Even if you could still get hold of a licence for EKI One you would still be unable to do anything with it without the SDK. I detailed this in a previous post above:

Quote:EKI One is no longer being developed or supported although it is seemingly still available to buy. This has been the case for about a year and a half now. The company that bought out Artificial Technology (the creators of EKI One) have their own competing product range and to my knowledge have not agreed to the release of the C++ SDK since then and this is required to interface EKI One successfully into game engines.

I was provided with the SDK through agreement with Artificial Technology due to the integration work I did with Leadwerks. I am unfortunately not licensed to distribute that to others. So that is an effective show stopper for anyone not currently in possession of a licensed copy of EKI One and a legal copy of the SDK.

Masa, as far as I can tell, have completely buried EKI One now that their replacement software Masa Life is available for evaluation. Unfortunately I'm guessing one has to negotiate a price with them as no price is stated. This leads me to believe its probably targeting established game studio clients initially. There was talk of an Indie version but I have seen no evidence of that to date.

I don't have any experience of other 'off the shelf' AI engines but bare in mind any engine would need to be interfaced into Esenthel too. Certainly interfacing EKI One is non trivial. You are then looking at a 3 month learning curve to use it effectively

Pixel Perfect : Thank you for taking time to explain everything.

Yea.. Im looking into this also.. Thanks!

Having completed all of the initial interfaces between my AI engine, Esenthel and my animation routines I decided to give the whole system a performance test by having the AI spawn 100 NPCs and have them path find and steer to randomly generated locations in my test level.

I have to say I'm pleased with the results I'm seeing. I'm getting between 35 and 50 fps (locked to 29 fps in the video below).

Looking good.

Also, at 00:38. Are they going in for a bump?

Thanks Tottel.

The 0:38 occurrence is two NPCs walking right up to the wall, sure looks like they are about to collide . I've noticed the function that produces random positions often chooses locations on the edge of nav mesh boundaries. I need to move the nav mesh out from the walls more. They do actually stop with their faces against the wall but it leaves the weapon embedded in the wall! In normal game play there would be a lot less use of random generated locations anyway.

The path finding and steering system is not perfect by any means, it does cause some unnatural looking movement at times and when extremely congested collisions can occur, but then I've never seen a system yet that's infallible, especially as the NPC density grows. The physics deals with collisions pretty well though.

Early days yet

(10-09-2013 08:28 AM)Pixel Perfect Wrote:  Thanks Tottel.

The 0:38 occurrence is two NPCs walking right up to the wall, sure looks like they are about to collide . I've noticed the function that produces random positions often chooses locations on the edge of nav mesh boundaries. I need to move the nav mesh out from the walls more. They do actually stop with their faces against the wall but it leaves the weapon embedded in the wall! In normal game play there would be a lot less use of random generated locations anyway.

The path finding and steering system is not perfect by any means, it does cause some unnatural looking movement at times and when extremely congested collisions can occur, but then I've never seen a system yet that's infallible, especially as the NPC density grows. The physics deals with collisions pretty well though.

Early days yet

I wouldn't worry too much about the weapon clipping the wall. Seen it in so many AAA games that it's probably ignorable to the players.

Looks good

It looks really really good.
As Scarlet said, you shouldn't worry about that weapon in the wall. that became an acceptable issue in pretty much every game I know.

But they really look damn well. The game starts to look "alive", which is awesome.

I'm pleased to see you are making steady progress and without too much trouble. Good work Pixel, do continue.

As for the weapon clip issue, replace them with lightsabers and you can label it realistic.




P.S.: just a little heads up - I've been meaning to spam you with some integration questions regarding EKI One for some time now.. expect a PM soon

Cheers guys!

Yeah, I'm not too worried about the weapon issue, as you point out it's not like we don't see that all the time in games.

It's been a great opportunity having moved engines to really take my time and ensure I write things properly this time round. I learnt a lot from my first attempt at putting game mechanics together but it's so much nicer to construct the code again from the advantage of 'hind sight'. I don't get a lot of time to work on this project but I'm determined not to rush it this time but rather code it well and ensure it's robust and flexible. I feel I have a really solid foundation to build on in Esenthel Engine and that motivates me to keep pushing the project on. I'm also inspired everyday by seeing what other people are doing with the engine too! There are some great projects.

I'm currently working on getting lip synced animation in before then moving on to the real AI design which is the part I'm most looking forward to. I really want this to be a living breathing believable world with sophisticated AI. At some point before then I need to give the level a face lift too by adding all the furniture, doors etc to the buildings and more foliage, all of which will be needed before the final nav mesh can be generated. There's a whole market area to go in too. I'll then have sufficient to test out all of the AI mechanisms I need to develop including ambient behaviour for villagers and animals, specific key game character AI and character interaction, cut scene functionality and a mission system.

Special effects/eye candy, a detailed player character, fancy interfaces, full game design will come last. Without the essential game mechanics up and running and pretty well tested there seems little point in adding the gloss and implementing further level design.

(10-14-2013 06:40 PM)para Wrote:  I've been meaning to spam you with some integration questions regarding EKI One for some time now.. expect a PM soon

Feel free, always happy to discuss these things especially seeing that you've helped me many times already

Pretty cool. Keep up the good work.

I now have basic lip synced speech up and running.

Excuse my 'programmer' rigging skills, this was my first attempt at facial rigging of a model. I'll hopefully get some artists to do a better job at a later stage.

wow, very cool how difficult was it to make?
how are you feeding the animations based on the sound?
is it just one prepared animation for the entire sound? or do you have different animations/face expressions for different A E O U .. characters and blend them together based on sound data? if so, do you analyze the sound wave or work on text data and setup the animation based on the text itself?

Nice job

Thanks guys

(10-23-2013 11:37 PM)Esenthel Wrote:  wow, very cool how difficult was it to make?
how are you feeding the animations based on the sound?
is it just one prepared animation for the entire sound? or do you have different animations/face expressions for different A E O U .. characters and blend them together based on sound data? if so, do you analyze the sound wave or work on text data and setup the animation based on the text itself?

I wish I could take credit for it but the clever stuff is done by a company called Annosoft who provide the tool to analyse the wav speech files and output a text data file containing a series of time stamped phenomes representing the speech.

The models to be animated just need to have these basic phenomes available as poses (single frame animations) to enable you to then drive the speech animation from the data. Using standard interpolation between each frame smoothes the mouth movement and gives a reasonable illusion of speech.

I'm mapping the 39 different phenomes it produces to a smaller subset of 10 which dramatically cuts down on the amount of animation work required but still gives a reasonable approximation of lip synced speech. You can increase this to improve the realism.

You can buy the convertor directly from Annosoft who produce an SDK or a version is also licenced to TGC in their Dark Voices package.

In my implementation I simply build a configuration object containing speech files and associated speech data in the editor which is used to store the UIDs and a unique numeric identifier in my ConfigManager.

I then call a PlaySpeech function in my NPC object from my AI engine passing the desired identifier. The sound file is loaded and executed and my animator object associated with the NPC handles the loading and processing of the phenome data and the blending of the normal and facial animation.

The UIDs for both files are provided by the ConfigManager by performing a lookup using the provided identifier.