much everything else. Unfortunately Igneus is NOT available for download just yet but I will be posting more information in the near future about this. In the mean time, check out Indigo, Kerkythea and Sunflow for some fine examples of home-brew renderage.
There are also a couple of new renders showing off the fluid simulation integration. You can find them in the new gallery. I would like to be able to provide Aqueus as a fluid solver extension, however this is complicated and will take time.
Expanded Aqueus into 3D ready for integration with Igneus. Check out the microsite for details, screenshots and movies of the new features. I also noticed that the Hungry Cat MLT renderer appears to have a very similar logo to Igneus; I guess flames are in fashion at the moment! ;-)
Added support for volumetric rendering using ray marching which means participating media such as fire, smoke and fog can be simulated. Both homogeneous and inhomogeneous media are allowed with volume datasets being imported from file. All this is in preparation for integration with Aqueus over the next few days. In addition, I also totally overhauled the photon mapper which was still using hacked-together code from well over a year ago. After a bit of optimisation and refinement I added full support for caustic, diffuse and volume photon maps which are compatible with all the BRDF shaders.
Unfortunately, the photon map still uses an arbitrary constant to determine the power of each individual photon. This basically means that it's trial and error to compute the correct power emitted toward each photon target. Implementing projection maps should solve all these problems.
Finished implementing a couple of new features. Firstly, black body emitters which allow the irradiance of a light source to be specified simply by temperature. In addition, the colour temperature of ordinary lights can also be specified which means that incandescent, fluorescent and daylight sources can be realistically simulated.
Secondly, I implemented Preetham, Shirley and Smits' Practical Analytic Model for Daylight. This is the same model Maxwell and Indigo use to create natural-looking outdoor scenes. Although I've got the sky component working correctly, I'm having trouble with the atmospheric filters resulting in the sun's power and colour temperature dropping far too quickly (this is why the latest image appears much too pink). Also, since Igneus currently features no kind of tone mapping, the massive difference in solar irradiance at different times of day is really making it hard to set up scenes correctly. Therefore, tone mapping is next on the agenda.
Besides working on Aqueus I've found time to make a couple of tweaks to Igneus. Firstly, all sampling methods are now fully deterministic and use quasi-random sequences to achieve greater convergence with fewer samples. Specifically, I'm using a Halton sequence with a Faure permutation applied to improve randomness. The results are a vast improvement over pure Monte Carlo methods although the irradiance cache still relies on traditional stratified sampling in order to work properly.
A while ago I broke the DOF blur code doing something else and since it sucked so badly I decided to completely rebuild it. Igneus now has a physically correct camera model. The user can now configure the film size (defaults to 35mm), f-stop, field of view and focal distance of the lens. After reading a couple of articles about Bokeh I also implemented a customisable shutter. The user specifies the number of shutter blades and Igneus renders the resulting equilateral polygon to a bitmap and uses it to constrain the outgoing rays. The effect is subtle and only noticeable on certain objects under certain lighting conditions, but it makes for some realistic renders! You can see two tests using the ubiquitous Stanford bunny HERE and HERE.
|
