|
Great Joe posted:This is a scam and everyone taking it seriously after all these years needs to get their brain checked. Why did it take so long for this to come up?
|
#
?
Nov 15, 2014 11:48
|
|
|
|
| # ? May 4, 2024 23:59 |
|
|
Will this new engine be able to handle Hooker-Bots?
|
|
#
?
Nov 15, 2014 12:00
|
|
|
I can't believe these clowns are still around, i guess they're still finding marks who actually give them money after all these years. Also the Title/OP reads like a bad marketing PR note copy&paste. Did you just land a job there or what? RoadCrewWorker fucked around with this message at 13:58 on Nov 15, 2014 |
|
#
?
Nov 15, 2014 13:54
|
|
|
With the current advances in scanning technology, this seems like a great idea for the future of gaming.
|
|
#
?
Nov 15, 2014 14:23
|
|
|
Mr Scumbag posted:Why did it take so long for this to come up?
|
|
#
?
Nov 15, 2014 14:52
|
|
|
In LA Noir's case I wish they'd stuck to traditional methods because that poo poo is just terrible.
|
|
#
?
Nov 15, 2014 14:54
|
|
|
I liked it as a first step kind of thing, but now it looks like L.A. Noire will never get a sequel and Team Bondi might well be dead.
|
|
#
?
Nov 15, 2014 15:12
|
|
|
Shouldn't even bother in this lovely thread, however; the ability to scan with hyper realism (which this isn't anyway, but it's pretty decent) will never be a substitute for being able to create high quality assets from nothing. So them showing off their this-took-a-month-to-scan-and-clean-up church interior is pretty worthless, at least for gaming applications. Is the OP and their company seriously suggesting that, say I want to make a shield for my video game, I'd have to hire a blacksmith, armorer, or woodcarver to create an actual shield in real life, just so I can scan it? Or what if I want some sort of huge spaceship hanger bay, better get that construction crew started now! Basically this is all pipe dream bullshit. And films use polygons not because the artists can use workarounds to make it easier on the render farm, but because they're the best at doing their job. The only big change to the age old system I've seen is something called Ptex, which is a system developed by Disney to eliminate the need for UV mapping. That is a genuine improvement the majority of big studios use for their films (if not all studios). fake edit: and yes voxels are long dead too. real edit: don't know why this is in Games either, should at least be in CC. edit 2: the reason scanning isn't very popular in games is because it requires a lot of tech know-how and specialty equipment to pull off. Like that face scanning rig for LA Noir, I doubt that was cheap or easy for any of Bondi's artists to work with. For Forza, considering that is Team 10 Studios only IP (and the only thing they seem to work on) it would make sense for them to invest in such technology. Plus scanning a track is less about looking good (it's mainly asphalt after all) and more about getting accurate track data. But for average companies, scanning isn't worth the time or resources. edit 3: and also I should point out for that church interior, or any scanned scene of any kind, the lighting information you see is baked on the textures themselves. This means that, should anyone want to change the lighting, it would be nearly impossible. Those highlights are there forever, the shadows there forever, etc. etc. This is why for scanning you have to get the most diffused lighting possible with no shadows (like the face scanning rig again in LA Noir). KiddieGrinder fucked around with this message at 15:34 on Nov 15, 2014 |
|
#
?
Nov 15, 2014 15:23
|
|
|
These guys appear to have also captured ghosts
|
|
#
?
Nov 15, 2014 15:24
|
|
|
I have my doubts about that engine, but the process of capturing assets from the real world instead of building them from scratch is already happening and is only going to get easier. As mentioned above, those technologies are not very useful (yet) when you need to make a game, but it's coming. We'll continue to use polygons for awhile because change of this magnitude takes time, but, yes, I think the [gaming] industry has already realized that it cannot just continue to throw more artists at the problem. Un-l337-Pork fucked around with this message at 15:39 on Nov 15, 2014 |
|
#
?
Nov 15, 2014 15:35
|
|
|
Un-l337-Pork posted:I have my doubts about that engine, but the process of capturing assets from the real world instead of building them from scratch is already happening and is only going to get easier. As mentioned above, those technologies are not very useful (yet) when you need to make a game, but it's coming.  point clouds (what a terrible name for voxels). Compounding the "problem" is the fact that improvements in looks matter less to both customers and console makers as time goes on, just adding detail to game scenes is in itself a game of diminishing returns, meaning it'll take much longer before consumer hardware gets even close to that point.
Great Joe fucked around with this message at 16:02 on Nov 15, 2014 |
|
#
?
Nov 15, 2014 16:00
|
|
|
You can't animate them, and you can't light them. But 'point clouds' are the future, because, polygons are an evolutionary dead end, because... The technology isn't going to magically improve and make these fundamental problems go away. Voxels are bad, except in Comanche, which was good.
|
|
#
?
Nov 15, 2014 16:10
|
|
|
DebonaireD posted:Voxels are bad, except in Comanche, which was good. They were also ok in the Delta Force games. Voxel grass was pretty cool back in the day. You can probably use them to calculate true LOS, which might be nice for games with wonky AI like ARMA.
|
|
#
?
