Welcome, Guest. Please login or register.

Login with username, password and session length

 
Advanced search

1411273 Posts in 69323 Topics- by 58380 Members - Latest Member: bob1029

March 28, 2024, 02:17:21 AM

Need hosting? Check out Digital Ocean
(more details in this thread)
TIGSource ForumsCommunityDevLogsWobbledogs
Pages: 1 [2] 3 4 ... 37
Print
Author Topic: Wobbledogs  (Read 229688 times)
ActualDog
Level 3
***



View Profile WWW
« Reply #20 on: February 25, 2016, 06:47:43 PM »

This is looking spetacular. I hope you can pet the pets because I can't wait to pet the pets.

Thank you. Being able to pet the pets is a very important feature and I will do my best on it when the time comes.

Don't be coy. I knew the answer to this question before I even read your response. You plan on keeping the graphics basically the same they are now, because it looks fucking hilarious with these animations. For what it's worth, I stand by this decision 100%. :')

lol, to be fair I really do want to experiment a bit with the graphics later on. But yeah, in all honesty I'm very into the abstract physics beast aesthetic for this.

How much variation will the genetics allow? Will they all still be dogs or will you allow for things with different physically shaped bodies? I guess making procedural animation would be pretty tricky if you did.


Most likely I will only be making this single species. It's going to be a ton of work to get it to where I want it, and the only reason I can justify all the time I'll end up spending on it is because it's the only thing I really have to put that time into. I have some plans for aesthetic modifiers based on genes, but I dunno how far I'll take it into the physical realm. Like you mention, any structural changes will require new fine-tuning, so varying that stuff too much is tricky.

I have been playing with the idea of having different bodies for these guys depending on age though. Not gonna think too much about that until things are farther along though.

As for the graphics, you'll probably have a better idea once the coding is done.  If you want to keep the gameplay comical then the current cartoony, colorful style would fit well.  If you want to go in a more serious direction,

Have you seen the following robot videos?  They seem to have some resemblance in terms of movement.

Definitely want to keep the whole thing goofy! And yeah, all the Boston Dynamics critters are great. Super fun to watch them in action! Honestly makes me pretty happy when people bring them up in the context of this project, because their stuff is so ace.


The main thing I did today was to start in on a system for leg fatigue. The colors are just for debugging, but I have legs getting tired the more torque they apply. After a certain time period below a certain fatigue threshold, or if they reach their fatigue limit, they give out for a bit. Additionally, legs exhert less torque when fatigued, and less fatigued legs will try to exert more torque to compensate up to a point. It's not done yet, but you can see the basics in action.


Logged

io3 creations
Level 10
*****



View Profile WWW
« Reply #21 on: February 26, 2016, 10:22:02 AM »

Most likely I will only be making this single species. It's going to be a ton of work to get it to where I want it, and the only reason I can justify all the time I'll end up spending on it is because it's the only thing I really have to put that time into. I have some plans for aesthetic modifiers based on genes, but I dunno how far I'll take it into the physical realm. Like you mention, any structural changes will require new fine-tuning, so varying that stuff too much is tricky.
Would the current code allow at least size variation?  If variables can be adjusted linearly than that could work but I'm guessing some calculations are not linear so even that might not be that simple?
Logged

ActualDog
Level 3
***



View Profile WWW
« Reply #22 on: February 26, 2016, 05:52:10 PM »

Would the current code allow at least size variation?  If variables can be adjusted linearly than that could work but I'm guessing some calculations are not linear so even that might not be that simple?

Theoretically, but I haven't tested it yet! Each dog's movement pumps through a central function call so I can scale it however I want, but ultimately I'm still at the mercy of PhysX and it's possible that it won't be as simple as just scaling all the values. I would love it if it was as simple as a linear scale, but I kind of expect it to not be. Will definitely play with it at some point though!


Nothing crazy new today, but I did a few things.

First of all, I updated how stability steps work. The rules for when to take a step previously were kinda weird and resulted in wobbly steps in weird situations when they shouldn't have been required.

The new rule is that a limb will take a step for stability in the case that its attached body's rotation is within a window I've defined, the limb is not grounded, and at least one other limb on the dog IS grounded.

It's not perfect yet, but it works much better!



I'm still getting little flutter steps sometimes though, so I think my next pass I'm going to try making it so that a limb only takes a step for stability in the case that its adjacent limb is grounded, instead of just checking to see if ANY limb is grounded.

