Testing irrEdit Global Illumination

Posted on:October 30 2006

As I already blogged, I implemented global illumination into irrEdit last week. Unfortunately I wasn't able to really test it out and didn't know if it really worked. So I just built a test scene using anim8or and lighted it with my irrEdit development version. The result:
The same scene rendered using direct lighting (left) and global illumination (right). This feature will be included in the next version of irrEdit.

Looks great, IMO. I wanted to test if the color bleeding worked using the red cube. Some texels on the floor are a bit more red, but I'm not sure if it really does work as I wanted it too. So I built a better test scene:
Again, a scene rendered using direct lighting (left) and global illumination (right).

Obviously, color bleeding works as expected, yay. As you can see, there are some artifacts, maybe really something Steve already suspected in a comment on the previous entry. But I'm happy it looks like this so far, hope I've some time to improve this until the release.


Great Work! :-)
2006-10-30 22:49:00

Im hoping to see a more complex scene :), this one isn't showing its potential.
2006-10-30 23:46:00

Another trick I have seen used before to get rid of those "edge artifacts" is to use the alpha channel of the resulting lightmaps as indicating whether or not there has been a front facing polygon written into it (from the light bounce). Then you simply post process the light-map textures by spreading/blurring the colour based on this alpha (i.e. ignore areas where front facing polygons have been written but spread/blur the colours for areas where backfaces have been written).

Eternl Knight
2006-10-31 00:38:00

That's a very very great work!

By the way, how much time did it take to render the second scene? Because the quality is excellent.
2006-10-31 06:15:00

Nice work, that really shows up the effect (you could have used the Cornell box for easier comparison with other results, but let's not get picky ;)

Personally I'd love to see a more detailed discussion about the techniques you're experimenting with, paper references, practical issues etc. I realise you like to keep your blog posts short and sweet (compared to rambling old me anyway ;) and that the screenshots keep the fanboys happy ;), but a more meaty discussion would be welcome, if you have the time. And not because I want to steal your ideas ;)
2006-10-31 13:03:00

Wow that was what I expected you would do: very high quality! ;-)
Great work!
2006-10-31 19:03:00

I think I'll need to optimize the renderer before I'm going to render a very complex scene, it's still quite slow. It needed about a minute for the scene with the colored cubes. But the cornell box would be an idea. :)
About the technique used: Sure, going to write something about it soon, didn't think of anybody would be interested in it. :)
2006-10-31 20:26:00

Count me as another interested party :)

I actually wish *more* developers would discuss the theory & mechanics behind the screenshots. How else are we going to build on each-others ideas? "On the shoulders of giants" only works if the giant let's you piggy-back :)

Eternl Knight
2006-11-01 00:53:00

hello niko

one question:
does the development of irrlicht engine also go on?
when will you find some free time to put some features into irrlicht like HDR, patches, community work like: CloudSceneNode, RPGCamera and so on?
2006-11-01 10:00:00

Well id love to know how it works, it looks really scary to make to me! BTW The screens looks simply amazing!
Rapchik Programm
2006-11-01 12:21:00

sanjin: other people work on irrlicht besides niko ;-)
2006-11-02 06:14:00

*bulbazaur:* Yes, but Niko is the man behind the "direction" of development. The other two primary developers are mostly bug fixers at this moment. It makes sense to ask Niko this question given that state of affairs... For example, I would ask Steve Streeting about OGRE's development, although there are more developers helping him than there are helping Niko with Irrlicht.

On the other hand, I think that Niko has said all he has to say on the matter in the thread about the same subject in the forums.

Eternl Knight
2006-11-02 23:39:00

Which is not all that much. His post there didn't give me a general idea of the direction or of the level of commitment. I really want to know, for instance, if anything is planned to do about the vertex format.

Even if nothing is planned, what kind of solution would be approved?
There have been some discussions in the forum on it about vertex format and hardware buffers. One of it has been controversly discussed. The proposer went out of his way to get it approved by the community. Good! But no comment from Niko, and only non-commitically from hybrid. Not good!
Klasker's last patch (after the big-waste-of-time discussion) wasn't even commented on at all. I didn't bother then anymore.

I already proposed something regarding this myself, and I am interested to start to design it in earnest or give input when someone else does so.
But this isn't going to happen without a broad discussion and this discussion will be moot, without comments from the team. All it needs is a bit more communication here. Two-way. Come on!
2006-11-03 00:14:00

i agree to you Saturn.
2006-11-03 10:01:00

I do also agree. I would love to get some informations on what is planned with irrlicht in the (near) future and I'm sure others would like to know that as well...
2006-11-06 15:31:00

Hm, I think I'm going to write something about the development plans soon. But as usual: There are no great plans at all, we are only adding stuff when we like and have time to, as always, as this still is really just a spare time project.
2006-11-06 17:51:00

Add comment:

Posted by:

Enter the missing letter in: "Inter?ational"




Possible Codes

Feature Code
Link [url] www.example.com [/url]
Bold [b]bold text[/b]
Quote [quote]quoted text[/quote]
Code [code]source code[/code]