New technology: GameToolkit

[DonMeck]

posm/eric, EvoG,

I am slowly understanding the whole concept and it’s really interesting. I can see that some stuff is existing things described with other words, but because its looking from a different view-angle - a differen metaphor.

Beside your graphics/game engine, you introduce a new fusion of ai and behaviour : ai driven behaviour - using weightmaps.

(Behaviour is programmcode that acts on the data given - like a state-machine. In OO design these behaviours are bound to objects, thus allow reuseing, aggregation etc.)

I can really see that this is a next level/step of realizing the behviour of objects/characters. Nothing that cannot be done with existing techonolgies, but you give right from the beginning the architecture, framework and tools to do so. Nice!

I’ve now been working with the Virtools-Technology for some years now - EvoG, you know me in it’s forum under the nickname ‘dom’. It’s first aim was to give tools for non-programmers to create interactive content, but using the oo approach of behaviours, proposing a bigger range of ready-to-use components. But it’s clear, thats the full power of the technology could not be used without SDK programming. It has hacnged since they introduced a script-language with borader access to the core technology.

But - it’s very important to understand - although all the programms EvoG mentioned director, Virtools, Quest3D (where Virtools is really a different league) - there’s significant differentce about implementing behviours. And thats the concept behind the architecture AND the tools delivered with. Well beside the core technology, of course.

Well lets trash director, because it’s really low-quality. Lets compare Quest3D and Virtools: there’s a big difference in concept which leads to a big difference for time-to-market. Virtools has a much higher data- abstraction, and offeres difference in represaentation. This leads to eaiser reading of ‘behaviour-graphs’ and quicker implementation. In Quest3D everything is a channel - no difference between data-types, and data-types and behaviour. It’s a big difference, and thats why i don’t switched to Quest3D - same for many other Virtools users.

Although Virtools is quite unique with its technology and it says to be a rapid-development tool - its not always true … or to be more exact, you certainly will be faster thant starting without any framework etc. but in certain cases you need to code/modify a lot of components using the SDK/the scripting language. Additionally the workflow for integrating characters and character animation is not very flexible and it lacks tools for troubbleshooting.

Posm, explained that they are now looking for professionals wich will help them to develop a good, streamlined workslow/production-pipeline. And thats a very important aspect - some developers nowadays tend to say thats even getting the most important aspect. GameDev is iterative - a good streamlined dev-cycle (modelling/animation package <-> real-time framework <-> other tools) is important, because it’s getting more and more the bottleneck.

So to resume, whatif show an impressing technolgy and a new concept of creating interaction. They are now focussing on they production-pipeline … I guess we need to wait until they finished that, in order to see how flexible and fast (which is really the key) one can be using their tool.

I whish you the best. Too said that i am not in a prof. gameDev company. I’d love to test and critzise it

So, changing topic a bit:
I am impressed by the mesh/material morphing. I’ve an idea how this can be done using modern tech - but its certainly with a big overhead - thus not the way you are doing it: rendering both states in different renderTargets and then blending it using a 2d mask into the frame-buffer.

Looks like the animation morphing is like vertex-morph8ing but on a per-object/bone basis instead per-vertex. I guess you need to define quite a lot of states to have a walkcycle? Or did get something wrong?

Btw. you still need mirroring in Verty? I’ve no time … but maybe i can take some

cya, till next time (I am not watching forums that often anymore )
Dom

[EvoG]

HEY DOM!

Well see, my problem was about the vague specifics. I don’t want to read about theory and concept, I want to see execution of something applicable, as I mentioned in my previous post, before I can appreciate how it works or what it is. So far, I see a modeling/shader type dealie with a scripting language but nothing related to visual interactivity creation. This another SDK more or less? I’m interested in authoring(being an artist).

Hell I’m always open to something new, but show me something tangible.

Cheers

[PetterSundnes]

DonMeck: You are right about Director being totally inferior to Virtools, but then again Director costs 1200 US$ and not 5000 US$ pr module as Virtools does. I look at Director as a good all-round application for interactive media, and prototyping, with simple 3D capabilities. It’s definately not a good game engine, but very good to learn how to do interactive 3D in general. Good realtime graphics is also more down to texturing than anything else (like radiosity based shadowmaps++). Check out this demo me and a friend has made with 3DS and Director:
http://illustrata.no/

The new pixel shaders in Virtools looks really impressive. Wish I had that kind of control in Director…

I look forward to see what GameToolKit will end up like.

