Translator required

fCD9BOdYiRM

I'll do the first three slides, my English got a bit rusty

#1: Avatar Animation in Video Games (esp. the sandbox type)

#2: Genesis of animation technology for real-time presentation

#3: in AAA titles it is based on motion capturing, from hundreds to thousands of captures for each and every character

-the avatar needs to move [naturally] in each and every game situation,

--on plain ground or downhill, inside or outside,

--with whatever equipment,

--in whatever stance: standing up, crouched or prone...

-Its motions are tied to the player's actions; the avatar needs to react immediatly!

-Biomechanics is extremely challenging. Animating the distortion of a carbody is a piece of cake to this.

Deadfast usually provides the czech/English translations for videos.

http://www.youtube.com/user/DeadfastCZ

Try contacting him.

This seems to be more technical stuff than anything related to ARMA 3. Dunno if its worth it.

Thanks for translating Flattermann. I was hoping in vain that they mentioned something about the new animation system.

they are talking generally about animations, different types of anim systems, sometimes comparing theory with A2, mostly its problems - need to have lots of animations for combinations of different weapons, stances, movements etc, hover-shooting - shooting the sky while lying down, AI switching anims from walking/running, face anims, animations are directly tied to AI, multiplayer etc

they were surprised with 1000+AI videos on youtube

They tested lots of different animation middleware, their requirements were beyond all of those libraries

last slides were interesting though: (you can't see the titles of the slides in the vid)

- Arma looks visually like modern FPS

- We can record and process practically any movement

- In spite of that, presentation is not on the level of FPS AAA titles and it is practically impossible to change

- after finishing the FPS, there is no reason to play it again, Arma is fun for months and years

- characters in Arma can do a lot of things, but that is not the only important thing in gameplay

- we are concentrating on what player needs most

- milsim, enemy behavior, terrain size, freedom, lots of simulated stuff

- the price is necessity of simplification and abstraction

- conflict between expectations of random player and a critic (paradoxically, community doesn't seem to mind it much)

- impossible to simulate on CPU in high quality

- development would be very expensive

- every change breaks something

- animation system is already unbearably complex

- for old characters, animation sets are ready

- animation system needs to be redesigned from ground up for substantial improvement

- full body inverse kinematics needs lots of CPU cycles and hard to sync in MP

- fluid changes of speed have bad impact on control - players do not want this

- basically a trap, but we will try

no mentions on difference between A2/A3, no mentions of A3, short older clips of A3

Thanks frostwyrm333, they made some interesting comments.

From what I gather from frost's post it would be very interesting to get the full translation of this presentation. Not from a fan's perspective, but a technical and technological stand-point.

There is an another one:

UdHxIz1Df94

Do you guys remember Ivan's ragdoll demonstration from E3 2011 where he throws a grenade at four dummies and they are blown onto the ground? Well take a real close look at the exact same scene in the video I posted at 44:53 - notice that one of the dummies is cartwheeled into and over a small concrete wall with all four of his limbs flailing limply in the air. That looks freaking awesome!

The PhysX Stuff alone will bring Arma gameplay and immersion to a whole new level.

Nice presentation, although a bit more tech-ey than I expected (aimed mainly at developement, not gameplay aspects) but still very interesting to watch, I really regret i didn't know about the event until it was over, I would definetely attend.

I don't think I noticed anything new regarding to Arma 3, mostly stuff we already know. A few things that may be deserving of pointing out:

• Limnos will not feature any rivers or streams. Even though BIS tried to incorporate them, they still were not pretty enough to include
  
• I really liked the part where they discussed the charatcter movement and how it should follow the basics of physics - fluid movement (no warping), fluid speed transition (no stopping from sprint on the spot), fluid acceleration... and how full inclusion of this would create other problems, namely with AI pathfinding (that is very closely tied to animations). It could be possible for the player but the result would be different of what people are expecting, it would feel a lot more clumsier and the player would feel less in control.
  