Nov 15, 2014 17:01
|
|
|
Shima Honnou posted:I will enjoy this engine when it is added to Google Earth and I can feel like I'm really walking on streets, OP. this, this right here is legit.
|
|
#
?
Nov 15, 2014 17:50
|
|
|
KiddieGrinder posted:Is the OP and their company seriously suggesting that, say I want to make a shield for my video game, I'd have to hire a blacksmith, armorer, or woodcarver to create an actual shield in real life, just so I can scan it? Or what if I want some sort of huge spaceship hanger bay, better get that construction crew started now! No you sperg, someone would either find or make a replica. Do you think they do this in the movie industry every time they need a character with a shield? The Euclidean and scanning thing was just to demonstrate that there are alternatives to polygons and manual asset creation. The tech isn't currently practical for games, but it's becoming more so at the same time polygons are becoming less so. I'm speculating that sooner or later those paths will cross, and we'll start using something else. Killer-of-Lawyers posted:edit: And as to culling being a work around, welp, can't say much to that, except the most powerful visual computer to date culls and makes up so much stuff that almost everything it see's is an illusion. It's efficient. There's no drat reason to display things that you can't see, or have high resolution textures where they'd never get displayed on the monitor. Euclidean has a single culling algorithm that doesn't need to change or to be expanded with the level of detail. It runs off CPU, not GPU, so the only bottle neck is RAM and storage. Even if it's fake, it just suggests there might be a far more elegant solution than multiple culling techniques.
|
|
#
?
Nov 15, 2014 20:33
|
|
|
EvilGenius posted:there are alternatives to polygons and manual asset creation EvilGenius posted:Euclidean has a single culling algorithm that doesn't need to change or to be expanded with the level of detail. It runs off CPU, not GPU, so the only bottle neck is RAM and storage. Even if it's fake, it just suggests there might be a far more elegant solution than multiple culling techniques.
|
|
#
?
Nov 15, 2014 20:55
|
|
|
Mid term it looks like GPU ray/path tracing (over polygons) is going to be next thing, probably mixed with plain rasterization for awhile. Look at Brigade for a demo of 100% path tracing in real time; https://www.youtube.com/watch?v=BpT6MkCeP7Y
|
|
#
?
Nov 15, 2014 21:04
|
|
|
Great Joe posted:This is a scam and everyone taking it seriously after all these years needs to get their brain checked.
|
|
#
?
Nov 15, 2014 21:16
|
|
|
when will they be able to just scan good gameplay in?
|
|
#
?
Nov 15, 2014 21:19
|
|
|
EvilGenius posted:Do you think they do this in the movie industry every time they need a character with a shield? Yes?  Do you think VFX houses pop over to Turbosquid or something to get props and poo poo for the new blockbuster superhero movie? And as Great Joe and others pointed out; voxels aren't new. They've been around for a while, and the industry still uses traditional polygons. Scanning assets is not even suitable for every game/movie either, nothing to do with the technology behind it. You simply can't scan things that don't exist, and your above solution to this quandary is to "find or make a replica" which is not cost effective or even rational. It's great that the Ethan Carter team happened to be making a game that specifically called for those exact buildings and exact environments that happened to be in their area and that they could scan (what a coincidence!  ), but not every other game or movie would have that luxury.To use my example of a shield, a good artist could sculpt and texture an awesome shield using Zbrush or any other package in about an hour. I don't think anyone could argue that you would get similar results with craftsman creating an actual object by hand, and then scanning it in digitally. Maybe you should stick to programming? I don't know if you haven't seen what good artists can do, I don't know, but this "POLYGONS ARE DEAD!"  sky is falling rhetoric is really weird and you should feel bad.
KiddieGrinder fucked around with this message at 21:26 on Nov 15, 2014 |
|
#
?
Nov 15, 2014 21:22
|
|
|
amazing new gfx https://www.youtube.com/watch?v=WQ_Aqmkf3IU
|
|
#
?
Nov 15, 2014 21:27
|
|
|
Ray tracing isn't going to be a thing for quite a while still. It's expensive and looks bad in motion, though at least it's not as far off as using scanned voxel models for everything. What's actually going to be a big deal in graphics for at least the next 5 years is physically-based rendering and a few of the cheaper methods for global illumination. Stuff like this and this: https://www.youtube.com/watch?v=fAsg_xNzhcQ Unreal Engine 4 has these features already, so if you're thinking about what's going to improve graphics-wise in the future, it's probably moving onto more detailed models of global illumination, maybe another shader that adds detail to surfaces similar to tesselation and parallax mapping.
|
|
#
?
Nov 15, 2014 21:31
|
|
|
 And since those are all technically voxels, we can see which is truly superior.  edit: Great Joe posted:Stuff like this and this: See these I like. 
KiddieGrinder fucked around with this message at 21:34 on Nov 15, 2014 |
|
#
?
