Project Stealth (updated 2009/08/15)

[frvge]

Dear UT3 Community,

I'd like to point you at the Total Conversion Mod that my team and I are making.
It's inspired by the "Stealth Action Multiplayer" gametype.

It's got 2 athletic, fast and non-lethal Spies in 3rd person, who need to accomplish certain objectives, and 2 slower, but powerful Mercs (Mercenaries), who need to guard the map and kill the Spies as a regular FPS.

Both teams are aided by various gadgets from smoke grenades and distraction devices to frag grenades and mines and special visions like Thermal and Motion Tracking.


For the latest news, images and videos, visit ProjectStealthGame.com or our ModDB entry


---------- 2009-08-15:

We're happy to inform you that we'll be exhibiting at GamesCom. You can find us in Hall 08.1, booth B052b, so come by and say hi.

Today we launched our first real website, the informational part of http://www.projectstealthgame.com. We worked on it for many months and I think it paid off.

The latest concept art for the Camnet has also been uploaded, as well as an article in PC Gamer UK. Both can be seen at our ModDB page.

---------- 2009-06-06:
A few already released updates on our own forums that never made it here:

Music and screens (forum) consists of:
3 Fingerprints Merc 1 Merc 2 Spy 1 Spy 2

Basic Movement Animations. This shows off the various standard Spy animations like idling, walking, running and transitions between crouching and standing. Also a running dive as a little extra. The Merc shows his run and a wrong version of walk.

Custom Kismet Dev Commentary. Programmer MulleDK19 shows some of our custom Kismet nodes in action. Has a sneakpeek of our custom PS Editor and the start-up sequence of Project Stealth.

---------- 2009-02-16:
It's been some time since we had an update, and so we need to get the wheels rolling again.
As a small update, we released this Visual Effect of our Camo-system, which renders the Spy invisible with some refraction based on distance and speed.

Camo video

---------- 2008-10-19:
Revised Motion Tracking vision:



And our second Developers Commentary, which can be found at ModDB. Note: 7:51 minute!

Lakehouse Developers Commentary

In between this one and the previous one was a smaller update with a musical explanation from our composer and the announcement that we are working together with an AAA-animator who worked on many Hollywood blockbusters (The Matrix Reloaded/Revolutions, X-Men: Wolverine and so on)

------------------------------------------
2008/10/12:
Our Neutralization Objective has been released today, including an animation.

Links:
Objective #1
Objective #2
Objective #3
Objective #4
Objective #5
Objective #6
Objective #7
Objective #8

Objective Animation (8MB, 0:07 min)

More to come within 3 days ;-)

------------------------------------------
2008/07/25:
Today we released some more screenshots of Lakehouse, which is still pre-alpha, and a render of the security camera.
Here are the links.

Lakehouse #1
Lakehouse #2
Lakehouse #3
Security Camera
------------------------------------------
2008/05/09:
Approximately 2 weeks ago, we released a teaser of our first map called "Lakehouse".
You can see it
here
------------------------------------------
2008/04/16:
A few of the new pictures that were released along with the previous update. Check our ModDB for all of them:



----------------------------------------------
2008/04/15:
First video of EMF-vision with Dev Commentary: click here

-----------------------------------------------
2008/01/28 Update:
We've released some more gadgets in addition to some previously released ones, which hadn't been posted here yet. Also some wallpapers have been released, which I won't post here because of their size.

http://www.projectstealthgame.com/pu...gStatsheet.jpg
and the older
http://www.projectstealthgame.com/pu...mStatsheet.jpg

--------------------------------------------------------

Link to a higher-res render of the finished Spy model:


Merc renders






More Merc renders (great wallpaper-potential):
http://www.projectstealthgame.com/in...5.html#msg9345

And finally, a link to ProjectStealthGame.com where we invite all stealth-loving UT3 fans to take a look. There's no more adrenaline than sneaking out of the shadows to grab an unsuspecting Merc while your Spy partner rushes to finish the objective in the last seconds of the match

While it's more meant as a press post, we're still recruiting experienced Animators and Unreal Script Programmers to join our team.

