Welcome to Keen Software House Forums! Log in or Sign up to interact with the KSH community.
  1. You are currently browsing our forum as a guest. Create your own forum account to access all forum functionality.

the future of voxels in gaming

Discussion in 'General' started by toxi, Apr 8, 2016.

Thread Status:
This last post in this thread was made more than 31 days old.
  1. toxi Apprentice Engineer

    Messages:
    390
    taken out of an other topic ... the interessting question about voxels.

    voxels are nothing to new ... voxelman came out 1986 ... than there was quite a wave of games using voxels in the late 1990ies. than there was a 10 year silience and since minecraft we have the second big wave, a wave that SE is riding, along others.

    the thing about voxels is if you reduce the length of the voxelside, you need, to fill the same space not twice as many voxels, but 4 times as much, it grows quadratic.

    the good thing is that the performance of our PC grows also quadratic ... it doubles every 18 month.
    (its not exact, its not strict ... but it dont have to to see the point of it)

    wo know that building a world out of 1m blocks works quite well in minecraft ... but we are here in the SE forum so lets use this as a reference. if we suggest that a world made out of only large blocks work very well .. we can say that 2.5m, right now is our base. 18month later, an newer pc could handle a world build out of 1.25m blocks as well as our pc can handle 2.5m.

    we can keep on doing this and will end up in the range of 1mm after 11 iterations, or 17 years. for me, thats the point where the voxel is so tiny, i cant see that it is a voxel anymore.
    if our target is just 1cm, it just takes 8 years.

    i think thats quite an interessting point ... back in the late 90iest, some called that the voxel is the future of gaming ... and i did agree then and i do now .... but ... how tiny must the voxel become to be a real competitor to the classic polygon?
     
  2. Light_gemini Junior Engineer

    Messages:
    671
    If voxels were the only thing using the increase in cpu power.... but everything else is taking its slice of every cpu increase.

    Not counting limits in the other stuff involved like Ram, coding language etc.
     
    • Agree Agree x 2
  3. KissSh0t Master Engineer

    Messages:
    3,508
    Soft Voxels : D



    This kind of voxels are cool..

     
    • Informative Informative x 4
    • Like Like x 1
  4. The Q Apprentice Engineer

    Messages:
    247
    I dont think there is a better system, is there (if anyone comes and says minecraft, i swear to god)
     
    • Agree Agree x 1
    • Funny Funny x 1
  5. mze9412 Junior Engineer

    Messages:
    791
    I agree. Minecraft only works because it is very static. Blocks usually do not move at all (except via pistons or gravel falling down), so there is no big simulation. Also moving blocks are converted into entities and then converted back when they stop moving. That would be totally unusable for SE anyway ^^
     
    • Agree Agree x 1
  6. Spets Master Engineer

    Messages:
    3,209
    Planet Explorers also use voxels
     
  7. Teravisor Apprentice Engineer

    Messages:
    398
    It's already starting to hit heat and precision(I mean processor size in nanometers) boundaries. So performance increase will slow down a lot of times. Unless there's entirely new technology implemented (quantum computers for example) we won't see enough performance rise to see 1 cm voxels.
    What would really push voxels forward is hardware support for voxels. Imagine GPU drawing not triangles, but voxels...
    It doesn't really need to be tiny. In fact it doesn't need size at all. Google for "point clouds" (you could think of them as unstructured voxels which aren't aligned to grid), it's already competitor, albeit a weak one as noone really figured how to substitute triangles with generated point clouds and there's no hardware support.
     
    • Agree Agree x 1
    • Disagree Disagree x 1
  8. Taemien Apprentice Engineer

    Messages:
    468
    We had a surge of voxel based games recently. Some good, some not so much.

    Two noticeable examples by Daybreak is Landmark and Everquest Next. Landmark started as a community building event for EQN. They then tried to decouple it from its Norrathian roots and make the two games truly separate. The issue with Landmark is it was a MMO. MMO means basically one server that everyone can play on. How do you divy out the space? Well you have to make it fair so everyone gets a claim of space. The space is limited unlike non-MMO type games.

    Everquest Next never made it to an alpha build before being canceled. They wanted to procedural generate Norrath for the third title in the series. Not a bad idea, but they also wanted to make it voxel based. This way if you had a big fight with a dragon.. the ground could be decimated and even cause holes characters could fall into and into hidden caves and the like. Cool idea.. but I'll wager they spent too much time trying to figure out how to make it so the world isn't destroyed or maimed by destructive players. And trying to implement those ideas.

