It’s been a while since I posted shots of the actual Scout’s Journey levels. Most of what’s been on here has been test maps – small boxes to test assets, lights etc. with a fast turnaround time. The actual game levels have a turnaround time (compiling) of many hours.
I did an iteration on the terrain in the herdbase level. This was done for several reasons:
- Starting to brush up the level to get it presentable in a few months
- Testing if terrain made in Blender compiles reasonably despite its high polycount
- Seeing what basic lighting on terrain models turns out to look like
- Doing some standing rocks that are impossible to create with heightmaps
- Getting a better grip on how the terrain mesh should line up with the brushwork.
Then I let the compiler run for a long time.
The result is encouraging. It did all compile, with the terrain being lightmapped despite the fact I didn’t tick the “lightmapping” spawnflag (a q3map2 bug I’ve noticed previously.) The overall shape of the terrain is about what I wanted. There are a few things wrong still, such as the lines of the rock texture running vertically in a few places (due to really quick and dirty UV unwrapping, so this can be fixed), no texture blending (the more flat parts of the terrain should be mud and grass eventually), black spots (q3map2 bug / wrong shade angle in my Phong shader), and complete lack of vegetation (I’m still working on that, eventually there should be quite a few trees.) All of that is due to this being work-in-progress.
Another thing I’ll eventually do is create a better skybox with a slightly flatter sun angle so stuff is lit more from the side than from the top and shadows get longer. Always something left to do.
Above is a view from the building front across the helipad, the road, the warehouse and tanks toward the river. One of the longer sight axes in the area. The large angular wall behind the standing stones is where I masked out the start of the level to get it compiling faster.
View from the main entrance towards the standing stones.
View from the river bend.
View from the river’s end into the map. The road should eventually continue into a tunnel entrance to the right, that’s why there is a ditch in the terrain already. I’ll need a clever way to let the river leave the map without breaking the illusion. I’ve got an idea already. The spots are probably a q3map2 problem.
The cliffs seen from above. When the camera moves across this area (noclipping), the changing perspective and the depth of the standing stones creates a really cinematic effect (similar to parallax scrolling) as the cliff tops move across the picture. I’m very happy with this because these rocks are meant to appear in an important cutscene.
The three shots above are the result of standing on the top of the cliff and looking from left to right. Quite a large area is visible from there; this, too, is because the storyline requires a major battle between all factions to take place here. The helipad is situated quite a bit higher than the road, serving as a natural point of defense. The area where the buildings meet the sky is going to be masked by terrain and some roofs in the future. I haven’t done that yet because it would mean increased compile time, basically, and I want to do a hardware upgrade first.
This is the first impression you get of the terrain when you come from the start of the level, which is indoors. The concrete stuff is actually the fulcrum of a rotating bridge. The bridge itself got cut away here when I masked the entire first part of the level in order to shorten compiling time.
I think the water looks relatively decent now; I might scale down the surface a little because right now the waves are too big.
The entire terrain (except the standing stones) is one .ase model. Lightmaps mean that there is some nice shadowing (and vegetation would also cast shadows on it) but the big downside is collision. Ase models do not collide by default in Quake 3 BSP (which is what’s used here.) There is an auto-clipping feature in the compiler, but its results are not very good, so I’d be looking at building a clipping hull for it manually, out of clip brushes. That does kinda suck. A better alternative will be FTE’s heightmap terrain, which allows for proper collision and much easier texture blending, and the availability of shadowmaps in FTE would obsolete the lightmaps, but the realtime lights and shadowmap stuff still need a bunch of fixing before that can be done.
So much about that for now.
What else is new?
For one, I have continued to build assets, such as this:
I’m working on an automated gun turret for the game:
This will eventually require IQM fully working (there are lighting problems with that atm) and some trickery with inverse kinematics and some engine builtin to make it track the player in three dimensions.
In other news, I have finally installed some sequencer software and sample packs and begun to lay down some of Scout’s Journey‘s main musical themes. There is a Herd theme, a Goddess theme (which might become the menu music as well), and a Tribal theme so far. About 5 minutes of orchestral music.
Scout’s Journey uses an orchestral soundtrack, comparable to Tomb Raider. This means classical instruments, not electronic/techno/rock/dubstep/metal or anything. I’m still researching classical composition. I have already found a wealth of info about dynamic game soundtracks which got condensed into a new page to the SJ design document.
I hope to upload a video to introduce some of the music, probably combined with concept art and screenshots from the game, in the near future. I’ll probably write more about the software I use and about game soundtracks in general then.
FTE related news:
I’ve also seriously tried for the first time FTE’s work-in-progress shadowmap feature. I can say that the results are very promising and this might eventually enable me to use a realtime sunlight that makes vegetation etc. cast proper shadows onto everything (and makes the player cast shadows as well.) With a low amount of realtime lights visible at the same time, the shadowmaps performed relatively well even on my old hardware. I was impressed.
As you see, a proper player model is needed. I hope to have one at the end of this year. Fingers crossed. So much else to do.
Apart from that, I helped Spike fix quite a few bugs recently, which is always a good thing.
The big wheel keeps turning.