Author Topic: Phaeton (powered with Storm Engine 2): self-driving vehicle test  (Read 3983 times)

0 Members and 1 Guest are viewing this topic.

motorsep

  • Hero Member
  • *****
  • Posts: 1,039
  • Karma: +74/-131
  • Artist
    • View Profile
    • Kot in Action Creative Artel
Self-driving vehicle that can be hijacked!


argoon

  • Sr. Member
  • ****
  • Posts: 275
  • Karma: +21/-81
  • Doom Newbie
    • View Profile
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #1 on: January 24, 2015, 11:49:45 PM »
Very good work i like how the vehicle behaves.  :)

motorsep

  • Hero Member
  • *****
  • Posts: 1,039
  • Karma: +74/-131
  • Artist
    • View Profile
    • Kot in Action Creative Artel
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #2 on: January 25, 2015, 01:04:47 AM »
Thanks man! We are working hard :)

TeaMonster

  • Newbie
  • *
  • Posts: 25
  • Karma: +0/-0
  • Doom Newbie
    • View Profile
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #3 on: January 27, 2015, 02:41:10 AM »
You even got the suspension animation right.

motorsep

  • Hero Member
  • *****
  • Posts: 1,039
  • Karma: +74/-131
  • Artist
    • View Profile
    • Kot in Action Creative Artel
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #4 on: January 27, 2015, 07:41:33 AM »
You even got the suspension animation right.

It's not animation, it's physics sim :-)

solarsplace

  • Jr. Member
  • **
  • Posts: 63
  • Karma: +8/-0
  • Mainly a programmer...
    • View Profile
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #5 on: January 27, 2015, 08:02:26 AM »
Hi

Good stuff!

The only suggestion right now would be to add some kind of transition effect for entering and leaving the vehicle maybe? but you probably already thought of that and this is a first show, so very nice work!

Cheers

motorsep

  • Hero Member
  • *****
  • Posts: 1,039
  • Karma: +74/-131
  • Artist
    • View Profile
    • Kot in Action Creative Artel
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #6 on: January 27, 2015, 08:24:09 AM »
Hi

Good stuff!

The only suggestion right now would be to add some kind of transition effect for entering and leaving the vehicle maybe? but you probably already thought of that and this is a first show, so very nice work!

Cheers

:-)  thanks guys.

I think it's one of those things that is nice to have but total overkill for an indie game. Not only it doesn't contribute to gameplay in any way, it also slows down the flow.

Radegast

  • Newbie
  • *
  • Posts: 27
  • Karma: +3/-1
  • Alpha Labs scientist
    • View Profile
    • ET: Legacy
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #7 on: February 03, 2015, 02:09:11 PM »
According to the feature list at IndieDB you are using a proprietary physics engine for this. Is there a reason why you didn't use one of the open-source physics engines like Newton or Bullet? I know nothing about them, I am just curious.

motorsep

  • Hero Member
  • *****
  • Posts: 1,039
  • Karma: +74/-131
  • Artist
    • View Profile
    • Kot in Action Creative Artel
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #8 on: February 03, 2015, 02:13:04 PM »
Proprietary in this case means whatever physics iD Software had in id Tech 4. Since it's not a middleware, it's proprietary.

The way physics engine is made in id Tech 4 makes it really hard to replace with anything else. Plus there isn't much to gain from adding either of those engines. Nvidia PhysX or Havoc would make a ton of difference.

argoon

  • Sr. Member
  • ****
  • Posts: 275
  • Karma: +21/-81
  • Doom Newbie
    • View Profile
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #9 on: February 03, 2015, 05:31:19 PM »
Plus there isn't much to gain from adding either of those engines. Nvidia PhysX or Havoc would make a ton of difference.

How so?

IMO Neither physx or Havok have anything above Bullet or Newton but better tools maybe, in the performance, features departement theres almost no difference between them, PhysX is the only one with GPU physx support out there is true, but a low number of games uses that feature, and only for Nvidia users, bullet has it (for everyone) but theres no game using the feature, about performance Havok and bullet feels better than PhysX to me, but i will say that among them Newton is the better, any game that i played that uses it has really stable physics, no object becoming crazy and twitching, no ragdoll craziness like limbs stretching and stuff and more importantly it really feels like the objects have real height, no wonder being the most accurate physics engine of them all being used even for professional and scientific applications, the frictional Games guys use it on their games, that is why they were able to do that physics based manipulation system so stable.

About replacing the idtech physics, the source code for the physics engine is open sourced as well, i bet someone with good c++ knowledge and physics engines would have no problem perfecting it or even replacing it for something better, but it is really just a matter of asking yourself do i really need that?     

motorsep

  • Hero Member
  • *****
  • Posts: 1,039
  • Karma: +74/-131
  • Artist
    • View Profile
    • Kot in Action Creative Artel
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #10 on: February 03, 2015, 06:45:28 PM »
Afaik Havok and PhysX are both GPU accelerated. Havok has the best performance among physics engine. Bullet 3.0 that is fast and hardware accelerated are on PS3/4 only afaik. It's definitely not under FOSS license.

How do you know that other FOSS physics engine are better? Have you actually used another FOSS engine with Newton or Bullet or whatever else and at minimum recreated Doom 3 gameplay scenario with similar assets and ragdolls and such? I played with Bullet in Blender and while it's more stable than id Tech 4 physics, it's _slow_. I played with ODE in DP engine is ODE was awful. If you player any AAA game with Havok and such, there are plenty of cases of physics going out of whack. It's just more stable than id Tech 4. Existing physics can be tweaked to be as stable and that will be a way less work than gutting existing system and replacing it with something else. Not to mention getting it to work predictably just as id Tech 4 system works.