    From what I've seen with both SE and ME, as successful voxel games. You can make a simple theme around a voxel world and make it work. You cannot take an existing world (like Norrath) and make it voxels. And that seems to be the mistake most Voxel games had been making. As well as the obvious ones of hardware limitations of the consumers, and the programming limitations of the devs themselves.

    IMO voxel engines tend to make alot of interesting effects. Blowing a hole into the ground works great. Ripping a building apart where sides fall off, but girders snap at different pressures.. no so much. I mean its possible. But not easy to work with from a coder's point of view.

    It reminds me of a quote from a dev working with the physics in CryEngine in an overhaul game mod. "It allows us to do stupidly hard things easily. But the simple things are stupidly hard to make right." That's the effect I get from voxel based development.
     
    • Like Like x 2
  9. Me 10 Jin Apprentice Engineer

    Messages:
    463
    Cubic. 1 cubic meter = 8 half-meter cubes. Math is holy to engineers. Repent your transgressions, lest you be consigned unto "Zero" for eternity.

    No. That is a gross perversion of Moore's Law. Also, doubling periodically is not quadratic, it's exponential.

    The ideal size of a voxel is whatever optimally achieves the intended goal, and that can change in real time.
     
    • Agree Agree x 5
  10. waterlimon Senior Engineer

    Messages:
    1,499
    If we get better machines id just prefer freeform building (more in direction of KSP). Beyond a certain scale, you dont gain anything by being able to make bigger things with help of voxels (as you lose sense of scale, whats the difference between 10 km ship and 100 km ship?).
    And overall more detailed physics and simulation.

    Perhaps we can use 0.1 mm voxels for the individual components (which, btw, we should be able to do perfectly fine already, as they can be processed into regular meshes for fast rendering/collision)
     
  11. KissSh0t Master Engineer

    Messages:
    3,508
    This is a teaser of Star Citizens procedural tech, from what I understand it will have voxels... for mining.

    Not sure if this is voxels, but it sure is impressive for early work.....



    Big *__*
     
    • Agree Agree x 1
  12. toxi Apprentice Engineer

    Messages:
    390
    its was 1996 and i was siting in an lecture and we crossed moors law .. the speaker sayed thats all good and fine for the past .. but if we hit about 680MHz, this curve will stop, because its not possible to reach 700MHz ;)
     
    • Funny Funny x 2
    • Disagree Disagree x 1
  13. Teravisor Apprentice Engineer

    Messages:
    398
    Then tell me, why have we stopped at 4 Ghz as practical limit for ~5 years already?
    The answer is: in order to decrease heat we need to decrease size of tech process; however, we are close to the edge where at any time quantum effects can start to show up making calculations non-deterministic which would require complete technology change to make use of them. And in order to cool better we need better cooling systems (best we have is water cooling which is quite expensive to regular end-user).
    If you don't know where limit is, that doesn't mean there's no limit. Also I didn't mention Moore's law, he was talking about increasing transistor count, not frequency of processors.
     
    Last edited: Apr 11, 2016
    • Agree Agree x 2
  14. toxi Apprentice Engineer

    Messages:
    390
    moors law is quite fluid ... there are some who try to make it some kind of natural law ... it isnt ... the basic idea is that you get double the power out of 1000USD every 18 month. whatever that means.

    we end up in deathends ... and we compensate that. true, but we always did overcome that ... and it was not so often that it was clear how we will do that before it was done ... because there are always some attempts to overcome the issue but only a few, if not just one, gets a line in the book of techno-history. thats applied evolution.

    try to imagine moors law in a different way: lets call it "homers law". homers law should today say [that homers grill parties will increase the kcal count every year by a of factor 2]. as Me 10 Jin pointed absolutly correct out: thats not quadratic growth but exponential. however .... every year the amount of cals's are doubled. some times because he uses more fat, than because he added a truckload of corn ... or a lot of doughnuts.

    the thing is you never know what homer will do ... but you can be pretty sure that he will make ppl fatter than the year before ;)

    moors law is the same. i have no idea what they will do ... and frankly, i dont care anymore (i did .. with 20). i know that i will buy a new computer every 4 years and i know the new one will be stronger by 2-4 times, depending on the price of the past and the new computer. if that is a qanttumcomputer or a analog system or even some kind of biological stuff ... i dont care. ... because i have no time anymore to get into this.

    perhaps we hit the roof at some point, perhaps thats today. but something tells me ... the end isnt near.

    but hey, thats just a feeling and i cant bring a proof, but i would bet a great pizza that moors law will go on at least the next 10 years. :)
     
  15. Spets Master Engineer

    Messages:
    3,209
    I hope this game dont come out super perfect and with all the stuff they are promising. Because it will consume so many lives around the planet that zombie apocalypse will be real.
    in reality I hope so... because I wanted a game like this since I started playing open world games.
     