Hope to see you there,

frvge

[ne_skaju]

Why 3rd person view

[frvge]

Spies need to be able to see Mercs coming when they are busy doing things, in order to get away fast. So more recon.
The Mercs only use first person.
2 different points of view is also one of the distinctive features of the "Stealth Action Multiplayer" gameplay.

[WickedPenguin]

I've never been a fan of stealth / sneaking games, so I can't comment on the gameplay. However, the character models look incredibly good and professionally created. Terrific work on the normal maps and detail.

[frvge]

It's also action. It's more about outmanouvering (sp?), distraction and misdirection than sneaking.

[dirkduncan]

nice models. i'll certainly give it a try

[Lord Vipes]

looks sweet, cant wate to play it

[XenonG]

Finally, ever since Ubisoft messed up Splinter Cell....... Anyway, this will be good.

[goodkebab]

glad you guys liked it.

I am the one making the models, heck, I even had a chance to work at Epic but decided to turn the oppurtunity down when I realized I could not relocate for onsite work.

Our team realizes that the Project Stealth is not a pure FPS shooter because of the third person element.

Truth be told, its a 3rd person versus 1st person multiplayer game which is what makes the game unique and fun. This makes it a fusion of different gameplays, so the game requires great deal of complex tactics that require more then just aiming skills.

-gk

[JaFO]

I hope you do know that Ubisoft had to modify the engine just to get those extra vision-modes working ?

Now I don't know what can be done in the UT3-code without access to engine-code, but stuff like thermal/night-vision can be pretty tricky. AFAIK a few TC's and mutators did add stuff like this, but none used it as the core-requirement for the games like you do.

Do you have a working proof of concept for the various ideas especially concerning the viability of alternate vision ?

And what about the use of AI in your game ?
Even if it is 'multi-player' the AI could be used as basic guards, generic crowds and a few neutral victims (such as a hostage).

[frvge]

No AI. Maybe some AI for practising neckgrabs or other basic things. But NPCs would change the core-gameplay too much. It's the feeling that you and your partner are against 2 humans, so when a misdirection tactic works, that gives a nice feeling of accomplishment.

Vision modes is just applying several filters and shaders. Should be doable. We've got lots of ideas from the fanbase and some are already visualized. But for the real stuff we have to wait till UT3 hits the shelves. Roboblitz is kinda stripped down.

[meowcat]

I want to second JaFo's recommendation for AI use. Even if they could not act as spies, just having additional mercs when playing as the spy would be fun (especially coop). This is particularly important to fill net games when 4 players cannot be gathered together (and modifying AI is not super difficult, it just requires a really good understanding of the existing code base).

The vision modes should not be too much of an issue. I was able to get motion tracking, night vision, and a reasonably acceptable thermal vision working in UT2K4. Access to the native code would probably allow it to be done in a less processor intensive way, but I'll bet it will still work if done in uscript with special texture overlays and hud.

[frvge]

The whole idea of light and shadow gameplay and misdirection is unfortunately not really programmable. We've talked about it, but it'd require additional lightmaps and even that doesn't really make up for a brightly lit wall behind a Spy hiding in a shadowy spot.

Only when you can fool a computer (aka a computer thinks like a human), it's possible.

The concept is good as it is. Additional stuff may be added in Project Stealth 2.0

The Mercenary AI-bots in Ubisoft's SCDA prove that it's hard to balance. And even on Expert they are really dumb. 2 human Spies would always outmanouvre AI Mercs. So then they'd have to cheat by polling the Spy's location every x seconds.

Pinpointing Spy locations based on sound is hard without using too much info that the game is supplying ('they shot out gaspipe number y'), and for humans it's even possible to get a general idea of the Spy-tactics by the _absence_ of certain noises. It's too hard to program that in in a convincing way.

Filling 4 spots isn't that hard, when the game gets popular. The good thing is, for a full game, you only need to have 3 friends .

You're talking about Coop. We're not making Coop in the way Chaos Theory did, which was playing like Single Player with a friend. Maybe I misunderstood you, or vice versa.

But your input is appreciated

[JaFO]

The AI in the demo at least appears to be capable of ignoring 'invisible' players even when they are shooting at them (although the latter may just be a 'feature' for novice bots).