[DonMeck]

Hey all,

I just watch the pre-alpha tools video, and clarifies some things up for me. (I hate to be forced to install the media player )

But first to the other posts:

@EvoG,
yes I understand your aspect, but its not soo theorectical. It missing some info about the content creation. But we have the max tools in the above mentioned video, with an introduction about the possi-maps and a vague idea of how behaviours are scripted (with speech-to-text wow! ). Maybe it’s explained in the other videos - but i cant remember their content. Don’t forget that they are still working on it - alpha state i guess - or pre-alpha?

And i think there might be a misunderstanding - i don’t think that whatif is aming for the same market like Virtools, although Virtools is more and more aiming in direction of the high-segement of the market.
Maybe one can compare it more with Tools like coppermine, which looks like really having a great production pipeline. Btw. it’s the same price tag like jupiter (i think). Compared with the features and the cost of needed Virtools add-ons - it’s the same price-range like Virtools. Something one can think about

@petterms,
don’t worry I can see that director has its own markt segement and royality free distribution makes it interesting for agenties etc. But you cannot produce high-quality in a short time with i it, i believe. API is very limited and Lingo really sucks! Dunno if its easier to work with the new JavaScript integration.
Yes you demo looks nice - baked light solutions. With that I can even make VRML stuff looking great
The floor reflection, is it an engine feature or mirror geometry?

@posm,
you know that i once wrote a complete export suite with tools etc. Some things you have are similar to stuff i did. Like the quick export function. Actually i did it a bit different. Every export setting is stored within the 3ds max scene (using persitent vars). A dialog allows modifications. So calling quick export does not rely on hard-wired/coded params within the macro scripts. Additionally your batch export could rely on that. You should consider that change in your scripts. Another option would be to think in temrs of ‘projects’ and to have config-files per project, so several scenes share same settings. Same for your executable choice option. I also did quick launch or combined execeution shortcuts/buttons - like ‘export und open in 3dengine’ or ‘execute and open in editor’. These settings can also be stored in a scene or gloabally in config files, easily.

It’s good (especially for debugging) to have all these log-outputs. But from experience thats not enough. It may take a lot of time to go trough of them and find errors. Based on you know-how what can go wrong, you should add ‘inspector’-like applications/tools that parse and analyse the log. Thats faster and easier for developers. Moreover I would create scripts, that verify integrity of the scene and advise the artist/content creator what should be corrected.
With Virtools I had a lot of troubble exporting my character an my animations, as i went a different way than the usal way. It also creates warning etc, but they did not warn me about many issues = time costly debugging.
Especially all the info you know about mirrored bones, mirrored geometrey etc. Thats something that can be checked and should be included as inspector-script or so. You know how long this takes to figure out etc.

I know see that the morphing of poses (la bit like the ani-pose script i used - is it from Grant? ), need of course something you ‘called’ parallelism - which means the same structure but different attributes. So forget my comment about render-targets

[EricChadwick]

Looks like the animation morphing is like vertex-morph8ing but on a per-object/bone basis instead per-vertex. I guess you need to define quite a lot of states to have a walkcycle?
If you’re using bones, then you just lay out note keys to “capture” as many poses as you wish. If you’re using morph targets, then you’d need a great deal of poses for a walk cycle… not recommended, since it’s a terrible waste of GPU/CPU cycles. But hey, if your hardware can take it, by all means feel free to do so.

Blending between poses can be per-bone, per-mesh, per-material-ingredient, per-behavior, etc., even per-vertex. You can use pixels to control blending, so blending can be on a per-pixel basis. Well, depending on if the hardware supports it, otherwise it may be per-vertex. This content-adjusting-to-hardware is automatically managed by the system, since it is hardware-agnostic, and at the same time hardware-aware.