    • Like Like x 1
  16. StuffYouFear Apprentice Engineer

    Messages:
    416
    My god, this thread... so little facts, so many options...
     
    • Funny Funny x 2
  17. FlakMagnet Senior Engineer

    Messages:
    1,551
    Moores Law is fascinating ..... but it will require a significant change to current technology to hold true in the coming years.

    What has been happening is that we are cramming more 'cores' onto chips now, rather than making the things more complex. Manufacturing techniques continue to reduce the sizes, but other factors come into play when you get really tiny. Physics goes on holiday.

    Rather than a simple yes/no pulse no pulse setup, you start to invlove strange acts of poisoned felines and quantum possibilities that only collapse into real states when you measure them.....which will not help with logic or voxels.

    If you look at the scale of operation in the finest computer we all own ( the grey squishy one living between your ears ) then technology has a long way to go yet.

    Don't expect to see Moores Law holding any truer than Murphy's one, or the one about toast falling butter side down. There comes a point where people simply do not need the next level of tech, and the funds to develop it go away. Increases in production costs from ever smaller patterns etched in silicon ( many more thrown away due to manufacturing faults ) means the bubble has to burst at some point.

    When it comes to games development.....who knows. For me, i'd rather play a clever well designed game that ran on an abacus than a flashy super tech FPS game with shoddy levels and poor play. I mean...I'm still playing Freespace 2 for goodness sake!
     
    • Agree Agree x 2
  18. toxi Apprentice Engineer

    Messages:
    390
    the same was true in 1996.

    our top of technology right now is always the fundament of the next generation .. in 2040 they will say "jea well ... think about 2017 ... i mean: that was an easy jump forward, its nothing compared to the problems we face now!"

    its always hard to go beyond ... and most of the ppl will always say that there is nothing ... that the end is here and it is impossible to go any further.
    fact is there is no god, but there are H2 fuel cells in the world of engineering.
    (one of the statments i can prove easy.)
     
    • Funny Funny x 1
  19. FlakMagnet Senior Engineer

    Messages:
    1,551
    You have to look at what iscurretny driving PC development and CPUs in particular.

    I am running a mid-spec Core i5, 4 cores, 3.2Ghz.

    It oes just about EVERYTHING I want a computer to do, and it does it fairly fast. Compared to the dinosaur machine I have to use at work, it's miles ahead. The ONLY thing I do with it that pushes it anywhere near it's limits is gaming. I can run massive databases, spreadsheets and so forth which it doesn't notice. Games are what pushes PC design in teh current market.

    So....you limit the market to early adopter gamers with plenty of disposable income.

    People like me, who look at the midrange options ( 4690 for me please.... ) as the increase in performance for the next step up comes with an exonential price increase. I bought a Gfxcard that maxed performance on a monitor size I was likely to be using at it's native resolution, then I matched components to those key items so nothing else bottlenecked it. I was enver EVER going to buy the cuting edge gear with it's heavy price penalty. I just can't afford it, can't justify it, and probably wouldn't even notice the benefit for 99% of my PC usage.

    Get too many people doing this, and the push to develop flagship products diminishes. Largely because todays 'midrange' is so good, there is little incentive to repalce it. I put my rig together a few years back, and apart from increasing RAM and swapping to a SSD for windows, haven't needed to upgrade the basics.

    The further they push, the fewer people need the top of the range. That puts the brakes on with Moores Law. You end up selling to the small number of PC 'geeks' who have the money and only want the best.
     
    • Agree Agree x 2
  20. Hakon102 Apprentice Engineer

    Messages:
    300
    I agree Teravisor. If i you know the Physics behind, then you know the fact, that they already push around Atoms and the Space between the transistors is only "some" Atoms thick. Sure there are maybe some ways of "better position the transistors in the room/space". But the possibilites are limited by the (Tunnel effects(tunneling) in semiconductors. I expect only 1 or 2 big increases in "transistors per square centrimetre" in the next years, but not more.
    After that, we get 3D Processors "transistors per cubic centrimetre" , but here is the Problem the heat conduction and the space is also limited at high frequencies.

    And you can't get higher frequencies of processors, because of electrical resistance, reflections, Skin-effect, parasitic effects etc. . Sure, there are some possibilities. I don't think we get a 10 GHz PC for the private market in the next 10 Years.

    This all don't mean that the performance of Computer's can't increase anymore in the distand Future, but this mean we need an other concept of Computers and new inventions in quantum engineering/ materials technology for semiconductor etc. . And you can't say, this invention will happen. Moore's Law is not a scientific law of nature , it's only an empirical observation.
     
