Source 2 (next generation)

about the seamless world, infinite view and no loading from a to z thing, play the ps2 game Jak 3.
 
Pi Mu Rho said:
realtime radiosity? Give it 10 years....

I said realtime-ish, not realtime ;)
Reality engine is faking it, with decent results by viewing the vids (I assume they don't use ambient lights or else it would be lame).
 
I'd like to see hl3 take advantage of the physics processing cards that are currently being worked on.

It would be like taking hl2 and making the physics 30x more interactive
 
Flyingdebris said:
I'd like to see hl3 take advantage of the physics processing cards that are currently being worked on.

It would be like taking hl2 and making the physics 30x more interactive
Wouldn't that need different Game engine?
 
AJ Rimmer said:
Wouldn't that need different Game engine?

No, they would have to rip out Havok engine, and put in Novadek engine because the PPU's are made to only work on the Novadek engine so far... They aren't engine-independent yet because nobody knows how well they will do and such....

With Unreal Engine 3 and god knows how many games will come out using that engine, there will be lots of games taking advantage of it! :bounce:

Hmmm let's see for Source next-gen

XNA Support!!! Think about it... If they made Source with XNA support, and they released Half-Life 2 on the Xbox Xenon with XNA Support, then both PC and Xbox users could play multiplayer together!!!! I'm not sure if VALVe would be able to just take out the networking portion of XNA though.

Soft-bodies/hair/fur/etc etc support

Enhanced weather support

New demo support so the .dem files aren't so big

Maybe do some research into the map editor/map format... I'm sure there are better ways to handle maps (current way works, but of course there is more than one way to skin a cat ;) )

Improved netcode (so that elevators and such actually work smoothly... may just be the cause of the entity?)

Seamless worlds (I hate seeing "Loading" in missions!!)

More GUI's for modding... Instead of having to do command line stuff, it would be easier to maybe get an all-in-one tool so that the pipeline will flow better... Unreal has one, Battlefield 2 has one... If you haven't heard of them you must read up on them... Sure the engine is flexible, but it just should maybe be a bit easier, _especially_ to the guys who want to get started in mod-making by like tweaking simple values around...

Maybe some ways to make textures look less-pixelated when viewed up closely (I've heard lots of new things like new algorithsms and such as of late)

Physics enabled for objects on characters (like for a janitor their keys would jingle.. It would suck to animate and then make sounds for it if you had to do it by hand)

Well this dosn't relate to the engine directly, but a working Friends system for Steam plz :D

Ummm let's see... Maybe improve lighting models to be more accurate (static world, dynamic characters?), self-shadowing (not sure if this is in), well HDR and those lighting tricks are going in-game soon so that's been taking care of...

Graphical upgrades would probably be everyone's #1 thing on their list since it's what appeals to most users :thumbs:
 
XNA Support!!! Think about it... If they made Source with XNA support, and they released Half-Life 2 on the Xbox Xenon with XNA Support, then both PC and Xbox users could play multiplayer together!!!! I'm not sure if VALVe would be able to just take out the networking portion of XNA though.

yay! more sweaty teenagers with speech problems, commonly known as teenspeak.
Soft-bodies/hair/fur/etc etc support
hair/fur not really required, can be done already with tricks, better than wasting resources on proper hair/fur dynamics

Soft bodies, I suppose.. though can be faked to a point already and look/work fine.

Enhanced weather support

Not a lot of reason to have that in built unless its an engine geared towards RPG's or MMORPG's With some coding it can be done just fine at the moment, it just isn't for the newbie to do, just really a more advanced coded, but its quite possible.

New demo support so the .dem files aren't so big

Not really a priority.

Maybe do some research into the map editor/map format... I'm sure there are better ways to handle maps (current way works, but of course there is more than one way to skin a cat ;) )

Agreed, but then im of the school that says Brushes are old news and 100% polygon environments can work perfectly and better. Also easier to build in and texture and play with in general.

Improved netcode (so that elevators and such actually work smoothly... may just be the cause of the entity?)

Didn't Valve say they'd be releasing the source to hl2dm's netcode. I'm sure someone will fix that.

Seamless worlds (I hate seeing "Loading" in missions!!)

This would come under the map format. But even then, seamless isn't seamless. Think how badly HL2 stutters if it loads something, now imagine it loading a 50mb map file on the fly while your playing through it (since it'll still all have to be split like that, for technical reasons) So a good deal of things would need to be changed, not just the map format for invisible loading. General rule though is smaller the map, quicker the load, you could make a seamless game in source right now, if you lowered the details enough.

