Q3Map3 - Discussion and Development

[skinNCNmaster]

yeh i figure the lightmap version is a blck and white textured solid map.

id like an alpha version of that, the whites, with transparents for the blacks

ps check out the crystal grove in wizard101, they way theyve done crystals is doable in q3

you can use my account to view anywhere in wizard101 for inspiration if youre interested.. i will pmya

[Silicone_Milk]

After thinking about this for a couple days, I'm replacing the python/c code with boost's solution so I end up having less things to deal with / worry about.

boost::python keeps complaining to me though. Apparently I'm a slacker and never updated to the newer service packs that the library craves so much.

Updating my system and then we'll see if I can't get something out there by the end of the weekend.

EDIT: On a semi-related thought, I'm wondering if perhaps this thread should be tossed in to Programming Discussion

[deqer]

developed as a set of modules (plugins).

Careful! You're going down that road again, where you might be going more work than you have to, at this point.
You're going above and beyond, I feel. Enhancing the structure and framework of the tool. But, maybe that is what might burn you out again, thus not finishing the project.

I'm all for the "above and beyond" approach, but I end up burning myself out as well when I knew that the extra work wasn't really necessary, and I wasn't being paid for it.

Perhaps it's best that you initially take the quickest and dirtiest approach; ignoring the whole plug-in system.

I have to admit though, the plug-in system sounds awesome.

[obsidian]

deqer, agreed.

1) Get it working.
2) Add extra cool stuff.

You're still on step one.

[Silicone_Milk]

Hm? But it's already working! See q3map2

Remember, OpenCL is the primary goal here. My route to use it is through the Python/OpenCL bindings. So, to get what I want working actually working, the stuff I need to do will naturally also provide the foundation for plug-in functionality.

But for any of my goals to happen, I need Python embedded and functioning correctly. I had this working with the Python/C api but there was a lot of other busy-work involved to ensure everything was playing nice. I don't want to bother with that so I'm replacing it with the boost::python library.

[Grenader]

Not too long ago BSP compiling on the GPU had already been discussed but (afaik) it was dropped due to GPU's not being accurate enough (floats).

Load of horse radish or a real issue? I mean, it's not as if BSP takes a huge amount of time (Unless you have _alot_of models).

[obsidian]

GPU's are primarily a very specialized floating point processor. They should be ideal for this kind of processing, actually. You are right that BSP doesn't take a long time, nor does VIS. Light calculations on the other hand can take a while.

[Silicone_Milk]

Yep, I had mentioned in an earlier post near the top of the thread that the lighting stage would benefit the most out of this.

[fKd]

cant wait for this to get released! good work sir!

[^misantropia^]

C'mon SM, kick that source onto GitHub or Google Code. Sharing is caring, many eyes make bugs shallow and all that jazz. Plus, you might pick up a collaborator or two. If you worry about never finishing a project, nothing helps as well as not having to do everything yourself. Voice of experience speaking here.

[Silicone_Milk]

I will just be a little more patient

Things have come up in regards to job hunting that have left me unable to finish what I want to get done before posting the source.

Once I finish that stuff up it'll definitely get posted.

[fKd]

*bump

any news?

[Silicone_Milk]

*ahem*

Yes yes I haven't forgotten about this.

Shit had gone down that the Mavericks competitors are aware of that kind of threw everything off schedule significantly.

I just haven't had a ton of time lately to do much of anything. And the time I have had I've put towards learning to make 3D models.

[Eraser]

Ever since late March you've been saying you haven't had much time :-(

[Silicone_Milk]

Yeah well things like losing your job, living out of your car for a little while, then getting a new job and putting yourself in debt to go back to college to get a degree really makes it a bit difficult to sit down and focus on a hobby programming project :P

[Anthem]

Yeah well things like losing your job, living out of your car for a little while, then getting a new job and putting yourself in debt to go back to college to get a degree really makes it a bit difficult to sit down and focus on a hobby programming project :P

Indeed, bro. We all understand the situation. Take as much time as you need.

[^misantropia^]

What I don't understand is why you don't put the code you have so far on GitHub or Google Code or something.

[monaster]

@Silicone_Milk: My sympathies. Just put some badass boot virus in you next release to share the pain. No, seriously, life can only get better from that point, cheer up!

[Silicone_Milk]

I'm starting work on this again. That is all.

[Eraser]

\o/

[Silicone_Milk]

Made some headway today that I decided to share before heading off to work.

Here's a quick rundown of what went down after picking this project back up:
- Was rewriting the entire source in C++. Realized how retarded it was to heap up so much work for little return value. So, started from square 1 with the original q3map2 source.
- Got a git repo set up. It's private right now.
- Was able to compile the source (gtkradiant-1.6) on linux but not windows for a while. Worked on getting the source compiling on both windows 7 and linux. They're both working now.
- Fixed a buffer overflow issue I was seeing in the release build but not the debug.
- Python support is no longer working since I started from scratch. It will be added. But later.
- Not using any bindings for OpenCL. Working directly with it in C. Host code is ready for kernels to be written and executed now

Obligatory screenshot:
[lvlshot]http://www.siliconemilk.com/music/real.jpg[/lvlshot]

[fKd]

oh hell yeah! gl bro, geet er done

[Eraser]

Awesome man. I'm really anxious to see some compile time comparisons when this is done

[Silicone_Milk]

Thanks guys.

Wasn't feelin' it today but I did manage to get a little more work done.

Wrote some helper functions to load opencl functions and to deal with opencl memory buffers.

I didn't realize how much I actually liked C++ until coming back to C

[Silicone_Milk]

I've had a massive headache for the past couple of days. Not sure what's up with that but I didn't get done nearly as much as I had hoped before the new school term began.

Anyways, this thread needs another image.

Just making sure everything's loading fine before I compile a test program to run on an opencl device. What you see here is just a couple .cl files that have been loaded in to a dynamic array being printed out (just checking to make sure my code wasn't being stupid).

[lvlshot]http://www.siliconemilk.com/music/update.jpg[/lvlshot]