Jump to content
Devil

Devils DeFlex

Recommended Posts

Can't say for sure and I really don't want to put a deadline on it.  I'm just going with the flow.  It will be ready when it's ready.  Not having group selection really slows things down.  As of Build 28 we all still have to copy and move each brush individually.  It's a long and very tedious process.

Share this post


Link to post
Share on other sites

Ran into more map lag and lightmap issues after doing all the green strafe pads and pedestals.  Didn't even get done with 2 lanes before the engine started puking on me again.  Here are some changes that I went through in order to get the map back to working normally again.

 

  • Completed all pads and pedestals for lane 1 (12 brushes per pad) (5 brushes per pedestal) (119 total new brushes)
  • Horrible performance. Editor failed lightmaps and lagged horribly, almost unusable. 
  • Deleted all pedestals, slime, and teleporters.  (41 brushes and 23 teleporter effects deleted)
  • Gained performance and FPS.
  • Converted every brush in the entire map to nolight texture.
  • Massive performance increase.
  • Manually changed only visible faces (shift+m) to textures, map looking like normal again with added benefit of nolight faces everywhere else.
  • Still outperformed any prior version up to this point!
  • Added  6 slime brushes, 5 water brushes, and 23 teleporter effects back in to inspect performance. 
  • Slight performance hit.  Feels like player model gained 50lbs of weight, not as agile and responsive but still playable and editor still useable.
  • Completed all pads for lane 2 (72 total new brushes). 
  • Map still working well and editable.  Everything is back in except pedestals. Performing better than it has in a long time.  I attribute that to the massive performance gain by using nolight brushes on every non-visible (to the player) brush face.
  • Added 36 small round lights.  (Ceiling above every pad)
  • Added 12 large spot lights.  (2 at the end of every lane)
  • Added 36 pointlights.  (Illumination on every pad surface)
  • Added 432 brushes to complete all strafe pads.  Completed brush work for every strafe pad.  Strafe pads are comprised of 12 brushes, 6 pads per lane, 6 lanes.  That was a lot of work.
  • Map still performs well and lightmap is building correctly!

Share this post


Link to post
Share on other sites

Strafe room is completely finished and working great.  Took the ceiling off temporarily just for this screenshot.  :)

 

Decided against pedestals just because they are somewhat unnecessary.  The map works in it's current state I don't want to break it again.  Even if something breaks in the future we'll always have this alpha8_1 version to fall back on.  It's working great and lightmap built perfectly.

 

All 35 strafe pads have 18 brushes now.  630 brushes meticulously placed today just for the strafe pads alone.  That's not counting lights, a complete nolight brush rework of every surface, and replacing effects.  Massive progress today.  Woot.

 

It's like a run version of guitar hero lol.

 

2014-11-26_00002_zps19157425.jpg

 

We tested each strafe pad and can confirm they are all possible.  Dustin was able to do the yellow pads.  They aren't easy even for veteran defraggers.  Those yellow pads are spaced apart pretty far.  So there's a challenge for any skill level.

Share this post


Link to post
Share on other sites

Still plenty of lighting left to do.  Some rooms are still unfinished as you'll see in the video. It's a good place to make another progress update video. 

 

 

A mapping music video seriously?  Yes for the amount of work I've put into this thing so far, you're damn right.

 

Lightmap has some odd flickers but at least there's no more banding or brush face issues.  Not sure what to make of the flickering.  This is with the full version of the map and a good lightmap build.  It took over an hour to build the lightmap and yes that's with every mapping optimization technique I know of already done. 

Share this post


Link to post
Share on other sites

Will add some floor color.  Appreciate the suggestion, didn't even think about redoing the floors since devs removed all the hzrd textures.  That's what we used to have as the floor texture.  Figured they would add that back in by now so we could do them again. 

 

A DNF reference is an unfair comparison.  There's a huge difference between 3 weeks and 10 YEARS.  Most of that time wasn't even our fault it was the game engine which has been fixed now.  iD software used to say that too during Q3 development.  That's where we get that from definitely not from 3D Realms.

Share this post


Link to post
Share on other sites

Now that the developers have found a way to get everything working and we've finished most other areas we can finally start adding new rooms. 

 

Oh and we added a flat gray texture to the floors.  It's almost pointless to color walls, floors, or ceilings right now.  When textures are released all of that will will have to be done again anyway.

 

New Rocket Jump Training room.  First green doesn't require a jump just a rocket to the floor.  It's pretty low.  Same skill colors as strafe pad room.  Eventually will have everything color coded appropriately for every skill level in every room.

 

2014-11-29_00001_zps3a7ee92f.jpg

 

Same kind of style as circle jump room.  Each jump is incrementally taller. 

 

2014-11-29_00002_zps0912f2ec.jpg

 

When custom textures are allowed we plan on adding decals to show in units how far or high every distance is in the map for training.  That way you have an actual value in units of what your maximums are.  This is good for knowing your limits and coming back to train in order to break your own personal records. 