More GUI's for modding... Instead of having to do command line stuff, it would be easier to maybe get an all-in-one tool so that the pipeline will flow better... Unreal has one, Battlefield 2 has one... If you haven't heard of them you must read up on them... Sure the engine is flexible, but it just should maybe be a bit easier, _especially_ to the guys who want to get started in mod-making by like tweaking simple values around...

Well GUI stuff is being made by people around the net at the moment. But in a development environment, command line programs have their uses, GUI development tends to slow things down, its often easier to write a few batch files to take care of everything. It's also a good idea for newbie mod making teams to learn these things and not rely on drag and drop clicky interfaces

Maybe some ways to make textures look less-pixelated when viewed up closely (I've heard lots of new things like new algorithsms and such as of late)

This will sort itself out as larger textures become the normal, to the point you'd never notice any pixelation.

Physics enabled for objects on characters (like for a janitor their keys would jingle.. It would suck to animate and then make sounds for it if you had to do it by hand)

Well you could probably do this in Source now, question is, would you really want to waste the cpu doing that when it could be doing something important.

Besides if a modeler/animator knows his/her stuff, they can animate such things without a second thought and it look just fine (pre-calculated physics saves the day)

Well this dosn't relate to the engine directly, but a working Friends system for Steam plz :D

Now your just taking the piss, you know that'll never happen, be realistic ;)

Ummm let's see... Maybe improve lighting models to be more accurate (static world, dynamic characters?), self-shadowing (not sure if this is in), well HDR and those lighting tricks are going in-game soon so that's been taking care of...

Well you wont get better lighting in the current form. Because of how it works. Though it would have been nice if the realtime shadows had taken into account the location of the nearest light to the caster and then blend the shadow from its previous direction/angle to the new one, instead of being a single one per map.. Though the realtime shadows do have I/O controls for changing angles, distance and color. So perhaps that already works? I've not tried it.

Anyone wanna give that a go?

Graphical upgrades would probably be everyone's #1 thing on their list since it's what appeals to most users :thumbs:

If people are willing to pay a monthly fee for such upgrades then they would happen. Otherwise its expecting people to work on something that isn't paying, and if thats all Valve end up doing, its byebye Valve a few months later.
 
You pretty much described all there is to the next generation of source,but all those things you described is going to take a hell amount of time to take into a game. :hmph:
 
Wouldn't that need different Game engine?
Source was made to allow its core components to be swaped out with next gen tech.

So, no just some more coding and retelling what calculates what.(Like right now its Gpu or Cpu)
 
yay! more sweaty teenagers with speech problems, commonly known as teenspeak.

You could mute them and still play with that huge playerbase.

hair/fur not really required, can be done already with tricks, better than wasting resources on proper hair/fur dynamics

Soft bodies, I suppose.. though can be faked to a point already and look/work fine.

Main thing I would want is Hair... Why? Makes MUCH more believable characters, and considering valve is aiming for a truly emotional experience, it would be awesome to have this in. Fur is of course a lower priority...

Softbodies can be faked, just set up lots of boxes with constraints and put one material over them all so thus you can't tell... Go download the Novadek Rocket to see this in action.

Not a lot of reason to have that in built unless its an engine geared towards RPG's or MMORPG's With some coding it can be done just fine at the moment, it just isn't for the newbie to do, just really a more advanced coded, but its quite possible.

Hmmm I don't get what you mean by "geared towards RPG's or MMORPG's".... I mean are those two genres the only ones that should get good weather effects? If it enhances the experience, it should be put in... Take a look at SC3... amazing weather system, and it's neither RPG/MMORPG....

Didn't Valve say they'd be releasing the source to hl2dm's netcode. I'm sure someone will fix that.

Not sure about them releasing the sourcecode to the networking portion... I know VALVe knows about this problem though already, I think they said they were going to redo entities or something like I said. Not sure 100%

This would come under the map format. But even then, seamless isn't seamless. Think how badly HL2 stutters if it loads something, now imagine it loading a 50mb map file on the fly while your playing through it (since it'll still all have to be split like that, for technical reasons) So a good deal of things would need to be changed, not just the map format for invisible loading. General rule though is smaller the map, quicker the load, you could make a seamless game in source right now, if you lowered the details enough.

