 |
welcome, guest |
| Development Forum | ||||
 |
| |||
 Brynn  |
Being able to change limit perimeters like amount of kritters, progression of seasons, number of injectables, etc. could have interesting results. Sure, increasing maximums past "reasonable" limit could cause processor problems and would be hard to test, but I am envisioning theoretical simulations on university supercomputers to observe population genetics. It's always a feature that you can leave in with the disclaimer "You can mess with it, but if you do, it's no longer my problem." |
|||
 Geat_Masta |
I don't think it would be possible to change progression of seasons. Here's how it works currently: 24 minutes in a day 6 days in a week 5 weeks in a month 4 months in a year The time of sunrise/sunset changes with the world clock, so winter has shorter days than summer, BUT the functions for this need the axial tilt of the planet you're on, and the latitude you're at. So different metarooms can have different seasons by changing the hemisphere they're in/tilt their at. The problem with creature limits is: I don't think they're imposed by CPU speed, but RAM speed. And that changes things. CPUs have gotten significantly faster since C3/DS were released, the same isn't true of RAM. So while I can get a huge number of creatures in a debug world, it's unlikely that will be possible in a world full of cheese and critters and such. The other major problem is that scripts and time aren't tightly linked. E.g. if a script get into an infinite loop, all scripts will stop because they need to wait for that one to finish, but the physics and biochemistry will just keep on going. In typcial use cases this isn't a problem, because scripts typically finish in a few nanoseconds, but if the limit is too high, the result isn't lag as much as the creatures can't learn at all, because the scripts and timers have gotten too out of sync, so they can't take any actions. |
|||
Geat_Masta |
It shouldn't take me 3 days to get particle effects to use instancing instead of a loop, but getting any answer on the openGL forums takes forever, (on a good day). |
|||
 Prodigal Sock Ghosthande  |
Looking at stuff like this has me thinking. Correct me if I'm barking up the wrong tree, but I'm assuming the Kritters/Lagoraptors/whatever we're currently calling them will be able to speak in some fashion. Creatures Adventures did this interesting thing where a Norn's face was animated when it ate, so it looked like it was chewing. Granted, this only worked because the ears were separate... so it could re-use the "scared" expression for the chewing animation without the ears flopping around. The effect looked pretty neat. Norn audible speech is already linked up to the text they're saying, so that (some of) the letters in the words determine which sounds they make. If that's possible... what about syncing their facial expression, too? Drawing Kritters with eyes and ears on separate layers would not be any more work than drawing whole heads, and I doubt very many new facial expressions would be necessary for speech that wouldn't also be usable for regular emotions. Since people have already brought up having ears and eyes be separately hereditary, and this would serve the same goal, I'm going to pursue this line of thought unless you nix one or both ideas. It wouldn't take much more effort for me, so I'm happy to do it--and if it doesn't work out, no harm done.  |
|||
 Bifrost |
Yes! All of that! (And horns and hair separately as well.) We'll need to make new sounds. Personally, I'd like to have voices match spoken syllables, at least to a certain degree. So what you read reminds you of what you hear. Could voice pitches, raspyness and melodies be a genetical thing? And we'll need to figure out how mating will work. Kisspopping is a brilliant way of making the game innocent, but that's sort of a Creatures exclusive thing. I'm thinking a mating position where two krittoraptors stand back to back, interlocking or just ruffling their tails. Oh! Oh! Oh! Would it be possible to give the krittoraptors the ability to sing? Perhaps as a sort of courtship, or to calm their little ones or something. |
|||
 BaffleBlend |
Bifrost wrote: I think going too far on that would backfire and would sound creepy/annoying rather than endearing. But then I thought about what pulled it off successfully: Animal Crossing. The "Animalese" speech there matched the text of the dialogue while still being garbled enough to stay out of the Uncanny Valley. Maybe we could do something like that? |
|||
Geat_Masta |
I eventually got the particle effects to work, I'll share the problem because I love it so much: "the 'number of bytes between entries' doesn't refer to the gap in bytes between the first entry and the next entry, so if it's 0 then the 2nd entry directly follows the first, while if it's 1 they overlap, and 9 has 1byte between them." WHAT! And yes, this is the STANDARD for how these libraries are, and audio libraries are no better than the graphics libraries. So writing programs with either means running into dozens of problems like that. I'm not noticing a significant difference between the two, so I'm thinking I saw so much about performance gains regarding batching because mobile development is so popular these days. Anyway, I haven't added sound because I only have the vaguest idea of what's going on when I play a sound. The problem is that I need to delete a sound when it finishes playing so it doesn't just sit there occupying RAM. And I don't know how to tell when it's finished, really. So therefore I also won't know when a syllable is at what part of it's mouth movements. So I'll file that under 'cross that bridge when we come to it' rather than making any decisions now. |