You can have contact-based pixel generation, to control blending. So for example you could slice into a character’s arm and see the muscle tissue and eventually bone, and even slice all the way through. Where the sword contacts skin, “mask” pixels would start to be generated, which would blend in the “gore” model. You can use as many “levels” as you like, and as brighter mask pixels are generated by the contact, more levels would be blended into that spot. Does that make sense? Think of Resident Evil, with contact-specific damage. Or slicing bread. Or cutting an apple. We’re working on samples to show this off…

Btw. you still need mirroring in Verty? I’ve no time … but maybe i can take some
Mirroring isn’t all that necessary, there are ways around it. What I would love to see is some refinements for the Insert New Verts function. It gets the vertex order perfect, but I wish it would give a better position for the new vert, perhaps an average of its neighbors, or even better… an approximation of the Source vert’s position, relative to its neighbors. While I’m making requests… I would love to see UV and material ID and smoothing group info created for the new vert as well. Perhaps just copy this info from the source vert? Or maybe average from surrounding verts in the target mesh? Verty is great, thanks for sharing it.

I want to see execution of something applicable, as I mentioned in my previous post, before I can appreciate how it works or what it is. So far, I see a modeling/shader type dealie with a scripting language but nothing related to visual interactivity creation. This another SDK more or less? I’m interested in authoring(being an artist).
We’re working on it! It’s not just an SDK, it’s a whole system. Some of the tools are primitive right now, enough to demonstrate with, but we have a lot more ideas and tools in development. We’re keenly interested in creating tools for the creation of interactivity, since we feel this is one of the real strengths of GameProcessor. As soon as we have something to share, believe me we will share it.

Check out this demo me and a friend has made with 3DS and Director:
Hey petterms, that’s cool. It’s amazing what good lightmaps can do. I like the reflection in the hardwood floor too. I wish there was something more up those stairs…

calling quick export does not rely on hard-wired/coded params within the macro scripts.
Well, like the video says, the tools are pre-alpha. So these are preliminary tools at best, the quickest route for us to get functionality until we can start full tools production… soon! I like the persistent vars idea. Right now we save an ini file using the same dir and filename as the max file. Then anytime the exporter sees the two files together, it can use the settings stored in the ini (or you can override). But the ini files will probably be replaced too, we’ll see.

The quick export tools I don’t use anymore, they’re just quickie one-offs, will certainly be replaced by better tools. The one we do continue to use is the Mass Exporter, for auto-exporting reams of test content, which we do anytime the exporter code is updated, to easily check for any possible errors.

you should add ‘inspector’-like applications/tools
Yes, that’s the idea once we get to full tools production. Additionally, we’re continually updating the Opoc specification document, currently about 120 pages or so, as we develop the format and the tools. This contains a large amount of information about each feature you can export from max (and other tools not yet created)… what is supported, what isn’t, pitfalls to avoid, etc. I haven’t seen this kind of document for very many other tools, especially not for interactive apps. It takes a lot of work to develop and maintain… but ultimately it is worth every minute.

I know see that the morphing of poses (la bit like the ani-pose script i used - is it from Grant? ), need of course something you ‘called’ parallelism
In GameProcessor (new name, same product), “parallelism” is only needed when blending between two or more separate skeletons. It isn’t needed if you are blending poses for a single character… if so then you just animate as you normally would, then “capture” however many keyframes/poses you want to use (by laying down note keys), and GameProcessor will blend between them to create animation or interaction. Really pretty simple.

But if you have two separate character models, each with their own bones, and you want them to blend, then parallelism is needed… the same bones need to be assigned per vert. Also actually pretty simple, the tool does a good job of preserving vert weights.

Morphing is really not an apt term for what we’re doing, since the blending is a bit more complex. We use Akima splines to connect all the poses, so the space between them can be as curved or as linear as you wish. Curvature can distort both time and space between the poses, adjusting the way that the posm network is traversed. Very cool stuff.

Well, I could go on and on. But please do ask more questions if you have them, and I’ll do my best to answer them.

