Jeira is now closed. Chrisbk and I have stopped our collaboration due for the main part to our work.
I will continue my work on specific subject like mesh compression, vsd or photonmap.
I have open my own website here which mainly consist of an aggregation of web news and personnal works.

Mesh compression have been updated

Its astonishing how time goes by when you work. Last news are 3 months old.. (sic). So, here are some news before the forthcoming new year.

From chrisbk
Hi all, An interesting news : since 3 mounths I have a job. Yeah, a real job, with money and stuff ! And it just costs me 8 hours a day to earn this money ! Isn't life fantastic ? Anyway, this sucks me a lot of time and energy, so I can't keep on working on jeira as much as I would like to. When I can spare a little of both I try to improve my Kenaz-baby. Current work includes support of floating point (god, I hate this stupid x87 stack), constructor and c#-like delegate. It's sounds boring but in fact it's quite funny.

From Nithril
Mesh compression
I have continued with mesh compression during september. I have tested the positional approach for compressing vertices but angular approach gave me better and more predictable result. After a paused of 2 months I'm actually thinking of implementing an adaptative quantization precision in 2 ways:
- by giving more bits for quantization
- by having a fixed quantization precision but by mapping precision interval according to a mathematical function

3dsmax material
I have made a 3ds plugins for creating on the fly parametric material according to a definition. This definition was simply a Kenaz class with its attributes mapped on a 3dsmax material panel.

Next developpements will be mesh compression, and some 3dsmax helpers for VSD

Here are some qualities screenshots from my frog test :) (see previous news for info on size result) :
Original : Compressed :
Original : Compressed :

The compression algorithm was in a good way. There are some cases not supported like egdes owned by n (>2) faces but I'm entrusting.
For example a frog with 14940 faces and 7400 vertices gives :
176ko -> 45ko uncompressed (1:3.9)
66ko -> 16ko compressed (1:4) (7z compression)
ie. 176ko -> 16ko (1:11)

I'm back from what one may call "vacation in Germany" (that's not exactly the point, but anyway). During those pseudo-vacation I started to write our new scripting language. It's nicely call 'Kenaz', and you may find more information about it on its dedicated website.
There are still lots of think that I would like to code, but well, writing Kenaz eats all the time that I can spare....

I'm developping a lossy mesh compressor which should reduce by a great rate the size of the resulting file. The connective part was finished and now I'm attacking the geomtry. Take a look at the graphic which represents the distribution of the coded geometries coordinates. The distribution was fairly localized, which may allow a nice huffman/arithmetic compression ratio

Once upon a time in a far far kingdom, two gently knights were fighting against the evil warbug. Time passed since february 4. Chrisbk and me are coming back from a long quest where only few men are victorious. We are now in the real life, student life has ended... (sic)

Okay, Nithril keeps telling me on IRC that I should speak about my actual works. I had some kind of code-stall during those last two/three months, but I'm now getting back into codig stuff. So here we are. I'm always focused on terrain rendering, and I spent some times trying some high-quality texture generation algorithm. At the moment there is only one not-so-nice screenshot to see, but lots of feature are planned to be added. Just wait (in case you're interested). Also started to write a nice GUI for easy texture generation in c# (lots of works still needs to be done, as usual)

I guess once you started writing some scripting language, you can't do anything else than always looking to improve it. That's my major occupation at this moment. I rode some really interesting stuff here, and will try to implement a few in the languages. Improvement are : a real type system, even better integration with C++ code than the present one and, well, stuffs. I'll write more when it'll be fonctionnal (there's still a lot to do, althought ground work is already up and running.)

Started writing the VQ section of this site with some screenshot of what can be done in image-compression with Vector Quantization

Some minor updates with the design
I have put some screenshot of my work in the VSD section

I have put some screenshot of my work in the Lightmap section
More in the futur

The Scripting and VSD sections have been updated
We have put a new dynamic menu using Coolmenu

The zgja3d team wich you an Happy new year!!! ;)

I am currently developping a tool to generate lightmap.
More info in the future...

The general part of the VSD algorithm is finished and many optimisations have been achivied with very good result by using better structure and MMX instructions. I will put some result later with different scenes to show you the advantage of this technic. The main optimization to do now, is to implement an hierarchical incremental occlusion query method in order to reject more efficiently the hidden mesh.

I disabled lots of code that was too slow, and spend lots of time trying to improve the binding between C++ and ZGSL. Large parts of the ZGSL language have been redesigned, and I have to say that I am (at the moment) quite pleased with the result. Next step is to test the language in real condition. As the terrain engine that I am creating at the moment will make heavy use of vertex shader, I guess this will be a good test.

First screenshot from the procedural tree Generator. Based on "Creation and rendering of realistic tree". Until now it only feature stem split, with various parameters. Future works will include child split, leaf generation, and then Level of detail . For the trunk/branche I plan to use VIPM, and for the folliage I'll see if the FSA algorithm gives good visual results. That's going to be quite interesting