|||
 Merboy |
I ADORE the idea of the creatures singing for courtship. I've been watching Planet Earth because I'm a dork and singing is such a universal language anyway and it makes perfect sense. I could also record some voices and edit them with my mic and software if we need. I also think having the pitch and speed of the voice clips altered to specific breeds or individual creatures would create a more unique experience for the player. However I also know sound files can be pretty big and even little clips can add up quickly. But singing? Yes absolutely, I'm in love with that idea. The Lantern Light.com |
|||
BaffleBlend |
Merboy wrote: I think I know how we can avoid making songs take up too much extra space: text-based .abc notation files. That's what Starbound and LotR Online uses for custom music that characters play on their instruments. By using the creatures' voices as the instrument when reading from the files (perhaps randomizing the syllable used per note), it would pretty much just use the voice resources. |
|||
Geat_Masta |
test of particle emitters they don't respond to gravity yet, but you get the idea! Don't get so excited about the singing yet, I have no idea if I can do it. |
|||
Bifrost |
That is awesome! I've been working on a new waterfall in the new world. I'll scrap the actual water and replace it with what's behind it. As for singing, a "worst" case scenario could be that we just save a few tones as sound files and have them "spoken" into a melody. If we get individual pitching, we could get away with half the sound files, not having to have one set for males and one for females. |
|||
Geat_Masta |
@Bifrost only make a line drawing right now. I want to see the basic layout before you start doing things. |
|||
Bifrost |
Okay. I tend to work in the opposite way; details first and structure later, moving things around to fit together. |
|||
 Lodestar Doringo |
Ghosthande wrote: It was also likely intended to have a crying animation, you can see the remains of it in the head sprites, along with some other interesting unused assets. |
|||
Brynn |
Looks like we're discussing social behaviors for the kritters! I'm excited! (I have a lot of things I would like the lil guys to do, okay.) I'm fully aware that building on their intelligence and social behaviors will be far down the line, but I wanted to throw some ideas out there for the community to think about: Singing: Would hopefully be a behavior they start displaying at a young age in a reduced form. As they age, their singing becomes more elaborate, until as adults they have full "musical phrases". Each kritter might have a randomized set of notes that expands as they progress through life stages, so to humans it might sound dissonant, but to another kritter it might be what they're looking for! Recognize other kritters as intelligent individuals: The greatest annoyance to Norns were other Norns who saw them as play objects or sources of stimuli. Hopefully Kritters will be able to recognize other kritters have their own decisions to make and are polite about asking for things like attention, food, or play. This is a lot to put in their brains, I'm aware, but if we're going to improve on the current Creatures installments, I definitely want to see this sort of thing in some capacity. This sort of "it's a living thing!" recognition would also apply to the pointer. the word "play": A more complex concept I thought up while typing up the previous. Maybe the kritters can think of objects, specifically "living things", as more than push/pull/eat. "Play" would not be so much of a command as much as a request towards another living thing. If a kritter is bored or lonely and there's another living thing around, whether it's the pointer or another kritter, they can approach with "Play?" which is a low-drive-priority request for the other kritter. If the other kritter isn't busy taking care of themselves, they can pay attention and interact. This sort of request would have a higher priority coming from younger kritters that are offspring. ...I just realized I described something that the Canny Norns/Nova Subterra have been grappling with for years (Socialization vs other drive priority). oops. Also, I realized that Ghosthande's creature reminds me of a jerboa standing upright which made me look at their skeletal structure. They have such tiny t-rex arms!! |
|||
Geat_Masta |
@Brynn, mostly not possible. There is a mathematical proof that all thinking machine has distinct stages, each one requires the capabilities of the one before it to exist. It's called The Model of Hierarchical Complexity We can see that norns have stages 0-3 hard coded into them in the form of scripts, to allow stage 4 and 5 to appear. My modifications raises it to level 7 (and even then, not fully, because stages 0-3 were cheated in). To do the kind of stuff you want it would need to get to level 9. So first we need a way for norns to count, unfortunately, I have no clue how a brain learns to count, so I can't even wire in an ALU to fake it. |
|||
Geat_Masta |