BTW, I haven’t tried Ani Pose, but here’s the page…
http://www.chuggnut.com/scripts/anipose/anipose.htm

[SIGIL]

Pardon my ignorance, this post may or may not be response worthy…

From what I have read I think I most understand EVoG’s perspective, but this could easily be attributed to my utter lack of programming knowledge. To me this is all greek, but I get the feeling that is not the intent of the project. I kind of get the idea that what Whatif is trying to do is to offer something that would allow someone like me ( NO programming skill or knowledge ) to produce a game from the toolset built-ins and an Art generation package like 3Dsmax, Maya, and so on… However due to my lack of comprehension of, or the omission / complication of details in the explanation, I do not see how this product offers me the ability to do so… Am I correct in the assumption that the intent of this is to offer a kind of Game world Sandbox where in the artist focuses totally on creativity within the confines of the Sandbox itself rather than having to go out and get the sand to fill the box with?

Anyway. I am wondering; What exactly is Whatif looking for in a " creative team to partner with " in the creation of a game project?

Are you looking for a team of artists, a team of designers? Would Whatif be interested in a Designer/ 3D Artist / Writer individual with an original well developed IP and preliminary game design document?

If so what materials do you require for review and consideration?

Thank you for your time and good luck with the venture I am very intrigued by this and will be looking forward to further development.

[EricChadwick]

All posts are response-worthy! I think a dialogue is always good.

…intent of this is to offer a kind of Game world Sandbox…
That’s the eventual goal, except not confined to only game-type content. At present, we’re working to develop tools for a particular genre of game (RPG), to ultimately fit the needs of an experienced team of game development creatives. This development period will allow us to expand the toolset and features in order to eventually make the system easily-accessible to non-developers. We’d like to see the day when people can create interactive content without heaps of technical knowledge… although of course with any tool, be it a wood chisel or a software package, what you bring to it determines the quality of the output. We can create a very ergonomic handle, give it a laser-guided cutting system, etc., but it is up to the user to make something beautiful with it.

However for now the focus is on a particular niche (game development), which ultimately makes the most sense, from a tools perspective. Thus, some “Greek” is needed right now, to communicate with our intended audience.

What exactly is Whatif looking for in a " creative team to partner with " in the creation of a game project?
We’re looking for a full game development creative team with experience on AAA titles. Unfortunately, a team of one wouldn’t be able to complete all the assets for a high-end game title in a reasonable amount of time. But if you have a team and are interested, please don’t hesitate to contact us…
http://www.whatif-productions.com/contact.htm

Having said that, we would definitely consider a well-developed IP on its own, and work to find a creative team to develop it. Please contact us.

Thanks for the post.

[SIGIL]

We’re looking for a full game development creative team with experience on AAA titles. Unfortunately, a team of one wouldn’t be able to complete all the assets for a high-end game title in a reasonable amount of time. But if you have a team and are interested, please don’t hesitate to contact us…

I had assumed that you guys ARE looking for a full fledged team of artists, but I just wanted to clarify…

Having said that, we would definitely consider a well-developed IP on its own, and work to find a creative team to develop it. Please contact us.

Will do. I have an IP that I have been working on in some form or another for about 10 years…

[DonMeck]

If you’re using bones, then you just lay out note keys to “capture” as many poses as you wish. If you’re using morph targets, then you’d need a great deal of poses for a walk cycle… not recommended, since it’s a terrible waste of GPU/CPU cycles. But hey, if your hardware can take it, by all means feel free to do so.

sorry, misunderstood of expression. When i was speaking of morphing, I did not speak ofr per-vertex morphing, but more of th concept of the transition from a source value/param to a target value/param. So like key-to-key interpolation. Naturally with the possibility of using interpoaltion functions for non-linear transitions.

his content-adjusting-to-hardware is automatically managed by the system

That’s great.

You can have contact-based pixel generation, to control blending. So for example you could slice into a character’s arm and see the muscle tissue and eventually bone, and even slice all the way through. Where the sword contacts skin, “mask” pixels would start to be generated, which would blend in the “gore” model. You can use as many “levels” as you like, and as brighter mask pixels are generated by the contact, more levels would be blended into that spot. Does that make sense? Think of Resident Evil, with contact-specific damage. Or slicing bread. Or cutting an apple. We’re working on samples to show this off…

Sounds great. Unfortunally i’ve never played that game. Does it exist for PC also? I hate console-pads! Grr
Sounds also like the technique I made up earlier in the thread - didn’t know that this already exits. Great!

but I wish it would give a better position for the new vert, perhaps an average of its neighbors, or even better… an approximation of the Source vert’s position, relative to its neighbors.