After that I took a pass at rotation restoration. I'm not currently happy with how dogs behave in the air. They're snapping to perfect rotations too quickly and it looks unnatural. I added a few things so that rotation restore when no limbs are grounded is dampened. Again, it's not perfect yet, but it's feeling much nicer. We did a few flips to test things out. I'm manually giving an initial rotation torque to start the flip, but the correction is all autonomous. Before this change, the dogs would flip back to their default rotation in a much more rigid manner.


The last thing I did today was to add bucking behavior on belly collisions.



Super simple after the bucking I added a few days ago, but it means they won't get high centered. It also means that the dogs untangle themselves even easier than before.
Logged

jctwood
Level 10
*****



View Profile WWW
« Reply #23 on: February 27, 2016, 03:43:24 AM »

It is extremely exciting seeing this develop. Have you thought about having the pets learn some of their abilities during gameplay, say they only start bucking after a few minutes of having the object on their back.
Logged

ActualDog
Level 3
***



View Profile WWW
« Reply #24 on: February 27, 2016, 08:54:16 AM »

It is extremely exciting seeing this develop. Have you thought about having the pets learn some of their abilities during gameplay, say they only start bucking after a few minutes of having the object on their back.

Thanks!

And yes, but kind of the opposite for core behaviors like bucking. I'm planning on having certain behaviors that all my dogs will do because it's in their species makeup. But, I want you to be able to train away or adapt these behaviors (within reason). Taking the bucking as an example, you'd never be able to get these guys in their natural state to like, wear a hat. They'd freak out. But if you work with them for a bit and get them used to having things on their backs with nothing bad happening, they'll get used to it and stop going crazy every time it happens.

I also want a way to train completely new behaviors. I haven't worked out all the details yet but I have a few ideas about it, so more info as I get to the point where I can prototype that stuff.
Logged

io3 creations
Level 10
*****



View Profile WWW
« Reply #25 on: February 27, 2016, 11:12:00 AM »

The standing up looks pretty good.

But I wonder about the mid-air rotations when the dog is pushed up vertically and then suddenly starts rotating it's body.  That doesn't seem very natural.  With internal gyroscopes or thrusters that kind of movement is possible.  However, cats can do something similar https://en.wikipedia.org/wiki/Cat_righting_reflex  that seems almost unnatural but they do contort their bodies a lot more. 

If you're looking for possible behaviour (especially the mischievous kind), these types of videos may help  Grin


Logged

ActualDog
Level 3
***



View Profile WWW
« Reply #26 on: February 29, 2016, 03:41:03 PM »

But I wonder about the mid-air rotations when the dog is pushed up vertically and then suddenly starts rotating it's body.  That doesn't seem very natural.  With internal gyroscopes or thrusters that kind of movement is possible.  However, cats can do something similar https://en.wikipedia.org/wiki/Cat_righting_reflex  that seems almost unnatural but they do contort their bodies a lot more. 

Yeah, I'm definitely not completely happy with their air control yet. They don't currently take momentum into account when determining correction dir, but they should. That'll be the next thing I tackle when I get back to it! I also want to experiment with losing some of their limb tension when in the air. That being said, the sudden rotation you're seeing after they get pushed up is triggered by me for testing purposes. They're only correcting the rotation, not initiating it.


I don't have much new to show today, I've mostly been doing behind the scenes stuff: bugfixes, optimizations, re-org, etc.

I did make some time to revisit leg strength a bit. I wasn't happy with how rigid legs were when they gave out. Now they lose most of their tension and all of their bounciness.



He doesn't collapse instantly in the above gif since his setup is fairly stable, but in a situation where he's actually using his legs in such a way that they're in danger of giving out, the ragdoll effect kicks in nice and smooth.
Logged

HIDEJI
Level 0
***



View Profile WWW
« Reply #27 on: February 29, 2016, 04:18:46 PM »

I love this last gif. "Oh no, he killed it?!" Then it just bounces back up with it's stupid goofy face again. These things look so charming and bully-able, hahah. :')
Logged


If you like what you see, please help spread the word on twitter! Thank you! Beer!
fall_ark
Level 2
**



View Profile
« Reply #28 on: February 29, 2016, 08:02:11 PM »


Don't be coy. I knew the answer to this question before I even read your response. You plan on keeping the graphics basically the same they are now, because it looks fucking hilarious with these animations. For what it's worth, I stand by this decision 100%. :')

lol, to be fair I really do want to experiment a bit with the graphics later on. But yeah, in all honesty I'm very into the abstract physics beast aesthetic for this.


Have you considered making the face/body/leg feature customizable by the user? It could be a few MSPaint-like panels for each specific spot (both eyes, nose, mouth etc.), or even a single texture map for the user to draw on in-game. It doesn't need to be perfect, heck it's better if it's going to be a mess now and then.