    • Agree Agree x 2
  21. toxi Apprentice Engineer

    Messages:
    390
    i agree but my argument stands.

    back in the days we faced some problems .. i cant remeber what it was, but my lecture was pretty sure that nobody will overcome them. i had nothing to argue back than ... as i dont have anything now .. my point is that he was wrong ... because somewhere on this planet someone had an idea and this idea worked out, it become a product, than common knowledge and is now a part of our history.

    the central question is: will progress ever stop? and i dont think so. i cant prove that and i never will be able to.

    i agree that we right now manage the progress just with sharing the workload with more and more and more units and that this cant go on ... but ... the transistor is not the end of the road (remember the tubes? ;)) ... to be true .. i would bet my money to biological hybrides for the next 15-30 years.
    if this works .. i think will give us a boost that compensates for the lag of growth in the years behind us ...

    the thing is .. we cant extrapolate the past to know what the future will bring ... but the past gives us an idea what direction it goes ... and if we follow the technology over the last 3000 years, in 20 year steps ... it is a exponential growth .... in 100 years we did learn to fly, we let the soundbarrier behind us and we did our first steps into space ...

    if it stops, it is because of such nice ppl like al-ghazali come up, say that all mathematics is devilstuff and his marry fellows demolish anything that offends god or whatever.
    thats the reason the arabs lost the upper hand in technology in the middle ages ... and the same thing (with a different name) is the reason the arabs had 1974 a bigger industry than today. and this could of course happen to us too (with a look to europe .. it could be sooner here.) ... but ... in some place on earth the progress will go on ... and that will be the place to go .. and there ... some cool sh*t will happen ... and progress will go on ...
     
  22. Hakon102 Apprentice Engineer

    Messages:
    300
    Sure, I Agree.

    As long as the demand for better hardware is high enough, the progress will not stop. It's more an economic question, research is very expensive.

    I just don't rely on Moores Law. --> Some informations
     
  23. toxi Apprentice Engineer

    Messages:
    390
    to say that research is expensive is as to say that food is expensive. the alternative is worth.

    research is like eating ... if you dont do it, you will become weak and die ... of course: most of the stuff that you invested ends up in shit ... but thats no reason to stop eating.

    the thing with moors law is that it is an idea and curves are depending on the amounts of points. if you drive from berlin to some place, and you give me 10 000 GPS data for the first 100m ... i have no chance to know where you want to drive ... give me 1 after 50km and i have a way better idea if you want to drive to vienna, paris or moscow.

    moors law is no law .. its an idea .. like "west" and "north" is an idea. the idea is that we are not at the end of the road ... and i celibrate this idea.
     
  24. Burillo Junior Engineer

    Messages:
    648
    these are two different situations. if a particular professor couldn't imagine how transistors can be smaller than they were, and how clock speed could be higher, that's one thing. but you can't really make stuff smaller than a few atoms wide, or smaller than an atom. you just can't improve indefinitely - sooner or later you will hit basic physics limitations.

    consider this analogy: you can accelerate up to certain speeds a few decades ago, now you can go faster. and in the future, you'll be able to go even faster, and you'll build better, more efficient engines to make that happen, but at some point you'll hit the relativistic speeds, and you ain't going to overcome those. this just isn't going to happen. we've already hit the "relativistic speeds" with our clock frequencies. the fastest Pentium 4 was clocked at almost 4GHz, and it was released in 2004 - that's 12 years ago. and at the time, Intel claimed that the NetBurst arch would allow for clock speeds up to 10GHz, so the future looked bright... except it didn't happen. clock speeds haven't improved much since then - we're hovering at about the same 4GHz because that's the practical limit we can't really go over unless we're willing to cool it with liquid nitrogen or something.

    we kept the Moore's law going through other means - namely, OOO execution, hyperthreading, more instructions per clock cycle, power efficiency, more cores, stuff like that. the Homer analogy described above is really great for explaining this phenomena, but you have to again understand that you can't improve on that forever either. sooner or later you'll hit a practical amount of cores to have on a single chip, before you hit the various limitations like power dissipation, cache coherency... and that's not mentioning the fact that some tasks just aren't well parallelized, so you can't throw more and more cores and expect it to always work. you can do lots of tasks on lots of cores, but having lots of cores doesn't necessarily speed up execution of a single task.

    personally, i think we are only in the beginning of the era of parallelism, and games have only recently started taking advantage of multi-core stuff, so there's still much room for improvement. but at some point, you're again hitting a limit on what you can do given the constraints you have. functional quantum computers are years away and you most likely won't get a commodity quantum computer (like a PC) within 20-30 years, if at all, and again, not all tasks are well-suited for quantum computations just like not all tasks are easily parallelized.

    bottom line, i think voxels aren't the future. they're a good fit for a game like SE, but for something like, say, Crysis, a much better alternative would be some kind of physics acceleration (PhysX?) so that you can have partially destructible worlds without the burden brought about by voxels.
     