I spent all day trying to figure out CAs I don't know why fluid dynamics is so hard! Let me just think about this. Okay so the first things we need to do, to make any of that possible is change the brain to be more like a real neocortex. (undo the cheating at level 3). So: instead of a vision neuron referring to an object, it would have to refer to a location. Essentially a grid of neurons representing each area of the visual range of the creature, and it would select the most interesting object in that location, and each neuron would respond to some property of that object. Next I'm going to cheat, I added in a form of mirror neurons, so norns can hypothetically learn by watching each other, so I reuse that code, so that each section of the visual field knows what the others have learned. A carrot here is the same as a carrot there. Then that has to be wired into an object recognition lobe, and finally, the object recognizer into the concept lobe. What could be the list of things that the neurons are responding to? And object catagory has to be removed from the brain entirely, because including it defeats the purpose of undoing the cheating. |
|||
Prodigal Sock Ghosthande |
I'd like to supply useful advice but that is sadly way over my head. Bifrost wrote: Yes. I rather like that they're so chaste; not just because it keeps the game kid-friendly, but because people who don't want to have to see digital animals "mating" (myself included) don't have to. But kisspopping is definitely too distinctive. I rather like the idea that they nuzzle each other instead. Bifrost wrote: If Kreatures coding is comparable to C3/DS, I can think of ways to make it work... if it were just a sound they made at each other, not something with lyrics, it could probably be scripted just like any other interaction (eg. speech). BaffleBlend wrote: Croc 2 had characters speak like that as well. I always thought it was cute, although the Gobbos' voices were a bit high pitched. |
|||
Geat_Masta |
The Kritters are more than capable of learning sequences of notes from each other, and a surprising amount of birdsong can be generated from a small number of neurons. The problem is more: how does the kreature determine the "goodness" of a song? |
|||
Merboy |
Assuming we don't want the creatures to couple for life, what if the game automatically assigned a song to every creature, every day (24 minutes was it?). But the song was broken into two parts. And the game gave each part to one random male and one random female. And they were attracted to each other because they each had a piece of the same song. And omg this is too precious I can't even. I don't even care if this can't work I'm going to write a poem about this brb. The Lantern Light.com |
|||
Bifrost |
How do norns determine if they like each other? I think there could be much simpler ways to decide than randomly assigned pieces of a puzzle. |
|||
Prodigal Sock Ghosthande |
Birds learn their songs from family and neighbors when they're young, and are attracted to songs that sound "foreign"--which is a nice trick when you want to promote genetic diversity. There must be some way to make kritters learn songs from each other, since Norns can learn words that way--though I've no idea how that mechanism works. For checking compatibility, I suppose you could take the "song strings" for two kritters and check to see if any difference exists between them. If they're identical, the singer isn't attractive. This would actually add a nice level of challenge because it would give kritters a degree of reluctance toward inbreeding. If first gen kritters are assigned wholly random strings, then they would all find each other attractive, making the game easier at the beginning. And harder with each new generation, because you'd need to maintain enough variety to keep them interested in each other. Hmm. I wonder if there's a way to ensure that they only learned song from their own parents? |
|||
Geat_Masta |
The mechanism for norn word learning is this: they have a list of strings for each word they know and each has a byte from 0-255, when they hear that word for that object the byte gets incremented, when they hear a different word it gets decremented. If the value is less than 5 they replace some of the letters with other letters, "yes->yeth" and if it is 0, when they hear a different word it get's replaced with a different word. |
|||
Bifrost |
So, we could have children learn melodies in the same way, then? Would it be possible for them to stop learning new songs when they're adults, and rather than decrementing known melodies when they meet new ones, they get an increase in attraction and/or mating drives? |
|||
Geat_Masta |
If I added a receptor locus for "learning potential" then you could have the song lobe start with it maxed out then drop as the kritter ages, so it would only learn songs as a baby. Then there's already an emitter locus for 'anomaly,' 3D lobes try to predict their next input, so if the next input isn't the next note in the known song then it increases anomaly, so just link an emitter hooked to that to arousal potential, right? |
|||
Geat_Masta |
BTW, because the creatures are diploid, mutations are less likely to be disastrous, so I made the mutation rates much more aggressive than in the original games. So incest would likely produce non-viable offspring. |
|||
Merboy |
And the singing is used to stimulate arousal/sex drives/fertility AND in actually locating a mate? This seems an important distinction. The Lantern Light.com |
