The Last "Two-Level Demo" For Release Raptor!

[x4000]

Original: http://arcengames.com/the-last-two-level-demo-for-release-raptor/

...and I just realized I forgot to update the main menu text to reflect that.  Bah!  Always something. ;)

Well, as you can see from the release notes, we're getting darn close to being ready.

• We have a variety of new enemies (though there are lots more coming -- it was hard to tear myself away from making more right away, because it is super fun to do).


• The voice acting for the robots is now in place, for those robots that are in place thus far.  We took note from Bionic Dues and have been sure to have a really huge amount of voice cues for every robot so that they don't get too repetitive.  There are also a larger number of less-funny ones in there so that it's not too funny all the time, which gets tiring when that's the case.


• Among the new robots, we have our first flying robots, our first robots that divide, and the first ones that use ballistic trajectory shots.


• The procedural map generation code is now done, thanks to Keith.  There are likely still some glitches with it, but this is based around hand-crafted content being procedurally assembled, so that does help.  The system is really powerful and lets me design level flows with a really high degree of specifity so that the layout makes sense.


• The occlusion system has had a huge revamp as well, to aid in performance in general, and to prep for the procgen levels.  This is the major thing holding back procgen levels at the moment.


• Father Brain and the level flow that surrounds him has also not been implemented yet, and that's another key part of the feel of procgen levels.  He's actually one of the easier robots, though, so that's good.  And the stealth bits that play into his main mechanic have been done for a while.

So, are we on schedule for our release to EA on Monday?

At the moment I'd have to say... yes!  There are some things I still have to get worked out over the weekend, to be sure, mainly to do with the occlusion system.  But my hope is to have the procgen levels in the hands of press and testers tomorrow (well, today; it's 3am).  Anyhow, Saturday sometime.

Beyond that it will then be a focus on more content, and continual polish in general.  It looks like we're getting really well situated for me to be able to focus on content (in the form of enemies and game mechanics) next week; and Blue has been focusing on content in the form of level design for quite some time now (and it really shows).  I'm definitely excited about the build you'll have at launch!

[tombik]

I am excited and really curious about how did you deal with 3d procgen. All the other titles I know so far feels pretty repetitive. Since you really spent so much effort on that, this time I am really hopeful.

[x4000]

Our goal wasn't to avoid repetitiveness, but rather to avoid blandness.  In a lot of respects, it's meant to be like procedurally-assembled hand-crafted levels along the lines of what I'd hoped Left 4 Dead 2 was going to do.  However that didn't really manifest there except for a door or two being opened or closed sometimes, which was frustrating.

My goal from the start here was to have really unique spaces so it's not like you would not recognize whatever space, but to make them arranged differently and filled differently each time.  It gives a sense of exploration while at the same time not succumbing to the over-blandness that so many procgen games do.  In practice I'm really pleased with the result, personally, but it's a very different approach from other games.

[Mánagarmr]

That last shot of the raptor really makes you realize how awesome the textures and light work is. Nicely done.

[x4000]

Glad you like it.

[eRe4s3r]

I am gonna have to be nit picky (well better me, than your early access customers eh ;p)

In the first shot, where the green effect sits in the middle of the screen, you can see artifacts bleeding through the effect.
In the first shot, to the right in the bottom, where the blue/white texture is, you can see weird artifacting (isn't helped by the JPG compression though ,p) Looks very very noisy.
In the first shot, on the ceiling, the roof tiles look very very sharp and noisy, with 1 pixel artifacts that indicate a normal map that is peaking in some direction where it shouldn't (usually solved by reducing the "strength" of each normal map color channel) making the normal map less.. "strong" (not sure how else to call it, first and second color channel of that normal map need some reduction)

In the third screenshot, it looks as if there is no anisotropic filtering on the walls (really strange texture mipmap falloff) and there is also a weird "flake" of AO near the tail.

TL;DR, it needs Temporal AA + luma sharpening, SSAA or MSAA for the best possible final look. FXAA/SMAA could work as well, or DSR... I know seeing 100 fps is nice and well, but personally I'd rather be playing this at 40 and have it look pristine and without jaggies (so DSR 2x it would be, I guess..) ;P

Aside that, (I know, that's a weird transition) it looks really nice. It just needs some polish on certain effects.. or maybe lossless screenshots with some downsampling or something... I am pretty sure the game would look a dozen times better with DSR 2x, aka rendering internally at 2x the resolution of the output, downscaled with proper filter. Did you try whether this actually works for the game? You know some people are gonna be sticklers for graphics ;P

[x4000]

In this particular case it is actually doing temporal AA, so it looks really good in practice.  But it tends to really make the screenshots look a lot less sharp.

[eRe4s3r]

In this particular case it is actually doing temporal AA, so it looks really good in practice.  But it tends to really make the screenshots look a lot less sharp.

Oh... you planning on testing some 2x or 3x downsampling with your game? ;p For promo screenshot purposes? Might be better to disable TAA for screenshots entirely and boot up downsampling with lanczos filter via GeDoSaTo or what have you nowadays... it would make your game look vastly spiffier in screenshots. Which I am sure it must look in motion with TAA, but we can't see that ;P

Btw, I never knew TAA would cause such artifacts, lessons learned I guess.

[x4000]

Temporal AA can cause extra artifacts with things that are moving in particular.  But some of it is the nature of the specific CTAA component we're using, I think -- it's still in beta.

But yes, making the screenshots automatically turn off a few things and switch to FXAA would probably be a good idea.  Doing supersampling would be cool, but I wonder if that would be considered misleading, you know?

[Pumpkin]

Now I'm sold on feathered raptors.
There will always be a place for saurian predators in my heart, but I'm sold.

Remember?
Seeing that feathered raptors "in their natural habitat", I can only quote myself. I'm definitely sold on the feathered raptor.
I bet I can find some funny raptor/apache pic...

[x4000]

I'd certainly like to be able to do both at some point, that's for sure.

[Pumpkin]

raptor/apache

I'd certainly like to be able to do both at some point, that's for sure.

[Mánagarmr]

Temporal AA can cause extra artifacts with things that are moving in particular.  But some of it is the nature of the specific CTAA component we're using, I think -- it's still in beta.

But yes, making the screenshots automatically turn off a few things and switch to FXAA would probably be a good idea.  Doing supersampling would be cool, but I wonder if that would be considered misleading, you know?

Supersampling isn't more misleading than any other effect. Supersampling really just is a HUGE image (smaller artifacts) with a blur filter on it. Simplified speaking.

[x4000]

Supersampling is misleading because most games can't do that in realtime.  Aka, it doesn't look like the actual gameplay.  Whereas FXAA instead of temporal AA is an option that you can tick at any time in the game if you so choose, and most rigs can run that just fine.

Bear in mind that I'm using deferred not forward rendering, so things like hardware MSAA are not available, which means that cuts down on the number of machines that could do that.  That may not have been clear before.

Being able to do both with an apache/raptor: I meant feathered and not.