 

Maximum height reported is 420u by Eldrek.  Plenty of work left to do on this particular room.

Share this post


Link to post
Share on other sites

Enlarged the circle jump room from 12 lanes to 24 lanes.  Resized each lane to be smaller so no more visual plane issue. Room used to be 4200u long now only 3900u.  Added more lanes by widening the room instead of lengthening to avoid running into the visual plane issue again.

 

2014-12-04_00001_zpsda3886a1.jpg

Share this post


Link to post
Share on other sites

A DNF reference is an unfair comparison.  There's a huge difference between 3 weeks and 10 YEARS.  Most of that time wasn't even our fault it was the game engine which has been fixed now.  iD software used to say that too during Q3 development.  That's where we get that from definitely not from 3D Realms.

 I did not mean this negatively. Take your time. Although I am a huge Duke fan, I waited patiently for DNF all the time. I trust in 3DRealms and in you.

Share this post


Link to post
Share on other sites

Please understand that development was halted for this map for nearly 2 weeks because of engine limitations.  We can finally add new brushes and make new rooms without worry about the map crashing.  This is the first time in weeks that we've actually been able to work on the map like we were doing originally.

 

I think Aeon will attest to the fact that it's hard to comprehend just how massive this map is until you get in there and can see it in person.  It was neat seeing him bewildered by the sheer size of it.  It took him 10 minutes just to walk through every room start to finish.  We want this to be a comprehensive training facility.  In that regard it's only about 1/2 done.  I did consider releasing it anyway because of impatient commentary here.  Those comments are appreciated by the way.  It's motivation for working on it more and improving features. 

 

So in an effort to keep you guys from salivating too much and give you a look in person we're going to have another open beta test tonight.  It's the best we can do right now because we're still not ready to release it. I think Dustin's been running on it a server occasionally open to the public but that's a really old version, alpha 7 I believe.  We're up to alpha 9 now and some rooms have seen major changes since then.  The current alpha 9 version will be put onto 2 servers tonight as an open beta test.

Share this post


Link to post
Share on other sites

Open Live Beta Test Announcement

Tonight from 9pm EST to 1am EST

 

If you haven't already joined the do it.  We will push a live notification to everyone in the group when the beta test begins.  Anyone in the group can ask a steam group officer to make a live notification for you and your map.  You must make the request through the group channels and not through this forum.  This ensure that only members of the group who are interested in mapping and modeling are participants.  This is good for beta test announcements and will be used for other such type of mapping community driven announcements in the future.

Share this post


Link to post
Share on other sites

We're having a lot of issues with the server handling the map.  Think it might be too big again to be played online or accept stable connections.  We've had plenty of people connect then disconnect.  Only a few people have been able to stay in for more than a few minutes.  Keep getting dropped by steam too.   This is happening on the dedicated server too.  Sorry folks.

 

We've done every possible optimization we know of.  The game developers need to fix this to support larger maps.   We don't want to cut it up into smaller separate maps because we want to build a comprehensive training map.  Doing that would go against the entire purpose of this map.

 

2014-12-04_00002_zps882c2c0c.jpg

Share this post


Link to post
Share on other sites

We're having a lot of issues with the server handling the map.  Think it might be too big again to be played online or accept stable connections.  We've had plenty of people connect then disconnect.  Only a few people have been able to stay in for more than a few minutes.  Keep getting dropped by steam too.   This is happening on the dedicated server too.  Sorry folks.

 

ce69745e15.png

 

Script kiddies hit the Steam servers for a while tonight, caused a few problems over the course of an hour or so but it was just script kiddies so no damage (real or fake) was done, possibly explains people connecting and dropping.

 

On the topic of your map and optimization, I need to step in and defend Newborn and crew. I honestly do not know how to help you with your map because quite frankly, you're trying to have your cake and eat it too. You want a big map that is big, highly detailed, and targets a lot of specific jumps. You have parts of the map coated in lights and neon while others are pitch dark to the point where I refuse to build the lightmap for your map while exploring it. You have DOZENS of pickups to alleviate the fact that a trigger hurt system hasn't been coded in yet. You have, and I quote:

 

Added 432 brushes to complete all strafe pads. Completed brush work for every strafe pad. Strafe pads are comprised of 12 brushes, 6 pads per lane, 6 lanes. That was a lot of work.