Lol, having source code doesn't magically solves problems. I have worked with 3 AAA programmers on the engine, who know id Tech 4 inside and out (worked on the engine during different times). So I don't just come out here and say things. We evaluated a lot of things we could do, and the consensus is that it's plain better to get UE4 or even Unity than fool around with replacing parts of id Tech 4.

The engine is not designed as a lot of modern engines, thus having to implement things that are mainstream in other engine is not always possible. Finding a specialist that understands physics really well (not to mention its application in real-time game engines) is extremely hard. It's just funny to see all these hypothetical ideas "let's just do this!" while in reality no one even fixes / implements basic things that are a way, a way simpler than replacing physics engine in id Tech 4 :/

Seeing conversations like this one I am almost tempted to release my engine before I even release anything playable of my game as I have a feeling it will still collect dust on Github by the time I will release something playable :P

argoon

  • Sr. Member
  • ****
  • Posts: 275
  • Karma: +21/-81
  • Doom Newbie
    • View Profile
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #11 on: February 04, 2015, 07:29:35 PM »
Newton was not always a open source engine it was paid and proprietary before, so just because it is now open, like many 3D engines that the market forced open to survive, that does not mean the quality is worse, Bullet was or is, being made by a former Havok engineer, and on Blender if it is slow (never used it)  then maybe the implementation is bad or they have a more realistic solver for movies and such, just speculation tho. I know that Newton is good not only because of being used for professional applications, but also because i played with it on many indie games that use it, plus also the penumbra and Amnesia series that i consider AAA, all the physics were very nice, and afaik there's no GPU physics for it. And maybe i was not explicit, but on my reply above i was talking has a user, not has a developer, so no, i never implemented any physics engine but i did played with them on engines that i used through the years.

Havok is indeed a nice physics engine, and i forgot that it has GPU acceleration but that was never used in actual games (that i know of), so never passed through my mind, bullet because was being touted by AMD has the PhysX competitor for their GPU physics just sticked i guess, about Havok performance, it has many years of experience has a physics middleware for games, so no wonder it is a good physics engine but i played many games that used it and it felt different in all of them, for example on Oblivion and Fallout i saw many ragdoll and rigid body glitches, but on HL2 and Max Payne 2 i saw none, so developer knowledge does have an effect on this, but again the "feeling" of height, of objects between them (the physics engines) is very different, and to me and this falls in a matter of opinion Newton felt better, the objects more heavy and grounded.

About ODE even tho it was used in many indie games and even the Stalker series, i'm also of the opinion that is indeed the worse of all of them (but still better than even the Doom 3 physics), for example it has no support for triangle mesh collisions (meaning you can't use a triangle mesh as a dynamic body but it has collision detection for triangle meshes, for example the static world geometry) nor has it any support for automatic convex decomposition (making automatic physics shapes for objects), at least the last time i used it this was true and afaik it was open source from the beginning so that maybe has an influence in the quality.

Some links for you, they could be obsolete by now but they still help give a general idea.

http://blog.wolfire.com/2010/03/Comparing-ODE-and-Bullet

http://www.adrianboeing.com/pal/papers/p281-boeing.pdf


And about your rant i was thinking to not respond to that you are just getting some steam out, i hope, but has a courtesy i will say this,  imo i would not expect people to use your engine for third party games, not now that the market is so saturated and the competition from AAA engines is very fierce, if i was you i would work on it and make it the best i could for my game, and if no one uses it who cares, the important is that your game is successful or at least profitable.     

The Happy Friar

  • Happy Happy Joy Joy
  • Administrator
  • Hero Member
  • *****
  • Posts: 935
  • Karma: +44/-4
  • Semi-Newbie.
    • View Profile
    • EarthQuake
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #12 on: February 04, 2015, 11:26:24 PM »
The biggest "improvement" to id's physics would be using the GPU to offload the CPU, like PhysX does.  Besides that, it can do pretty much anything other physics engines can in relation to rigid bodies, ragdolls & collisions.  Obviously it's still dated to 2004 and I wouldn't compare it to much that's used now (I'd hope the other physics engines updated since 2004!).

For most applications, it's more then enough.

Either way, usable vehicles in Doom 3 engine (besides ETQW) is still completely awesome.  :)

argoon

  • Sr. Member
  • ****
  • Posts: 275
  • Karma: +21/-81
  • Doom Newbie
    • View Profile
Re: Phaeton (powered with Storm Engine 2): self-driving vehicle test
« Reply #13 on: February 05, 2015, 06:08:22 PM »
The biggest "improvement" to id's physics would be using the GPU to offload the CPU, like PhysX does...

No necessarily, first that would be a enormous undertaking and second the GPU's even using PhysX, Havok, bullet etc, can only accelerate non gameplay affecting physics, like particle physics, smoke, liquids, cloth, etc but normal rigid bodies physics, like boxes and stuff, ragdolls, continuous collision detection for ammunition and the player, AI collision and pathing, etc needs to still be made on the CPU. To many people are mistaken about what Physx is capable, they think "Oh look this game has PhysX engine so it uses my GPU!" but no, it needs to specifically be made to do that by the game developer, if they don't then it will automatically run everything exclusively on the CPU or CPU's, even if you have a PhysX supporting Nvidia card with it enabled on the Nvidia drivers, that option is only for PhysX games with extra GPU physics, but Nvidia knows that seeing that option ticked makes a unsuspected user think their GPU is accelerating Physx all the time so they make no mention at all about this.   

For most applications, it's more then enough.

I agree.

Either way, usable vehicles in Doom 3 engine (besides ETQW) is still completely awesome.  :)

Exactly, the video above just shows that doom3 physics are good enough for realistic car physics.