de_cathedral (CS:GO)

Started by Joris Ceoen, March 04, 2014, 04:48:40 PM

Previous topic - Next topic

Joris Ceoen

Quote from: func_Mathias on May 24, 2014, 06:09:52 PM
Wow that's a pretty darn good job! Too bad the lighting doesn't keep up :-/
Thanks for your kind comment!

I have fixed most of these shadow problems by now. The problem was mainly the window frames just before the glass that casted shadows which where too complicated for the engines to cast on nearby static props (such as the ceilings). By disabling the shadows it looks much better. I'm also going to add lights especially on the lower level of the cathedral which should instantly deal with shadow problems (as the shadows will be corrected by the light rays).

Joris Ceoen


New picture of the finished outside, at this side of which most of the playground will take place.

Joris Ceoen

More screenshots from the 3D Skybox as it is finished:




I can safely say that this was only achieveable with 3DS Max and Wall Worm Model Tools. The way I was able to match the 3D Skybox and Hammer in the most delicate and 'fragile' situations is blatantly mindblowing and I would never be able to optimize the map by nature in Hammer. I mean, I haven't yet applied a single optimization process (as the environments are in the works) and the map runs at a steady 280fps in and outside.

My initial idea of seperating the the cathedral in 2 (a clean cut to seperate left, right, front and back from each other when walking through the map) and still conserving the total picture as if you can see the whole cathedral from everywhere has become a reality whereas previously it was just a dream! And, even more, now when walking around the cathedral, it nicely culls away all inside geometry from the outside as well as the opposite side geometry (as I predicted). This means that even though the map currently consists of around 1.2 million polygons in total, at most only 300k get rendered at the same time excluding all of the incoming geometry as well.

That is to say, it runs at maximum speed for me so I have lots of extra reserve to add fancy stuff!

Now, I only have 8 days left, so 8 days are now going to be spend on finishing the map, making it playable and get it running for the CEVO contest. That means that new screens will only be added probably AFTER finishing the whole thing.

That being said, once the map is entered and polished, I will be starting all of the tutorials for WMMT that I wish to share with everyone, via Youtube and articles.

wallworm

Looks awesome Joris.

Two things:

1) There is still some misalignment of the roof tiles (as seen in the first image).

2) I am curious: is the polygon count in Max or Hammer? I'm guessing that if these are Max stats, they are incorrect because the scene calculations will include anything in scene (like hulls, entity nodes, etc)... and also, if you are referring to Max's Polygon count, it doesn't match the in-game stat of Tri-count. I'm just curious.

Joris Ceoen

Quote from: wallworm on June 01, 2014, 02:18:03 PM
Looks awesome Joris.

Two things:

1) There is still some misalignment of the roof tiles (as seen in the first image).

2) I am curious: is the polygon count in Max or Hammer? I'm guessing that if these are Max stats, they are incorrect because the scene calculations will include anything in scene (like hulls, entity nodes, etc)... and also, if you are referring to Max's Polygon count, it doesn't match the in-game stat of Tri-count. I'm just curious.
I said polygons again... I ment triangles. Selecting everything gives a triangle count of around 1.3 mill. It includes all of those things indeed so it's incorrect, but should be close to the real number I think.

Yea, the roof tiles are correct in 3DS Max, but on export are getting lost. I have no idea why this happens, and I already applied an Reset Xform modifier to it.

wallworm

Send me a max file with just those objects and I will look at them (I'm assuming it's world geometry?). If a model, they probably need to be rexported. (You can select one piece and then hit File > Save Selected for a new scene. Then open that scene and Archive it.) I'll see why the UVs aren't exporting correctly.

Joris Ceoen

Quote from: wallworm on June 01, 2014, 02:24:23 PM
Send me a max file with just those objects and I will look at them (I'm assuming it's world geometry?). If a model, they probably need to be rexported. (You can select one piece and then hit File > Save Selected for a new scene. Then open that scene and Archive it.) I'll see why the UVs aren't exporting correctly.
It's world geometry (Those are the func_details in the skybox.). I'll send it to you!