    • Agree Agree x 2
  25. toxi Apprentice Engineer

    Messages:
    390
    thats right ... but the same was true for tubes in 1960ies .. or mechanical calculators in the 1930ies.

    that means: i agree ... the transistortechnology is perhaps at its climax .. not in its end, but in its peek ... but does not mean that progress in computingpower will end there. thats the core idea.

    its like saying: "look, you can breed faster and faster horses, but there is a limit"
    thats right. i agree ... absolutly ... but than the modern car came up.

    the thing is: as long as you can build on an idea, as long as there is potential to get more out of the same concept, you will follow this concept. right now thats the semiconductor. but thats at least the 3. version of computers.

    what i say is a bit esoteric ... i give you that ... because i belife in progress, not matter what .. and i cant ever profe that this is correct, so i can understand everybody who says "nah. thats it. end of the pole." and perhaps i am wrong ... but just for the sake of the argument ... what about if i am right and in 20 years we have such powerfull computers? whats than? how do we waste this power? because making even better graphichs will now stop .. because we will hit the limits of our eyes :D
     
  26. Lt_Duckweed Apprentice Engineer

    Messages:
    417
    Until we actually have a viable transistor substrate replacement for silicon, mores law is dead. It is not the end of perf improvements, but unless we move to graphene transistors or silicon-germanium alloy transistors, the perf improvements are for damn sure not coming from single core performance. The next big jump will be storage/dram unification and non-volatility. Better multithreading, and higher core counts (but this sees strong diminishing returns). And finally unification of unified storage/ram with cpu cache.

    In overclocked single core performance and in total performance, a i5-2500k is only 20% behind a i5-6600k. The 2500k is five years old and only slightly falls behind a modern CPU. Here's to hoping AMD ZEN lights a fire under Shintel's ass.
     
  27. Teravisor Apprentice Engineer

    Messages:
    398
    Modern SSDs are already packed to a degree where they can't maintain data more than 5-10 years without power/overwrites because it loses charge. If we continue to decrease memory cell size further, we get usual RAM we use currently. Or we can just expand amount of memory cells, increasing cost proportionally. So why not use RAM in the first place?
    CPU cache size suffers from cost/efficiency ratio for last 20 years, that's why it's of size it is currently. Expanding it would require cost to rise extremely while giving non-proportionate benefit.
     
    • Agree Agree x 1
  28. Ronin1973 Master Engineer

    Messages:
    4,797
    Sure, we can build faster computers... but as far as gaming goes... we're still stuck using just one core out of arrays of 2,4,6,8. We want faster cars that are only using one cylinder of the engine.

    (and yes, getting a game to run on multiple cores is no easy task)

    But there's so much room to grow if there is some sort of change in the industry or CPU development that makes it easy to use multiple cores.

    The next front is GPU based physics engines and functions (other than graphics) pushed off to the GPU or cluster of GPUs.

    https://en.wikipedia.org/wiki/General-purpose_computing_on_graphics_processing_units

    The next leap in voxels will come from advances in GPU technology and processing. nVidia and AMD are set to release the next gen of graphics cards in a few weeks. They are going to be leaps and bounds faster than the current generation.
     
    • Agree Agree x 1
  29. Hakon102 Apprentice Engineer

    Messages:
    300
    Hm... i think it's a bad comparison. You are right
    in your core statement. "research will happen" . But, you can't say which kind of research. Think about electric cars, renewable energies. The researches started, as the demand was present. (environmental Protection, high fuel costs etc.). If the demand for better Hardware is to small (because "only a small group of Gamers ask for it" e.g. why need a office desktop 16 cores?), the researches of Computers will be slowing down and the money for research are invested in more important things /more profitable things.

    But under the line, i don't know if voxels are the Future of Gaming. Maybe... maybe not.
     
  30. mze9412 Junior Engineer

    Messages:
    791
    I see voxels the same way I see programming languages: Use the right tool for the given task. If voxels fit the concept, use voxels. If they do not really fit the task, do not use them. They are definitly not the final and best solution for everything. For SE they most probably are the best possible solution ^^
     
    • Agree Agree x 1
Thread Status:
This last post in this thread was made more than 31 days old.