Why load the entire map at once? You could just load it in parts that you need at the time (or else you would get major texture thrasing when trying to load an indoor scene with an outdoor scene connected)... UE3 uses this tech actually, it just calculates where you are and your current orientation/velocity and then loads the appropriate map for that area, or rather the parts you need.

Well GUI stuff is being made by people around the net at the moment. But in a development environment, command line programs have their uses, GUI development tends to slow things down, its often easier to write a few batch files to take care of everything. It's also a good idea for newbie mod making teams to learn these things and not rely on drag and drop clicky interfaces

Batch scripts are helpful, but still what if we have a total newbie to modding, and they don't know any of that? Like in my example, want to change the speed of a rocket... They should just have access to that value that is already created, then tweak and go in-game. Of course you couldn't build a new mod by changing values, but it would get people off the ground into the world of modding. So when it comes time for them to use more advanced tools (like bash scripts, cmd line) then they won't have to learn how to use those, and on top of that learn how to change simple values.

This will sort itself out as larger textures become the normal, to the point you'd never notice any pixelation.

Well you could obviously do that, but who knows how long it will take until you won't be able to see the pixels. It could be a few more years until it's impossible. Wouldn't it also be a fair assumption that also more powerful graphics cards will allow gamers to run games at higher resolutions? So thus then your method wouldn't work as it's more/less resolution dependent (like stretching out a 512x512 texture across a 1280x1024 space... Obviously these values would increase in the future)

Besides if a modeler/animator knows his/her stuff, they can animate such things without a second thought and it look just fine (pre-calculated physics saves the day)

Well in a lot of cases, just doing it pre-calculated would work, but it would be nicer for it to be dynamic, so it doesn't look so static (seeing the keys go to the same position every time at a certain step for example)


Double woot for long posts :D Woot Woot
 
Ue3 is simply too advanced for any "updates" to have any hope of competing. However, Source is scalable, which will be its selling point, Ue3 is not. This means the Source can get better for us who like to whine about it, and still be playable for peopel with geforce 2 cards. Ue3 is "upper class" as it is unable to work with the "lower class" computer.
 
Hyperion2010 said:
Ue3 is simply too advanced for any "updates" to have any hope of competing. However, Source is scalable, which will be its selling point, Ue3 is not. This means the Source can get better for us who like to whine about it, and still be playable for peopel with geforce 2 cards. Ue3 is "upper class" as it is unable to work with the "lower class" computer.

It's being based on the gf6600. By the time ue3 is out, the 6800's would be between mid-range and advanced.
 
Almost forgot. I miss the idle animations from Halflife 1.

And I haven't seen any acid drops either.
 
the very first thing valve needs to do it get steam not to suck and fix the friends list that has been broken for maybe 3/4 of a year now
 
AJ Rimmer said:
Gates also said that 640K of RAM should be enough for anyone, so I'll trust Bill Gates when he can blow trumpet playing soapmonkeys from his ass, until then, my view of him is as follows: He's full of shit.

http://www.nybooks.com/articles/15180#fn*

Quoting people out of context and regurgitating popular myths doesn't make you clever. Just a thought.
 
Firstly, true fluid dynamics are at least 5-10 years off, the pure computational power that is needed to calculate the physical reactions by each particle and then to create the volumetric surface mesh on the fly is extremely astronomical. Check out RealFlow for the industry-level standard of pre-computed fluid dynamics.

Secondly regarding the quips about lighting, just wait until HDRI is implemented into the Source Engine, which alone is a hugely monumental step in achieving photo-realism.

Finally, what exactly are we whining about?
 
having realtime fluid dynamics in the way of CFD modelling for aircraft and other serious engineering applications... most likely not going to happen for decades. A simulations working with a lattice of a few million elements... would take trillions of operations, and that's a fairly course and limited simulations. But there are far less accurate 'faked' methods that give the visual impression of real fluid dynamics.

Surface simulation for water (not taking into account the accurate behaviour of the underlying volume) seems like it should be possible.

But yeah, considering the number of computations required for even mediocre fluid sims, there are a lot of higher priorities for gamemakers, I think.

Having non-planar water seems like it would be a good one, first. Have fluid dynamics simulated water is pretty useless if the water shaders still look like crap.
 
To the original threader:

The screenshots you posted look like total crap. This proves that once again, superior design is better than superior graphics. As long as you don't have talented artists that can measure up to a technology, the technology is worthless. And in these days where kids are discouraged to be arttists, there is bound to be some problems.

<3
 
Back
Top