Actually that was already implemented, as you might have seen with ‘inerpolate pos’. It does take the pos of the neighboors and uses the middle. But it doesn’t work.
Maybe with max6 there now new maxscript tools for inserting things on the surface - which would be a better concept. So perfect invisible alignment could be the result. Dunno, any ideas?

I would love to see UV and material ID and smoothing group info created for the new vert as well. Perhaps just copy this info from the source vert?

Yeah - these maxscripts functions were broken in max4. Time for an update

but ultimately it is worth every minute.

And a lot of fun i guess.

I haven’t tried Ani Pose,

It’s slow, but a very util tool! Some of the stuff where one wonder why discreet doesn’t offer it within the core.

It’s very handy for skin-weighting! You can changes poses without leaving the sub-mode of the skin-modifier … it really speed up the procedure! No leaving the skin-modifer, choosing a handle (or how i did it before, with a swap object script), change pose, back into skin, adjust, back to handles, adjust pose, back to skin and so forth. Or no need of working with the timeline.

I kind of get the idea that what Whatif is trying to do is to offer something that would allow someone like me ( NO programming skill or knowledge ) to produce a game from the toolset built-ins

Well, they never claimed that. Only said, that the artist of a gameDev team can do more without the help of the programmers.

eric,
what about UI interface creation (widgets). I can see you got some stuff there. Reuseable?

[EricChadwick]

Think of Resident Evil, with contact-specific damage.
The technique I’m describing doesn’t exist in that game, I was just citing it as an example of how the genre could be taken further with GameProcessor. We can give location-specific damage, with wholly-artist-controlled damage states/transitions. Think of a future hack-and-slash game where you can really do visible --and unique-- damage to the zombies, based on where your weapon hits them, and how hard it hits. Even which side if the weapon… the flat of a blade vs. the sharp edge vs. the point.

Violent gaming is really just an easy example though, it could be used for more interesting things, perhaps carving ice sculptures, or using a chainsaw to carve a totem pole. Just thinking here…

It does take the pos of the neighboors and uses the middle. But it doesn’t work.
Sorry, not sure where to go with Verty from here. I don’t have max6, still in 5.1. I would think the average would work, but that’s probably a lot more math than I can handle!

It’s very handy for skin-weighting!
You should check out Skin Or Die… it has posable functions too, plus some other great skin tools.

would allow someone like me ( NO programming skill or knowledge ) to produce a game from the toolset built-ins
Actually that is not far off from the eventual goal, it just isn’t the immediate one. We do want to make content-creation more universally accessible… we just need to focus on a particular market at present, to help flesh out the tools, and show what the system is capable of.

what about UI interface creation (widgets). I can see you got some stuff there. Reuseable?
I’m not sure what you’re referring to here. We have some icons we made for the maxscripts, and we have a preliminary UI in the runtime… but the UI is still very early in its development.

[SIGIL]

The technique I’m describing doesn’t exist in that game, I was just citing it as an example of how the genre could be taken further with GameProcessor. We can give location-specific damage, with wholly-artist-controlled damage states/transitions. Think of a future hack-and-slash game where you can really do visible --and unique-- damage to the zombies, based on where your weapon hits them, and how hard it hits. Even which side if the weapon… the flat of a blade vs. the sharp edge vs. the point.

Okay I do not know how I missed that earlier but that is exactly the kind of thing I have been looking for in an engine… THAT is probably the most intriguing thing I have read about this engine so far. I would think that something like that should be demonstrated as a key feature… It relates directly to gameplay possibilities and at the same time evokes a lot of imagination… Not to mention I cannot think of any other tech demo that shows anything like that.

[EricChadwick]

I would think that something like that should be demonstrated as a key feature
We’re working on it! We haven’t developed art tools yet to exploit this feature, although I’ve seen it work with “programmer art.”

Btw, we are presenting a live public demonstration of our software in the Boston area next Tuesday evening, April 13. If you’re in the area, it would be a great opportunity to see our technology in person.
http://www.bostonpostmortem.org/

[bupaje]

This is really impressive stuff. I’m looking forward to seeing more -won’t bug you about pricing… only hope I can at least test drive it as a demo or maybe something like ZBrush does with a functional and useful but feature limited product. In fact this is one of those products -like ZBrush- that I like because it doesn’t try to duplicate functionality but really sets a trend -and the fact that it is ‘artist friendly’ is great as while not an idiot I find myself just using the level design and graphic related portions of game development products as that where my inetrests and major ability lay.

