DEdit tutorial video: curved surface, part 1

[xistence]

'kay, i did a small video tutorial, nothing fancy and no audio comment, just text comments in the video itself (so you might want to pause at some points). This seemed to be the best way for now, as i won't find the time to write all that stuff down, nor to make all the screenshots.
So, if at least a few are happy with that stuff, i might create some more vids (if i find the time).

Tutorial : Curved Surface 01

[TronFAQ]

Unfortunately, the video is marked as private so no one can view it at the moment. But nice to see you making a tutorial, especially in video form. I truly believe videos are the best way (like the old saying, pictures are worth a thousand words), but they do take a lot of time to plan and produce. Writing a tutorial is sometimes easier, if not always as effective.

[deeahchur]

Just embed the video in a post, X!

[xistence]

Actually i don't like embedding, as it reduces loading time of pages, and i hate waiting for stuff. I changed the private flag, so please try again.

[deeahchur]

You make that look SO EASY, X!

I seriously need to learn some of those keyboard shortcuts!

[xistence]

Hopefully this helps others to get to this point, to handle it that easily as well. And yes, shortcuts are most important for most programs. No mouse, trackball or other inputdevice can be that fast and direct as a single key press. And i recommen everyone, who didn't have done it already, to have a look at Menu:Edit>Edit Key Configuration. There are tons of functions/options which are not reachable via the UI. Lots of the most advance techniques are based on using these functions.

[Daddyo]

X = Merlin | X <> Merlin?

[deeahchur]

X = Merlin | X <> Merlin?

That depends on which interpretation of Merlin you're using . . .

[xistence]

Code: Select all

class Xistence extends Freak{
    public Xistence(Graphic g, Design d, Insanity i){
        super();
        this.init(i);
        this.setMadness(Freak.UBER);
        this.add(g, d);

        this.prepare();
    }

    public MadnessStuff run(){
        MadnessStuff ms = super.run();
        ms.randomize();
        return ms;
    }
}

Xistence X = new Xistence(new Graphic(Graphic.TRON | Graphic.TRON20 | Graphic.TRON_LEGACY | Graphic.DIU), new Design(), Insanity.ULTRA);
X.run();

....

[tronrose]

I noticed you removed the edges from the object that formally distinguished the edges on the z plains. Wicky, A NOLF2 mapper enthusiast from the Glue Factory forums encouraged to add any edge to a polygon so that every one would be edged as a triangle to, from what I assume, boost performance where the game's rendering functions wouldn't have to cut the polygons into triangles that could possibly cause some sort of polygon fragment anomaly such as texture "sparkles" from a T-junction.

However, I do know that other games, such as id Tech, doesn't concern the user to practice this method. But then again, DEdit is much different from Radiant and UnrealED as far as brushes are concerned.

[xistence]

This is exactly the point, where the mapper should know what he do. Triangulation can make some things easier, as preventing some sort of errors, but the count of polys is also a point of low performance if it is to high. Same thing, if you build polies not triangulated and containing error. If you have polies, being 'twisted', then you also giving the render a hard time.
I did 3d-render development, so i have some insights abut that stuff, so i remove edges, when i know that splitting a surface by the renderer is not:
- critical
- leading to a massive amount of polies
- the poly itself doesn't contain too many verticies
About such stuff, there are many more mistakes that could be done, just take the brush itself. At least it is recommend to put a good amount of polies together as a brush. On the other side, a brush determines for the renderer a lot of stuff about rendering order, shading and more. So, as example, i also split up or group up polies, to control the shading, as 'LightConrol' is not the only important thing about that.
Furthermore, brushed should have not several thousands of polies, but also not just a single one. Brushes should also not 'intersect' (something most beginners do, including me when i started) because this also force the renderer sometimes, to split things. So there are many unoffical rules like that.
While building a map, you have to keep a lot of information in your mind, respecting the current engine, the current level of hardware, your map concept, and a lot of more.

[deeahchur]

Tronrose, what you're talking about is "clean building", ensuring that all points connect and removing T-junctions. I would follow X's advice on this, as he frequently discusses clean building and removing as many T-junctions.



I, too, was surprised to see that X did not split up the corner faces, as I have seen from some of the roughs he recently released in his map pack. Given his explanation, however, I am put at ease to learn that there are some exceptions to ALWAYS cutting triangular brushes to connect points.



In conclusion, use your best judgement. If you believe a face is better with the triangles pre-established, then by all means do so; otherwise, leave the brush/face alone and allow the processor to determine the details.

[TronFAQ]

@tronrose: I'm guessing by "edges" that you are referring to faces. As in, a side/surface of a brush. While it's true that manually breaking up a larger face into smaller ones can help optimize your map when the processor breaks everything down into triangles, you more or less have to play it by ear.

You have to weigh the benefit of breaking a larger face into smaller ones vs. increasing polygon count, before compiling. By dilligently merging faces - thus reducing the polygon count - this too increases efficiency and optimizes your map. It's probably only worthwhile splitting a very large face, manually.

Deeahchur already said essentially the same thing, which I happen to agree with.


@X: After watching your video, I wanted to bring up a few points.

1) The text goes by very quickly. Do you think you could extend the length of each text caption a bit? Not so much for someone like me, but rather beginners who want to follow along. I'm guessing you used Youtube's annotation feature, which should make this easy.

2) I'm trying to remember: when moving one vertex to overlap another, does DEdit merge them together by deleting one of them? If DEdit does that automatically, that would be a nice feature and again increase efficiency.

3) At the end, I don't think you explain how you're joining the smaller faces together to create larger ones when you're finishing up. At least, I don't recall that. You should add a caption about using Shift+J or similar.

[Daddyo]

Code: Select all

class Xistence extends Freak{
....

After all this time I thought you were multithreading...

[xistence]

1) The text goes by very quickly. Do you think you could extend the length of each text caption a bit? Not so much for someone like me, but rather beginners who want to follow along. I'm guessing you used Youtube's annotation feature, which should make this easy.

I already thought that this could be a bit to fast, i'll will have a look. The only problem is, that isn't much time between some text segments (listening to spoken words goes quicker then reading), but i'll see what's possible.

2) I'm trying to remember: when moving one vertex to overlap another, does DEdit merge them together by deleting one of them? If DEdit does that automatically, that would be a nice feature and again increase efficiency.

In an efficient world with no bugs it should be like that. Basically verticies can be only merge if:
-part of the same brush
-exactly at the same position (which can be tricky with float val's for coordinate). The CTRL-V/CTRL-B use helps to 'match' two vertex on each other, but sometimes also not.

3) At the end, I don't think you explain how you're joining the smaller faces together to create larger ones when you're finishing up. At least, I don't recall that. You should add a caption about using Shift+J or similar.

Humm, well i'll have to look. I had some weird problem with setting one comment, maybe it was this one. Hopefully i can fix that.

After all this time I thought you were multithreading...

No, still single threaded basic (i was compiled in the 70's). I just pretend to be that way, so no one realize how old my code already is.