• The other thing was a problem where players expects some realistic things from the AI and something different (unrealistic) for themselves. What was specifically mentioned was player's running speed which is enhanced to about 22km/h (nor the weight of equipment neither the ground surface is even taken into account, so you can still run 22km/h across a muddy field with full gear).
  
• Another thing was a problem with long animations, namely getting in a tank. On one hand, players do not want to just press the "enter" button and be teleported inside the tank, on the other hand they do not want to go through the long animation of climbing into the tank which makes them very vulnerable. Among these lines another problem was also mentioned - how do you get out of a car stuck in a really narrow road between buildings and there is not enough space on either side of the car to open doors? Animation for crawling out through the back or the windshield?
  

I say it again, it was a pretty cool presentation, also great to hear Lord Ivan again! :)

Sounds like the new animation system is a nightmare to build and will require at least 10 years of patching to get perfect....

*grabs popcorn*

Also probably why they're not showing off the build as anything that complex drawing endless sideline banter from the fans would be annoying to no end.

No Rivers? :(

• The other thing was a problem where players expects some realistic things from the AI and something different (unrealistic) for themselves. What was specifically mentioned was player's running speed which is enhanced to about 22km/h (nor the weight of equipment neither the ground surface is even taken into account, so you can still run 22km/h across a muddy field with full gear).
  

It's interesting why they chose not to include terrain as a factor to a character's speed. I'd welcome terrain "friction" as it could change the way people and AIs would enact their strategies, it would certainly make MP/SP missions in Takistan a bit more interesting (though that's more of a elevation issue then what type of ground surface you have). But there is the thing with AIs, however: how would you make them choose paths that doesn't involve stomping trough a wet marsh for the next few hours, but will still choose to go trough that marsh if it's to their advantage?

There is also the thing with the illusion of speed, and I think BI could reduce the speed without having people feel they are running/jogging/walking slower. I find that the default FOV create a feeling of slower speed if my character is driving or jogging around compared to playing with a broader FOV - of course, not to the extent that I'm getting bored or thinking it's taking too long. This illusion seems to be quite a lot of less if I change the default FOV to a broader FOV (using the minus "-" key). This does, however, make seeing enemies at a distance a bit harder. But I don't see much of a problem if the current feature of zooming in and out is still possible. The mid-way solution is to have a FOV in between the current default view and the view you'll have if you press on the minus button.

Out of curiousity, does BI have a picture of the river project? It can't be THAT bad. :bored:

Regardless on how much we might whine about it, it would be too late to implement a river if BI ends up showing a picture. :p

Not sure if I understand the technical limitations for rivers when they have 500Km squared ocean around the place, just throw in a simple current to move waves one direction downriver at a reasonalbe speed so people/boats who turn off engine and float will drift when sitting still and swim/boat slower when going against current by a constant factor of say 3knots ressistance and faster than normal when with current. I dont think we need Real time dynamical stream flowgraph technology. Just use the same physics model they have for ocean but with different some values.

Not sure if I understand the technical limitations for rivers when they have 500Km squared ocean around the place, just throw in a simple current to move waves one direction downriver at a reasonalbe speed so people/boats who turn off engine and float will drift when sitting still and swim/boat slower when going against current by a constant factor of say 3knots ressistance and faster than normal when with current. I dont think we need Real time dynamical stream flowgraph technology. Just use the same physics model they have for ocean but with different some values.

Problem with rivers seems to be far more deep than just "not wanting" to incluce currents. Think of getposASL (which some use to not spawn stuff in water, etc.), the ASL value would need to be rethought completely, as rivers likely won´t be on the visitor 0m level, for that they´d have to tinker with probably a dozen other points of intersection - and then theres:

- every change breaks something

No rivers decision sucks, but it´s understandable from a "coming Q2 2012" business oriented point of view.

edit: And that´s thinking of rivers! Not bays as we know from nogova or pools of water, be it objects or simple dips in terrain at exact sea level.

Guess with rivers we all would want something thats appropriate for wild water rafting or a scenic rhine tour, water flowing down from mountains, that kind of thing.