Keep updating this thread, cats out of the bag anyway and I already posted a few links to your preview page on forums I frequent. Very cool…

[bupaje]

@Eric - I was reading your description on this future feature “We can give location-specific damage, with wholly-artist-controlled damage states/transitions. Think of a future hack-and-slash game where you can really do visible --and unique-- damage to the zombies, based on where your weapon hits them, and how hard it hits. Even which side if the weapon… the flat of a blade vs. the sharp edge vs. the point.”

It reminded me of a suggestion I had made for a texture design tool some time back. I had been tossing ideas with someone who makes 3ds max plugins and thought it would be good to be able to create a model of a medieval castle door for example and be able to select axes, swords, fire, bullets etc and scar the door and be able to export the texture with any maps. If people could define tools like hammers, chisel etc and how the material reacts as far as hardness etc it could be a good way to make some interactive texture design -for example create a plaster wall, add several layers -burned, rotted, wires and pipes etc and then use your tools/weapons to ‘sculpt’ it realistically and export your texture. Figure it would be easier to create more 3D looking 2D textures than by standard 2D drawing programs or even easier than trying to model all those layers by pushing verts etc

[EricChadwick]

define tools like hammers, chisel etc and how the material reacts as far as hardness etc
This is intriguing. Thanks for sharing!

One of the difficulties with this approach would be defining all the material properties for your various surfaces. For accurate breakage you’d want to define the physical attributes for your surfaces… mass, fragility, elasticity, roughness, susceptibility to fire, etc. Also the physical properties for each of your weapons/tools. Then you’d need a system that took all those into account and produced accurate physics, break patterns, particulation, wear, burn effects, etc. As of yet, this is not a trivial task.

Interestingly enough, the idea you’ve presented is very much in line with the original inspiration for the GameProcessor system architecture. It has always been our goal to provide the type of interactive creation you’re describing. We don’t have the tools for all this, as of today, in our system. But this “accurate surface material” technology is one of the features we’re working on. It is a more distant goal of ours, but like the slicing technology I mentioned it is working in our system already. We just need to develop the tools to make it accessible.

One of the goals will be to simplify the material-assignment process, to allow quick attribute-creation, probably through a system of presets that can be tweaked as needed. There’s a lot more about this technology I can’t share just yet, but it allows precisely the kind of interactivity you’re describing. I’ve seen a preliminary version of this in action already, very intriguing.

At present GameProcessor already lets the artist combine textures in materials to create complex surfaces, which can then be combined with variations to create some very interesting interactive surfaces. More info to come soon

Thanks for sharing the ideas. I feel that we’re at an exciting crossroads in interactive graphics, things are just starting to heat up.

Let me know if you have any other thoughts.

[EricChadwick]

Just an update to let you know we have more video content available on our website.

There are more blending examples, some sample gameplay from recent small game demos, many short videos showcasing different applications, plus some new tutorial movies showing our content-creation methods. More of the latter are on the way, that will cover how we setup bones, use materials, blend UVs, create artist-controlled interactivity, etc.

There are also recent screenshots in the gallery.

More documentation is available as well. In particular it should give you some idea of the depth of what’s already incorporated into the system. For example, tons of Concepts that have been designed so far.

There’s so many exciting things we’re working on that I wish I could share right now. Well, when it’s ready… when it’s ready…

[blankslatejoe]

holy cow guys… those videos have some awesome stuff going on. Impressive!

[EricChadwick]

A new video has been added to our website, showing our most recent project:
http://www.whatif-productions.com/video.htm

You can download the actual game here:
http://nte.navy.com

You can see some of the assets from it in this recent CGTalk thread…
Game Art from NTE: Strike and Retrieve *warning big images
http://forums.cgsociety.org/showthread.php?t=274096&page=1

[blankslatejoe]

welllll heyyy! what a coincidence… look who last posted a year ago here…

heh. small world, eric.

[EricChadwick]

Tell me about it! Big grin on my old face.