Advogato blog for id

21 Nov 2010

Sun, 21 Nov 2010 01:34:53 GMT

I have been working on these for a while after looking at that SNAP
thing I decided to put up this premature announcement.

First, I am working on a framework called AP3, which abbreviates
nothing but stands for ape coz I love monkey.

AP3 is a 3D application development framework which currently included
a basic scene graph, a sample OpenGL view to display 3D information
currently just some meshes, but all the goals are

Data model, built around COLLADA, has a quite complete and working COLLADA parser which I barely use, however.

Basic object or view, heavily inspired by Blender's internal data structure (unnecessarily), these basic view objects are in so many way like NSView, just in 3D. The model data above can be shared and exchanged and ID'ed between basic view objects (Think of duplicate linked, if you know Blender). They might have a single parent and children objects just like NSView... and any scene graph system. The APIs will also try to imitate NSView's, but with additional Z axis, having 3D ray "hitTest" to map the mouse action, well you get the picture.

Animation system, time-based with graphs, FK/IK

Rendering engine backend(s), quite like GNUstep's rendering backend but more visible and can have more than 1 of them working at the same time. The rendering engine backends can be either used for shading 3D data model which attached shading language (eg. GLSL or Renderman) over an NSOpenGLView or invoking other photo realistic renderer, the first idea is to support Aqsis renderer. The default real time engine should be extensible enough for general purpose rendering but it will not define a strict display API like GNUstep's display backend.

2D raster engine. This engine is a rewrite of my own maliwan project. It can be used to implement procedural texture generation also a GIMP clone if you want to make one. The data model was built around lcms library which dealt with ICC color profile management. Mainly the old node system will be rewritten since I think I have a better design for that based on DAG'ed invocations (which will also be used in the procedural 3D modeling)

This project can be described as a fork of GNUstep 3DKit but didn't
share the code (not anymore, I think).

The second project, I called this one 3Σ, which will utilize the AP3 framework, is just like GORM but in 3D though has way cooler name. It will include many 3D modeling functionalities as I want to make this one a Houdini killer and ... I can just dream about that for now, the first phase would just be able to use it to layout the scene and linking methods and properties' relationships and generate a GORM file?!?!

The website of the project is at http://mus3.sourceforge.net/ which currently has no source code pool (or it did? must be pretty obsolete). I'm managing the code that some part was GPL'ed but the project should be at least LGPL so this will take some time to make a release or put it in public. Beside that, most parts need a clean up and I am migrating clean source (now 2 files) into my local git and will have it on-line ASAP.
If anyone is interested and want to join the development you can write me.

Links

http://mus3.sourceforge.net

http://qstx.blogspot.com/<wbr>search/label/3Σ

26 Dec 2006

Tue, 26 Dec 2006 20:57:38 GMT

Today I decided to quit my activity on GNUstep project due to a political reason. But then since Objective-C has became a large part in my life, it wouldn't make much sense if I have to quit coding in it. I'm still deciding between looking forward to Cocotron project as a new foundation for my stuffs or if I should just fork GNUstep without a feeling of stealing stuffs from people I really disagreed with. In either ways, I'll still be contributing to Cosmetic as long as other developers are still working on it.

5 Nov 2006

Sun, 5 Nov 2006 10:29:04 GMT

I am getting involved with a new project named Cosmetic in a way. Not much there left for me but designing. Some old/some new garma developments are voluntarily going on by "the others". Please contact them if you are interested.

16 Feb 2006

Thu, 16 Feb 2006 15:54:48 GMT

Big Big Thanks to Steppy for helping me with GSWeb! After spending 6 months trying to get such thing to work, a single line modification in the config file did the job!

7 Nov 2005

Mon, 7 Nov 2005 21:13:05 GMT

Thinking of starting an OpenGL backend for GNUstep. This is for an experiment that I'll write a new 3D view, GS3DView. With this GS3DView, I should be able to map an NSView right into a GS3DView. So I can have all dialogs and buttons with perspective representations in an OpenGL-based view. Yes, I'm thinking of Open Croquet. This will be done after I write a backend that run as a separated server.

6 Nov 2005

Sun, 6 Nov 2005 15:01:18 GMT

New project. IK, Illuminataions & Kinematics (optionally Illusion Kit) soon to be a framework built around Maliwan and COLLADA . IK, which can be used as generic 3D engine (and may invoke external renderers) is being written in Objective-C for GNUstep and Cocoa, as usual. This project has educational purpose. Hopefully it can be used to teach students about how 3D works in future. If I am lucky enough, may be I could even finish an all-in-one application that can mix raster, vector & 3d mesh editing. But that is not the current goal.

3 Oct 2005

Mon, 3 Oct 2005 02:28:46 GMT

Didn't write here for sometime. I took a job as a sculptor recently. It is interesting to work seriously on something I have not touched for 15 years. Quite a surprise that I am still get along well with plasticine. I'm also working as a director of a small game. No coding since all the proprietary softwares are evil. But I'm thinking of making my own version of the game based on Quake 3 engine. (The original version was designed for cellphone)

17 Jun 2005

Fri, 17 Jun 2005 16:23:03 GMT

NotingTutor

Spend half an hour to write this totally worthless app (shot). Well, the notes were streaming in random and you just type c d e f g a b to clear them.

DiagramKit

Now I guess I already waste enough time on craps. Hope I can start porting my DiagramKit (from GNUstep) to OSX this Sunday so I can continue my development.

Maliwan

A side note here that Maliwan is going to be redesigned, from DAG to changeset model. This involves eliminating ivar from factories and intensively exploiting command patterns in Objective-C with invocations. You don't know what I mean? Neither do I! Ha!

16 Jun 2005

Thu, 16 Jun 2005 18:54:41 GMT

Shhh

Commit new patch to make interface more scalable. I guess I need a way to control ratio of a subview's geometry to its parent.

Wesnoth

It's kinda fun but it seems it's quite buggy (on Ubuntu).

Money

Enough to do a living. Not enough to travel. May be I should hunt for them from GNUstep's potentially future bounties. I'd survive a month here with $50. Or may be I should do icons?!?!? :(

30 May 2005

Mon, 30 May 2005 17:35:40 GMT

Blog

I've started my Thai language blog called "..transid34plethora". Today I added an article on free software and a portrait of RMS

GNU Arch

I guess I am really keen on using GNU products. Beside a GNU Hurd fanboy, is this yet another attemp to be proving myself a GNU zealot? Yes, this is the 30th time I try to deploy tla, and I am considering writing a Cocoa/GNUstep frontend for it. (Well, what actually attracted me was the file-id thingy.)

Mashima

I revisited the idea with a friend (since we were in accounting school) 2 days ago. I was showing him what quantrix can do and I was thinking, it would be great if we can have an underneath framework that implement the quantrix's idea which serves our double entry accounting application. May be I should take a look at Flexisheet, but I have a feeling like rewriting it from scratch is a better approach, in case that flexisheet isn't maintained. Looking in their CVS makes me that impression.