IMHO the real problem for a stealth-based gametype is the need for quality maps.
If the objective is in a room with limited access then the mercs only have to wait for the targets to walk into the trap (especially if the objective forces the spy to be immobile for a few seconds).

Even if there are multiple objectives then the guards would only have to trap one as the spies will eventually have to pick that one, unless the game allows for them to complete n - 1 objectives. And even then the spies would need multiple escape-routes in order to prevent being trapped.

Also keep in mind that in the demo nothing prevents the players from disabling all effects. As a result any vision-modes would have to be coded in such a way that this cheat doesn't work.

[frvge]

The SAM gameplay always has more objectives that can (usually, not always) be taken in a random order. Number of objectives to win <= total number of objectives - (number of mercs +1). So the Mercs will always have to split-up. The tactical decisions from the PoV as Spy are mostly: which objective should I try to get now, so that the Mercs will need to cover a lot of ground, so that lateron, I'll have an avantage using my superior speed and moves. The tactical decisions as Merc can be: How can I cover as many objs as possible? Should we 'give' them the objs at the first floor and place all of our static defenses at the second floor (easier to defend, but no room for errors)?

The UT3 bots are programmed to just shoot at everything. Maybe dark places are less 'seen', but if the backdrop is lit, it still gives a weird situation where the bot is in a shadow, but is still very visible as a silouette.

Good point at the vision modes. I am not sure about this, but since it's a vital part of the gameplay, I don't think you'll be able to switch it off for performance reasons. We'll have to see if that's possible.

[frvge]

Major update:






More renders:
http://projectstealth.splintercellne...5.html#msg9345

First renders and screenshots (= in-game quality) of the Merc, and a render of a Spy sneaking up on a Merc in a map.

[goodkebab]

AI would be cool, but the maps are designed for 2vs2 and bots would never work because unlike simple defense tactics that you could code, the Spy equipment would lay waste to bots. You just could not program enough intelligence for the merc to use his equipment intelligently. What carries this game is skilled use of the equipment, because more often then not, the game is decided when a merc is killed or unconscious.

Bots would be a lot of fun in DM though.

Indeed, this kind of game play requires balanced maps but it also requires a significant amount of psychological tactics like diversions, baiting, and other traps at the disposal of the spy.

A balanced map will always leave 2 mercs defending the last 3 objectives, and the spy is only required to get one.

Our lead level artist is also working on an AAA title liscensed with the UR 3.0 engine...so hopefully any technical problems will be solved from experience.

[innociv]

You better make a cool shader for that guys glasses. That yellow-gold is ugly as sin.

[frvge]

It'll probably be glowmapped, environment-mapped and maybe with an animated texture. It still will get tweaking

[xentrumx]

not a big fan of splintercell but aslong as its free ill give it a try ofcourse. U never know maybe ill find a new cool game or mod :P

[goodkebab]

Different Visions are shown here. The previous images were "nearly" done. Will change face texture next.

[goodkebab]

And for an example of how skilled spies disable mercs when they have only non lethal weapons:

http://www.youtube.com/watch?v=HJypCA6S_Ok

Keep in mind, the Merc is using a standard FPS field of view.

[warlord57]

This is looking good, I can't wait to play it If the coding is anything like UT2004 then alternate visions can be possible with the right amount of work involved. I have created night vision for UT2004 and plan to do the same for UT3. As for other vision modes I have seen people create different ones such as thermal. Keep up the good work and keep us posted

[KewlAzMe]

Stealth FPS are the way to go... Take games like Thievery for UT99 .. nearly 6 yrs old and still being played cuz the gameplay constantly changes and there's always a challenge.

I love you for making this mod! And I agree about having AI guards too. Even iif they are just dumb guards that alert others or shoot rampantly with bad aim. Plus you can't always count on a 4 person game, so u may need an AI to take the balance

[iservealot]

Stealth FPS are the way to go... Take games like And I agree about having AI guards too. Even iif they are just dumb guards that alert others or shoot rampantly with bad aim. Plus you can't always count on a 4 person game, so u may need an AI to take the balance

You have to understand that creating AI with simply Uscript is nearly impossible. Actually, creating AI is a difficult process all together. Ubisoft hired a special company to do their AI for them in Double Agent Versus mode (which was the only online mode that had bots)

With this type of game mode, the AI has to understand when you are in dark places, and understand that their flashlight can be used in dark places. With dynamic shadows this becomes a huge problem, as the AI would need to updated several times a second to understand weather or not you would normally be visible (which would mean many engine ticks which = more lag).