@froggyluv they said that they will try (full body inverse kinematics), the steering wheel is anims are probably something like that. but thats all summer 2011...

rivers are actually interesting, there was something like river in A1 and there is something that could actually pass for a river in proving grounds.

Another fun fuct, if you go north outside of chernarus on a plane you will see something that basically is a river, although a bit weird but still interesting, probably some leftover from satellite data/terrain processing.

Sounds like the new animation system is a nightmare to build and will require at least 10 years of patching to get perfect....

*grabs popcorn*

Also probably why they're not showing off the build as anything that complex drawing endless sideline banter from the fans would be annoying to no end.

I agree although 10 is abit much, I say 5 years.

i wouldn worry, insiders seem to be happy with development

@froggyluv they said that they will try (full body inverse kinematics), the steering wheel is anims are probably something like that. but thats all summer 2011..

I would really like to see somekind like this when comes to animations:

The body actually adapt hisself when comes to terrains and small spaces (like against the wall) instead of the stretching, clipping and weirdo positions that ARMA have now. (Ok, not the best photos to demonstrate that but those that I have at hand)

And about the lack of rivers: they throw the towel when comes to "AI pathfinding+Bridges"? :D

if their arent river with moving trigger fingers im boycott arma 3

@Smurf - that actually is the definition of inverse kinematics

http://en.wikipedia.org/wiki/Inverse_kinematics#Inverse_kinematics_and_3D_animation

Im afraid that if I will see hands properly aligned with ladder Ill get a heart attack.

• I really liked the part where they discussed the charatcter movement and how it should follow the basics of physics - fluid movement (no warping), fluid speed transition (no stopping from sprint on the spot), fluid acceleration... and how full inclusion of this would create other problems, namely with AI pathfinding (that is very closely tied to animations). It could be possible for the player but the result would be different of what people are expecting, it would feel a lot more clumsier and the player would feel less in control.
  

To me, the current ArmA2 system of complete stop to instant full bore sprint to instant complete stop again is about as "clumsy" as it gets! Please BIS, any attempt to model the acceleration from stop to sprint and back would be a vast improvement over what we have currently, regardless of what players may be "expecting" in the short term.

If you haven't noticed how abrupt things are currently in ArmA2, setAccTime to slow-motion, go into 3rd person, use free look to view yourself from the side, and start sprinting. The instant you let off the forward key, your character will cease any forward motion. The ridiculous part is, the slow-down animation still plays, even though you aren't moving forward anymore! Looks especially silly in slow-mo.

In any event, there's definitely lots of room for improvement with regard to human body acceleration/deceleration (player and AI) in Arma3. Let's hope they get it right. :)

• The other thing was a problem where players expects some realistic things from the AI and something different (unrealistic) for themselves. What was specifically mentioned was player's running speed which is enhanced to about 22km/h (nor the weight of equipment neither the ground surface is even taken into account, so you can still run 22km/h across a muddy field with full gear).
  

This also explains BIS's rational in having two different animation speeds for the same animation. One that you see, and one other players will perceive. It you're not quite sure what I'm talking about, try this:

It is quite comical to have a single ArmA2 multiplayer game running side by side on two computers, with two playable men (one you control and the other a stationary spectator). Go into 3rd person, and run around. Although your ground speed is the same on both computers, the animation speed varies greatly between the computer you're controlling the man on, and the spectator computer. What you see is a sped up version of what they see. Move around in a crouch to really see the discrepancy.

CfgMovesMaleSdr values relSpeedMin and relSpeedMax control this, I believe. In addition to these adjustable values, we could really use an "animationSpeed" value to simply speed up or slow down the animation speed (different from the "speed" value, which currently affects both speed of animation and ground speed).

Hopefully with their release planned for Q4 2012, they will have a great animation system. I personally believe they are pretty much done with the environment, models and story; they are just working on stuff like animations and inner workings of the game. Also, Take On Helicopters already has inverse kinematics from what I hear.