|||
Geat_Masta |
For locating a mate? I would have a gene switch on during youth that inverts the song anomaly and increases crowdedness. On the other hand it would make sense for the family song to increase comfort. I'm also not sure how to get from recognizing the sequence of notes to producing it. It sounds easy enough: assuming when the creature sings it always starts with the same note: it will recognize that note as the start of a song, the lobe will predict input, you link the predictive to note output, and it will do it's tune. The first note will match to multiple songs, so the song lobe links into the drive lobe to decide which one to sing, sounds easy. Then we can make a mirror song lobe, and link it into drive, so that if a creature hears a sad song he recognizes he starts to feel sad too. Right? The problem is the sexual songs are the opposites of all the other ones, it's not what he recognizes, but what he doesn't, so how does he know it's not a sad song he doesn't recognize, but a love song? |
|||
Geat_Masta |
I've got most of the CA code written, all that's left is some vorticity methods. (i.e. the flow should work, but currently nothing causes it to start moving). And I'm not sure how to implement the CAs themselves, basically there's 2 kinds of CAs: 1 grid-based (fast but low resolution) 2 tracer-based (slow but high resolution) The problem is that neither of these are particularly good for smells, because the CAs were written using a fluid simulation, they should have updrafts, and downdrafts, and you'll be able to see particles blowing in the wind, forming little swirling patterns, etc. Furthermore, because rooms are basically gone, replaced with collision geometry, 'linking' elevators isn't really possible anymore. And if CAs just follow horizontal motion it could lead a creature off a cliff. Also, I'm not sure how useful smells actually are, they may be best just left out. Can someone make a norn with the smell lobe removed and tell me if the ability to find things is significantly reduced? |
|||
Geat_Masta |
rudemantary particle physics Particles can be set to go through their phases based on time, or change phases on collision, thus rain can be represented as particles that change to another phase when they hit the ground. Instead of permeability being all or nothing like with objects, probably a permeability of 50 would mean 50% of particles can get through. |
|||
Prodigal Sock Ghosthande |
Geat_Masta wrote: Norns can definitely locate objects without using a scent as a guide, because there are various things (eg. elevators, buttons, toys) that lack scents entirely. Norns are just limited to noticing such objects when they are within visible distance, and must wander aimlessly if one isn't around rather than being able to follow its scent. I don't know exactly how much that would impact them if it also applied to things like food... I don't know that C1 and C2 Norns could follow smells, and they survived okay. |
|||
 Wingheart |
I've played with carnivorous Norns unable to follow smells-they can survive, usually, but tend to find it much harder than Norns that can smell their food. |
|||
Bifrost |
I wouldn't use song as a mating call, where kritters sing to locate potential mates, but rather as something they do to eachother as a means of communication. Children (and adults?) get calmed down by familiar melodies, and adults get aroused by exotic melodies. There should also be a mutation rate in learning melodies, so a population won't get stuck with one melody, and nobody finds any of the others' song attractive. |
|||
Geat_Masta |
I could use an undirected cyclic graph, each vertex would check it's vicinity to see what objects are there, and when a creature is trying to find an object it would do a breadth-first search along the graph to find the closest object of that category. However, it wouldn't be possible to procedurally generate this graph, the locations, links and search radii would all have to be entered by hand. |
|||
Geat_Masta |
On second thought that would take O(N^3) and i need in O(N log(N)). I can't think of anything that will work. What it comes down to is this, if i did: For each creature: set all smells to 0. For each object: get distance of object to creature add distance divided by length of distance squared to object's smell How messed up would that be? Would the creatures be able to smell a lot of things that they shouldn't and wander off cliffs, or would that work reasonably well. Other problem: if I make it so smells slowly move from place to place, then a creature could follow a scent to an object that no longer exists. 3rd problem: there's very little I can do about smells going through any wall that's thinner than the grid resolution. |
|||
 replies cannot be added because this thread has been locked. | ||||
| downloads cobs adoptions creaturelink metarooms breeds gallery art wallpaper screenshots graphics promos sprites |
dev hack shack script reservations dev resources active projects dev forum community links advice chat polls resources |
creatchi forum bookmarks general news help development strangeo survivor |
mycaves log in register lost pw |
0 online |
| creatures caves is your #1 resource for the creatures artificial life game series: creatures, creatures 2, creatures 3, docking station, and the upcoming creatures family. contact help privacy policy terms & conditions rules donate wiki |
||||