There are of course other major obsticals like allowing the AI to understand what sounds are loud, soft - motion tracking visions etc.

So one again, at the moment we are not planning to have any bot support what so ever. However, if there is someone out there that thinks they can pull this off efficiently with Uscript, then we are not by any means stopping you from trying.

[meowcat]

iservealot: those are all very valid points. And you are right it could take a lot of code running every tick to get it to work. That being said, you again would have very few of these bots (4 or less?) running around, so you might be able to afford the extra processing power.

I can definitely see the dynamic shadows and silhouetting being the biggest 'obstacles' in the AI programming (may just have to be lived with). For merc bots: sound volume can already be handled (one of the parameters passed to the AI HearNoise function is the volume, at least in UT2k4). For static dark areas in maps you could easily subclass 'Volumes' to designate these dark regions. Then when bots enter the dark areas simply decrease their sightradius and peripheral vision and increase the likelihood of using Motion tracking vision or electro view (which would then increase these values). In the SeePlayer function, depending on which vision mode was activated, just do specific distance, item activation and/or velocity checks of the player controlled pawns to verify whether or not the bot actually "sees" them, goes to investigate etc.

[iservealot]

iservealot: those are all very valid points. And you are right it could take a lot of code running every tick to get it to work. That being said, you again would have very few of these bots (4 or less?) running around, so you might be able to afford the extra processing power.

I can definitely see the dynamic shadows and silhouetting being the biggest 'obstacles' in the AI programming (may just have to be lived with). For merc bots: sound volume can already be handled (one of the parameters passed to the AI HearNoise function is the volume, at least in UT2k4). For static dark areas in maps you could easily subclass 'Volumes' to designate these dark regions. Then when bots enter the dark areas simply decrease their sightradius and peripheral vision and increase the likelihood of using Motion tracking vision or electro view (which would then increase these values). In the SeePlayer function, depending on which vision mode was activated, just do specific distance, item activation and/or velocity checks of the player controlled pawns to verify whether or not the bot actually "sees" them, goes to investigate etc.

So is this you applying for the position? If it is, feel free to showcase your modding/scripting work on our website ProjectStealth.splintercellnetwork.com

If not, then trust me when I say we have enough performance issues to face with the kind of environment we are aiming for - Light/Shadow game play. In case you haven't noticed, the lighting engine (even Composite lighting) is dog slow.

[goodkebab]

Stealth FPS are the way to go... Take games like Thievery for UT99 .. nearly 6 yrs old and still being played cuz the gameplay constantly changes and there's always a challenge.

Same goes for the splinter cell series. Multiplayer of Pandora Tomarrow and Chaos Theory are pretty much the same...and it has been around for quite some time now.

The fault lies in UBI getting greedy to try and please a larger fan base, the whole continuous growth thing.


About Bots, perhaps it would be easier to have Merc vs Bots in the beginning....and worry about Spy vs. Bots later.
Attracting players that are not into stealth gameplay is also imporant.

[KewlAzMe]

You have to understand that creating AI with simply Uscript is nearly impossible. Actually, creating AI is a difficult process all together. Ubisoft hired a special company to do their AI for them in Double Agent Versus mode (which was the only online mode that had bots)

With this type of game mode, the AI has to understand when you are in dark places, and understand that their flashlight can be used in dark places. With dynamic shadows this becomes a huge problem, as the AI would need to updated several times a second to understand weather or not you would normally be visible (which would mean many engine ticks which = more lag).

There are of course other major obsticals like allowing the AI to understand what sounds are loud, soft - motion tracking visions etc.

So one again, at the moment we are not planning to have any bot support what so ever. However, if there is someone out there that thinks they can pull this off efficiently with Uscript, then we are not by any means stopping you from trying.

This has all been done by the AI coder for ThieveryUT (99). They can patrol, they can see you in light and not see you in dark. They can know how to chase you and know what weapon is best for the range. They even know when they've lost you and go into search mode. They even have bump testing that can tell if they walked into someone hiding. They can hear and light a flare to see in the dark. They can even relight torches and call out for help based one whether they are low on health or being sniped from a distance.

