Havent had the chance to work on Titanic the last few days due to life But I was able to put in a bit of time tonight.

I'd say about a third of E Deck is now built, and now the brush count is at 5656.

Two more pics for the new progress, although now its somewhat more difficult to show newer sections as Im building the decks from the top down.

Worked on the Second Class staircases some more today, as well as sections of D Deck. Then compiled and made another 12 minute video giving off a quick tour of the ship thus far.

You can view it at this link: http://youtu.be/jHMj6Py6JkU

Let me know what you guys think!

More of the D Deck areas are complete now: Second Class Dining Saloon, and the area immediately aft of it.

Also started detailing both Second Class Staircases. That will be a pain to get right since measurements dont exist and I have to do best guess based on only the one or two photographs which seem to exist of the staircases.

Im going to be making another video set perhaps tomorrow. Although this time I may break it down to one video per deck or something.

Got the forward section of D Deck completed, along with the First Class Reception and First Class Dining Saloon. Skipping the galleys for now and started on the Second Class Dining Saloon.

Tonight's brush count is 4800 on the nose.

Earlier today I experimented with loading the portal file to check out the visleafs, neat stuff that is. Converted a lot of the staircases to func_detail for now, and got the numportals down further from 16000 to 10000 or thereabouts. Good enough for now, I'll do the more fine tuning on that later, but the majority of the areas that needed the fixing were the two Grand Staircases, especially the forward one since I created all the railings and posts that go along with it (and didnt func_detail the railings, just the posts)

I've only got two options here:

1) AMD Phenom II X6 2.8GHz w/4gb RAM
2) Intel Pentium M 1.6GHz w/ 1gb RAM

And option 2 isnt really an option as its my Linux server.

Its interesting though, last time I ran the full VVIS with the killbox it was consuming almost 2gb of RAM and now its consuming 400mb.

Dunno if this is an option but, you may need to find a large intel machine to compile on in the end.
Even though AMD has lots of cores now, I have noticed that they still suck when it comes to crunching lots of numbers.
I have 2 laptops. A 4 year old intel dual core and a 1 year old AMD dual core.
The amd runs faster but takes twice as long ripping movies.. Ripping books and/or converting books to MP3.

Your call though.

Spent the last few hours getting rid of the killbox. It only remains on the top outside portion of the map. Everywhere else is sealed. Plugged up about 30 or so leaks, some big but most of them around 1unit in size (variable lengths).

Last compile log shows the numportals down to 16209, which is significantly down from the previous 30473. As well, the waterindices dropped from 91.6% to 66.5%.

The FPS also went up significantly. Whereas they used to drop to around 20 FPS for the most part, they now only dropped to 100 FPS but only while at the bow and stern of the ship. Elsewhere it was back up to 300.

Huge improvements.

Late now. I'll try another full VVIS on all 6 cores and see how it fares.

I'll attempt a couple of those q's

func_detail has no i/o capability, cheap; where func_brush is capable of toggling solid, colours, visbility, rendermode etc; more expensive.

func_detail used to have a bug where light would bleed through intersections, I am unsure if this is fixed in engine updates. They cast shadows correctly as expected. I don't remember how func_brush deals with light except that it takes on a slightly different 'hue'.

idk about waterindices. I've never seen issues with it.

Models will block light depending on what their attributes are. Most static props will cast correct shadows if -staticproppolys and -staticproplighting is used at final compile. Generally you disable shadows on these because they suck, try a bush/tree and you'll see. Dynamic props will cast dynamic shadows so use sparingly. All of your ideas for model candidtaes make sense to me. Basically anything you want to have an advanced LOD over plain brushes, and be able to fade from engine rendering while still being present in the PVS.

yea I know I have serious problems in the map , for one, its one big giant killbox, about 1480 units wide, and 16000 units long, and probably 3000 units high.

For two, I hadnt really func_detail'ed much, just enough to see if it brought down my numportals, which it did.

Im going to work later on closing off unfinished areas of the ship and getting rid of the giant box. That'll for sure help.

Question: I finally opened up some of my old RMF files (It impressed me the work I had!) I noticed I had a lot of brushes as func_wall which from reading on VDC is legacy and has been replaced by func_detail and func_brush. But I cant really get a clear difference in when to use a func_detail versus a func_brush, or advantages and disadvantages of either.

I know in the HL days, func_wall brushes did not block light, is this the same case with either func_detail or func_brush?

Another question, I did a quick Google and found that "waterindices" are caused by overlapping brush work which (at least back in the HL days) equates to sloppy mapping. While I do have about 4500 brushes currently, I can say with almost absolute certainty that none of my brushes overlap. I've always been extremely meticulous about that. However, on last compile, my "waterindices" were around 98% stating "VERY FULL". What else could cause this and is this a concern?

A few questions on models:
Is there a limit on the amount of models you can have in a map, or in a certain area? Do they block light now? Do players bump on models or do you still need clip brushes? What are some examples of things to model? In HL it used to just be NPCs, weapons, sprites. Would a complex door mechanism (such as the water tight doors on the ship) be good candiates for models? What about the iron work in railings of the Grand Staircases, or the outside railings all over the ship? What else would be good candidates?

Thanks!

4 days yuck.

You have some serious geometry problems, or too much information somewhere....

Try converting everything to func_detail on a save copy and see how vvis chops it up. Its hould just be in 1024x cubes.