Joris Ceoen

Currently I am working hard on Aperture Tag (http://store.steampowered.com/app/280740/) and as such I have no time until the end of this week to catch up on the cathedral. However for future purposes, and because I currently don't feel like playing anything or even mapping, I will enlight on the future plans of the cathedral:

1) Making a new layout

One of the biggest complaints that the current version of the map received was that the layout was just not good and didn't make sense. 2 bomsites being outside the cathedral while it would be more obvious to bomb the cathedral itself. There were also big gaps at crucial door openings that were so large that a smoke couldn't even cover both sides of those openings. Mid is the quickest meeting point but yet takes so long to reach. The bombsites themselves are far away from both teams and rotation times were simply impossible. This made me realise that I'll have to change the speed of the general distances by moving one of the bombsites inside the cathedral (placed at the center where CT's spawn.). This will put both bombsites closer to each other and make rotation times a very possible solution.

For me it seemed logical to scrap the fountain bombsite, it had no meaning at all and was making the map too large to handle. So now the path left from the cathedral that initially led to the fountain bombsite will be used to flank from the left to get faster to the inside bombsite! I will also make sure that the opening leading to that bombsite will come earlier because otherwise it's the same problem for being a very long traject just to reach inside the cathedral. This will also get me rid of the unnecessary routes I have been putting there.


2) Changing the general scale

I find it hard but not detrimental that the map has this big scale, making you feel like a 'rat' (being in a rats map). The only way to change this is to make the players feel that they are appropriately walking in an environment that with the eye makes you feel comfortable, so that the only thing that is huge, is the cathedral. I generated this feeling in players because of the texture scale and the giant stairs in the cathedral which have to be jumped in order to pass. Honestly said, that is because it's really hard when you're mapping in 3DS Max to judge the scale from players. I had in fact measured the player model in Hammer to put and reference them in 3DS Max, but the view seems really different. It looks like you're mapping something that is actually on scale for players, but once inside the map it looks huge!

This is especially the case for textures. The bricks on the floor could easily fit 2-3 CT's and the bricks on the walls cover almost 2 CT's! Not to mention that many textures had different scales which I didn't have the time to fix before the deadline. This is something to get used to, and for future maps I'm pretty sure the same problem won't happen again now that I know what to do. To get rid of the huge feeling of the cathedral, I will be putting objects in such a way that they convey an interesting layout inside the cathedral, give a sense of direction, a reason to be bombed and generally give you the feeling that all the space in the cathedral is not just there to be (as now it's pretty much empty).


3) Testing!!!!

I have been stepping into the same error for about 5 times now. ALL of my maps went through one-way idea passages and were actually never tested (Himeji was but only after this one-way process). I will stop starting like that. After changing the texture scales and having the new layout done, I will throw it on a server and ask many people to test. I know plenty of communities with servers that will happily put my testversions on them and test for feedback. After gathering all feedback, I will adapt, compile and test again. Again again again until I get a layout that works, that is generally accepted by almost everyone and that plays well. I'm not afraid to 'lose' that time anymore as I realise it just takes that much time and effort to come at such a stage.

4) Adding custom soundscapes

With WWMT I can now realistically option for custom soundscapes. I will be taking a lot of time to learn making the right sounds and give you a feeling as if you're really being in a cathedral-like environment. I will also be playing an organ piece on request of Shawn which is an amazing idea!


5) Optimizing, finshing up errors, adding cubemaps & optimised .nav file

The final step will be all of the above, which doesn't need explanation I think as everyone knows what it means, but I still want to enlight on optimization:
This was one of the more positive comments on the map. I'm happy people realised that the map is actually very well optimized, despite the high polycount it suggest at the unfinished/unoptimized stage it currently lives. That is because of 3D Skybox which I could only have realised with WWMT. One side of the cathedral can never get rendered, and as such completely cull all of those objects. Yet, from every POV you'll never be able actually see this process happening because the map is literally cut in 2 (horizontally) and as such makes an interesting, natural optimization. With natural I mean without having a single manual optimization. I could compile the map, not optimize, and it would still run at almost 300fps at any spot (I think the lowest I had, with 3DS Max and Hammer open at the same time, 250).

It doesn't mean I will do that, because I definitly spot some areas in the map that could have been easily optimized and weren't yet. A few hint/skip and areaportals will quickly get rid of those.


That's the big plan for this map. In the meantime, I'm about to start writing down my own experiences/discoveries on WWMT and see to publish them on mapping communities and keeping them up-to-date with the explanations. Much like FMPone did with de_crown which had quite a lot of succes and generated much interest!

wallworm

Great info Joris.

When I first started mapping in Max, I too felt a little bit awkward about scale. But this was probably just a familiarity issue as well as FOV. (But if I recall, the first steps I took into Hammer also had scale problems... and it took some experimenting to get the right feel.) Luckily there are controls in Max to adjust the FOV for the perspective views.

SMF spam blocked by CleanTalk