 

That's JUST strafe pads. This is fine but you are also going into hyper detail mode with brushes as well on a few of your training levels (the telejump area, for example). It's little wonder why it's taking you an hour and a half to build light maps besides the fact that you're on a low/mid-range AMD processor from 2010. As a map size comparison, for the map Benroads and I are making we have 438 brushes. Your strafe pads alone are almost equivalent to our map. I just recently put together my new tower after my old one died and I was relegated to a laptop, currently have an i5-4690k sitting in here @ 3.5ghz. I built the lightmap for your map just to see what the difference in power between your tower and mine is: it took me 10-15 minutes at texel_size 128 to build the lightmap for your map. I'm in no mood to think about building it at texel_size 64, let alone 32 or 16. Putting this altogether, you need to understand that something's gotta give when it comes to your design philosophy. You can't have your map be big AND detailed AND filled to the brim with jumps. You need to economize from a complete and utter design prospective, Devil, not just having a "let me just cut these corners" attitude. Your tower just isn't up to spec for the detail you so desperately want: I'm not only an outlier but I used a larger texel size to save lightmap build time. Just think about how many people already can't play Reflex as-is purely because of the prototype state it is in now in comparison to their specs; Newborn and company have their work cut out for them on the road ahead, and quite honestly it IS a lot of work. The current lighting solution might not even see the light of day (heh) for the Alpha build. If you're still deadset on having your map out in a respectively short timeframe, you're simply going to have to rethink your entire approach to the map. This happens with anyone doing any kind of creative work: I have to do it for papers, Ben and I had to do it for our map once or twice, my CS friends had to rethink entire project designs. You're just not going to get away with having every last thing you want considering all of the factors that go into this.

 

Good luck.

Share this post


Link to post
Share on other sites

 

You have parts of the map coated in lights and neon while others are pitch dark to the point where I refuse to build the lightmap for your map while exploring it.

No part of the map is pitch dark without a built lightmap because everything else is full bright with real time lighting.  The only way it would be pitch dark is if you tried to build the lightmap and THEN enter those unlit rooms.  Yeah all the lighting isn't done yet, never claimed it was, not for every room.  I can understand wanting to build the lm as a way to increase performance but you can't fault me for having incomplete lighting for an open map test that was meant to be done with real time lighting.

 

I'm pushing engine limitations on a continual basis.  I would think it provides the developers with invaluable information on the current limitations of the engine and something they would be thankful for.  It gives them a good metric. They are not limitations of my computer.  Dustin has a really nice computer and awesome GPU.  We cross-check issues to ensure they are map or engine related first of course.  Has nothing to do with my computer otherwise I wouldn't mention any issues that crop up.

 

I do not want to release the map early but I'm getting pressure to do it.  As one person put it they are tired of reading about updates and just want to play it already.  That was the reason for the test.  At least give people a way to play it temporarily.  There does seem to be a rush to get maps out there and I really don't want to be a part of that rush but I am getting pressured.  If I have to redirect that pressure to the developers fine.

 

All the item placements being used as a replacement for self_damage 0 is a problem for performance but there's nothing I can do about it right now and still have a playable map even if it's just for myself during play testing.  Yes I've deleted a lot of items many many times in fact.  To the point where I'm always going backward as much as forward trying to find the current sweet spot and limitations.  Every time they've raised the ceiling I've hit it again pretty quick.  I just don't think they accounted for someone making such large maps so soon.  At some point mappers will want to be extremely detailed and yes it's obviously too early for that. No one was able to prove what those limitations were.  I think I've provided an invaluable service to the developers and this community by finding those limitations and pushing the boundaries.  I've found the limitations and will continue to do so with every update until Deflex can be released at its full potential.  If that day never comes so be it, at least people will know where the ceiling is for the engine.

Share this post


Link to post
Share on other sites

On the topic of your map and optimization, I need to step in and defend Newborn and crew. I honestly do not know how to help you with your map because quite frankly, you're trying to have your cake and eat it too. You want a big map that is big, highly detailed, and targets a lot of specific jumps. You have parts of the map coated in lights and neon while others are pitch dark to the point where I refuse to build the lightmap for your map while exploring it. You have DOZENS of pickups to alleviate the fact that a trigger hurt system hasn't been coded in yet

The bottom line is, there were a dozen of defrag maps with at the least this much detail in Q3. If we want the Reflex's defrag be even half of what Q3's was then the game really needs to catch up with the creativity of the community, not the other way around. Bit too early to demand that yes, but not something the devs can avoid forever which nobody is accusing them of doing in the first place by the way.

Share this post


Link to post
Share on other sites

Thank you, exactly my point.  Q3 can handle about 3 times what Deflex is currently even with it's thousands of brushes and hundreds of lights.  Again, there's no rush! Just saying that eventually the developers will have to make more progress.  Currently I've hit the ceiling and just waiting patiently for it to be eventually raised to the same level as Q3.  That's all.  Not saying that has to happen right now but eventually.  We can all wait very patiently for little updates that get us closer to that goal.  I'm in no rush.  It's people that are pressuring me to get it released asap that are rushing.  Not that I don't appreciate their enthusiasm it's just a bit premature from my standpoint of where I want to eventually take this map fully textured.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×