Maybe?
Logged

Chinese localizer and influencer. Translated Dead Cells, Slay the Spire, The Count Lucanor, Katana Zero, Dicey Dungeons, and involved in the localization of Reigns, The Curious Expedition, Desktop Dungeons, etc.
If you have questions about Chinese loc and publishing etc., find me at Twitter @SoM_lo
michelmohr
Level 1
*



View Profile WWW
« Reply #29 on: March 01, 2016, 01:18:46 AM »

I feel the little poor things are stress relief balls for you. Tossing cubes at them, using them in weird experiments...
Do some more gifs Wink
Logged

RujiK
Level 2
**



View Profile
« Reply #30 on: March 01, 2016, 07:38:19 AM »

Really beating a dead horse in that last gif!
Logged

io3 creations
Level 10
*****



View Profile WWW
« Reply #31 on: March 01, 2016, 10:01:34 AM »

Yeah, I'm definitely not completely happy with their air control yet. They don't currently take momentum into account when determining correction dir, but they should. That'll be the next thing I tackle when I get back to it! I also want to experiment with losing some of their limb tension when in the air. That being said, the sudden rotation you're seeing after they get pushed up is triggered by me for testing purposes. They're only correcting the rotation, not initiating it.
Oh, I see. Smiley 

Overall, depending on the direction of the game, sometimes it can be helpful to be able to "manipulate" physics.  For example, in realistic car racing games, once you start flipping over then there isn't much you can do.  But in some arcade style car racing games, you can even manouver mid air. 



I did make some time to revisit leg strength a bit. I wasn't happy with how rigid legs were when they gave out. Now they lose most of their tension and all of their bounciness.



He doesn't collapse instantly in the above gif since his setup is fairly stable, but in a situation where he's actually using his legs in such a way that they're in danger of giving out, the ragdoll effect kicks in nice and smooth.
When you posted the previous pics, I would've liked more ragdoll physics applied to the legs.  Of course, I know that the game's in early development and maybe you didn't get around to work on that. It's nice to see that you do consider that aspect.  The result so far looks good.
Logged

ActualDog
Level 3
***



View Profile WWW
« Reply #32 on: March 01, 2016, 04:40:35 PM »

I love this last gif. "Oh no, he killed it?!" Then it just bounces back up with it's stupid goofy face again. These things look so charming and bully-able, hahah. :')

Right now they're very easy to mess with, but I wanna give them emotions and memories. Eventually throwing blocks at your pet will probably be a lil' sad!

Have you considered making the face/body/leg feature customizable by the user? It could be a few MSPaint-like panels for each specific spot (both eyes, nose, mouth etc.), or even a single texture map for the user to draw on in-game. It doesn't need to be perfect, heck it's better if it's going to be a mess now and then.

I couldn't find a great image, but I'm picturing that segment in Okami where you draw a mask for the wolf. It's a fun idea! I want some amount of variation to come from each dog's genes, but it might be neat to have something like a cosmetic shop where you could take your dog and give it a makeover. That would be way far in the future, though! I'm not anywhere near a point where I want to think too much about it yet.

I feel the little poor things are stress relief balls for you. Tossing cubes at them, using them in weird experiments...
Do some more gifs Wink

gifs aren't stopping anytime soon! I'll be working on building up behaviors and AI for these guys for a while still.

Really beating a dead horse in that last gif!

 Hand Thumbs Down Left

When you posted the previous pics, I would've liked more ragdoll physics applied to the legs.  Of course, I know that the game's in early development and maybe you didn't get around to work on that. It's nice to see that you do consider that aspect.  The result so far looks good.

Yep, I have an ever-growing Trello project full of little notes and observations about stuff like this. Literally nothing about these guys is polished to a point where I'd say I'm done with it yet. There's only so much time in the day!


No new progress to show right now, but I'm starting in on some basic AI. All the current behaviors I've been posting are either instinctual (bucking, standing up after being knocked over, stability steps), or triggered by me (walking, standing in a pyramid). I'm busy setting up that system so it'll be extendable and easy to work with, and I hope to have some simple behaviors going by tomorrow. Once I get the basics out of the way, I really wanna try and get the dogs interacting with each other!
Logged

ActualDog
Level 3
***



View Profile WWW
« Reply #33 on: March 02, 2016, 04:36:49 PM »



This is from last night, but it shows my first pass at piping basic follow behavior through the AI system I'm building. As with everything, this is all super WIP.