Take a look at what they did for AI, it is quite advanced, and all done with "mostly" Uscript (maybe a few dlls).. but all can be, and was done by a mod team

[Psycho1989]

Models are extremely nice, however a complete copy of Splinter cell. But I would enjoy it :P

[goodkebab]

^^that was the intention! The multiplayer series is no longer being developed in the style of CT.

Sort of like UT being a clone of its predecessors.....just upgraded graphics and a few new features.

That is exactly what the SC community wanted in the sequels, but UBI decided to go the Asassins Creed direction.

[iservealot]

This has all been done by the AI coder for ThieveryUT (99). They can patrol, they can see you in light and not see you in dark. They can know how to chase you and know what weapon is best for the range. They even know when they've lost you and go into search mode. They even have bump testing that can tell if they walked into someone hiding. They can hear and light a flare to see in the dark. They can even relight torches and call out for help based one whether they are low on health or being sniped from a distance.

Take a look at what they did for AI, it is quite advanced, and all done with "mostly" Uscript (maybe a few dlls).. but all can be, and was done by a mod team

Again, we do not have the resources to spend on recreating an AI system from scratch. If someone would like to invest their time in doing this, they are more than welcome.

None the less, thanks for the information.

[meowcat]

@ iservealot: Let me see if I can get that light based bot vision system working in my own UT2k4 mod (check my sig), if I have reasonable success, and UT3 runs on my other computer, I may be able to help out. Spy bots would obviously be extremely challenging, but I think Merc Bots would be very programmable.

[goodkebab]

wow Cat!

That sounds cool....introducing bots to our project would be a significant feature and introduce an entire game mode for us.

DM is a popular game mode amongst a lot of noob players and this would be a great way to satisfy beginners and let more experienced players play amongst themselves (they despise DM because it was not designed for it).

[Blinco]

Cant wait man

[Lord Vipes]

And for an example of how skilled spies disable mercs when they have only non lethal weapons:

http://www.youtube.com/watch?v=HJypCA6S_Ok

Keep in mind, the Merc is using a standard FPS field of view.

How u make an ingame trailer if UT 3 isnt even out yet?!

[MR.Mic]

Greetings, I am Mr.Mic, one of the environmental artists for Project Stealth.

How u make an ingame trailer if UT 3 isnt even out yet?!

This is a gameplay video of Splinter Cell Chaos Theory, a game in the same genre to which we are closely adhering. The main reason we are making this mod is because even though Splinter Cell Chaos Theory was a very deep competitive experience, it has been just a plain bug-ridden piece of software with nonexistant official support from day one.

I still play Chaos Theory pretty often, and I think I am pretty decent at it. and it's still my favorite game after 3 years. I just hate to see a community die because Ubisoft gave up on us a long time ago.

A common misconception about the multiplayer stealth genre is that it is just slow-paced shadow-camping. This couldn't be further from the truth. My playstyle is, in fact, rather fast paced. You can see this in my merc and spy chaos theory demo videos.

http://www.youtube.com/watch?v=41tkSOPVkkk
http://www.youtube.com/watch?v=LNh88Bk5RXo
http://www.youtube.com/watch?v=OiHCEQg-Gg4

(For better viewing context, headshots are instant kill, and cameras shoot gas that put the mercs temporarily to sleep)

[frvge]

Small update without many details: our animations are coming along nicely.

[MulleDK13]

I hope you do know that Ubisoft had to modify the engine just to get those extra vision-modes working ?

Now I don't know what can be done in the UT3-code without access to engine-code, but stuff like thermal/night-vision can be pretty tricky. AFAIK a few TC's and mutators did add stuff like this, but none used it as the core-requirement for the games like you do.

Do you have a working proof of concept for the various ideas especially concerning the viability of alternate vision ?

And what about the use of AI in your game ?
Even if it is 'multi-player' the AI could be used as basic guards, generic crowds and a few neutral victims (such as a hostage).

It should be quite possible! Just an overlay!

[goodkebab]

I dont see that as a bad thing. It was always our intention to make it look and feel like Splinter Cell, without violating copyrights.