I've gotten a lot more done today, but the visible differences aren't enough for a new gif so you just get this one for now.
Logged

crusty
Level 2
**



View Profile WWW
« Reply #34 on: March 02, 2016, 05:44:15 PM »

hooooooooch
Logged

Caravan: devlog | website
Bluebutton
Level 1
*



View Profile WWW
« Reply #35 on: March 02, 2016, 11:56:43 PM »

This is technically very impressive stuff!

What's the actual gameplay going to be? Build your own creature? Obstacle courses? You should try putting legs on all sides of those things!

Following for sure.  Gentleman
Logged

ActualDog
Level 3
***



View Profile WWW
« Reply #36 on: March 03, 2016, 06:29:20 PM »

lol, thanks guys.

What's the actual gameplay going to be? Build your own creature? Obstacle courses?

Actual gameplay isn't 100% locked down yet, but the main idea is that you'll be able to raise these guys as pets, train and interact with them, and then enter them in events. Specifics are evolving with the tech right now, but I'm feeling confident and trying to leave room for options as I'm developing the dogs right now.

You should try putting legs on all sides of those things!

One of the first things I did when starting out was to put legs on all sides of the dogs, actually. My original idea was just to build a super generic creature builder that would adapt to whatever params you put in. At this point there's too many assumptions about these guys' leg setups to just add more to them, unfortunately. Give me a day or two and I could have something fun working, but I don't have that time to spare right now!


Progress on the AI system has been steady but there's a lot to do. It's the biggest individual system in the game so far and I need it to be extendable. The dogs themselves are complex but built up of multiple straightforward behaviors, so no one part of them is too crazy.

The basic idea behind the AI is that dogs get lists of possible behaviors. Each behavior has self-contained info about priority, requirements, execution, etc, and each dog's AI controller evaluates everything and figures out what to run. Not too complex in theory, but there's a ton of details to sort out.

So far I've hooked up a few super straightforward things to test with.





This is a *very* early pass at playing. It's embarrassingly simple and serves as more a test for the system than the behaviors themselves. Literally all that's going on is that dogs will find a nearby dog to target and will walk towards that dog's face. If a dog tips over it takes a sec to right itself, and each dog regularly re-evaluates its state to see if there's something better it could be doing. I kinda think even this super simple behavior is fun to watch, though, and it's not too hard to imagine them actually playing with each other.

To be honest, I don't want the final interactions to be too much more complex than this. I think the fun of the physics animations comes from layering behaviors and leaving room for them to come together in funny ways, not from scripting extremely complex one-shots.

I've just started in on a more complex play scenario (though still relatively straightforward and keeping in line with what I just said), and hopefully it'll be in a place where I can show it off soon!
Logged

Hephep
Level 1
*



View Profile WWW
« Reply #37 on: March 03, 2016, 11:13:59 PM »

Please be gentle with them  Screamy
Logged

ActualDog
Level 3
***



View Profile WWW
« Reply #38 on: March 04, 2016, 07:04:47 PM »

Please be gentle with them  Screamy

It's ok, they are very hardy.

Some progress. I've got the base for paired behaviors set up. Dogs need other dogs to be either not running a behavior or running an interruptible behavior in order for them to engage in stuff like this. Still need to work on a "play bow" to replace the bucking you're seeing, but it's getting close to what I actually want.



After I get this stuff a little more solid I'm going to try and get in some variable outcomes here. Like, a dog might decide not to return the play bow and walk away instead of initiating play, or a really excited dog might just jump straight into a play session without asking, which could make the other one mad, etc.
Logged

ActualDog
Level 3
***



View Profile WWW
« Reply #39 on: March 08, 2016, 03:33:24 PM »

Progress has been slower recently, but here's a quick update to show I'm still at it.

Mostly I've been working on AI stuff; tweaks and system adds but nothing really worth showing yet.

I was trying to get a "play bow" in the game (still not in a good place), and I got them to flip.



Also, https://twitter.com/marcoapc_art contacted me a few days ago and wanted to make me a model to test out, so he sent over a really nice vox dog. I did a quick and dirty job of overlaying it on top of my cubes and the result's pretty funny. Since he gave the whole thing to me in parts, I also added some springy joints to the tail, head and ears to make them flop around (though that's messing with the balance a tiny bit).



I don't want to use voxels for the final art, but I like the goofy dog and it was really great to be able to test out this flow. I think this type of model overlay could work really well if I decide to go that route.
Logged

Pages: 1 [2] 3 4 ... 37
Print
Jump to:  

Theme orange-lt created by panic