Nov 15, 2014 21:32
|
|
|
Polygons aren't going anywhere. They can produce photo-realistic graphics efficiently already. As a counter to other tech; if there exists another form to produce photo-realism, why do offline renderers use polygons? Expect raytracing hardware to become a thing in the next few years/decades. This will still rely on polygons, though. edit: Also, gently caress the Euclideon guys. They have extorted millions from the Australian Government for grants that were meant to be used to build up the games industry. How this scam continues, I don't know. Sillybones fucked around with this message at 21:39 on Nov 15, 2014 |
|
#
?
Nov 15, 2014 21:34
|
|
|
NecroMonster posted:when will they be able to just scan good gameplay in? Games are art.
|
|
#
?
Nov 15, 2014 21:36
|
|
|
As was already said, the problem is animation and lighting. Notice how dead this otherwise realistic looking forest is? Completely still and frozen. How would you animate it? How would you identify the subset of points under millions which make up this tree you want to be swaying in the wind? How are they connected, how do they move? Same for shadows and lighting. Can you have every point cast a shadow under changing lighting conditions? Now, the same problem exists for polygons, but for them they are largely solved. To animate a textured, bump mapped, pixel shadered etc. polygon you have to move 3 points, with the point cloud method you'd need to move hundreds or thousands. So, no, I don't think "The Death of Polygons" is imminent. But the point clouds make pretty "3D photographs".
|
|
#
?
Nov 15, 2014 22:18
|
|
|
Euclidean is horseshit and always has been. They've been hawking their bullshit for like five years now. Don't put much stock in them. That being said, is there any news about Ptex for realtime applications? It was mentioned earlier for film/vfx work and I keep wondering if it'll be used in real time work at some point.
|
|
#
?
Nov 15, 2014 22:25
|
|
|
Ptex is a good idea for non-realtime rendering but the approach is different for realtime rendering. For realtime rendering more prep work is worth doing, since you do it once and then it saves you time on execution for forever. When you're spending relatively enormous amounts of time processing each frame in non-realtime rendering, a tiny percent more to save your content creators a bit of time is well worth it.
|
|
#
?
Nov 15, 2014 22:28
|
|
|
Tintifax posted:As was already said, the problem is animation and lighting. Notice how dead this otherwise realistic looking forest is? Completely still and frozen. How would you animate it? How would you identify the subset of points under millions which make up this tree you want to be swaying in the wind? How are they connected, how do they move? Same for shadows and lighting. Can you have every point cast a shadow under changing lighting conditions? This is spot on. Watching that video was like scrolling through a panorama picture. Everything looked completely static and in no way like a living, breathing world.
|
|
#
?
Nov 15, 2014 22:38
|
|
|
K8.0 posted:Ptex is a good idea for non-realtime rendering but the approach is different for realtime rendering. For realtime rendering more prep work is worth doing, since you do it once and then it saves you time on execution for forever. When you're spending relatively enormous amounts of time processing each frame in non-realtime rendering, a tiny percent more to save your content creators a bit of time is well worth it. Ah, I didn't realize it was so computationally intensive. I had heard that it allows varying levels of detail on a per-triangle basis which seemed like a big win for videogames but if it is that expensive then maybe not. Interesting.
|
|
#
?
Nov 15, 2014 22:46
|
|
|
You've always been able to do that. UV mapping is how you decide to map a texture to a polygon (or more typically, a set of polygons that comprise an object). You can set the coordinates any way you want. Ptex is basically a system for making this process easier, but it isn't directly applicable to games because it's designed for the needs of movie rendering and not realtime rendering. That doesn't mean that techniques for content creation don't carry over between movies and games, it's just that the final format is going to be different because games and movies have different needs.
|
|
#
?
Nov 15, 2014 22:54
|
|
|
Great Joe posted:Ray tracing isn't going to be a thing for quite a while still. It's expensive and looks bad in motion, though at least it's not as far off as using scanned voxel models for everything. What's actually going to be a big deal in graphics for at least the next 5 years is physically-based rendering and a few of the cheaper methods for global illumination. UE4 cut this global illumination system because it was too slow. Q-Games is using a variant technique that builds cascades of voxel lighting data instead of an octree for their new game The Tomorrow Children, it's very cool but I wonder how much their scenes make it usable. It's possible it's not a viable general-purpose realtime technique. http://fumufumu.q-games.com/archives/000934.php
|
|
#
?
Nov 15, 2014 23:22
|
|
|
"It's a scam!" declared man who became a multimillionaire off of voxels three years ago.
|
|
#
?
Nov 16, 2014 01:47
|
|
|
EvilGenius posted:Euclidean has a single culling algorithm that doesn't need to change or to be expanded with the level of detail. It runs off CPU, not GPU, so the only bottle neck is RAM and storage. Even if it's fake, it just suggests there might be a far more elegant solution than multiple culling techniques. Do you not understand why we use GPU's? Of course the system uses the CPU, because there isn't specific hardware to do it yet. Everything can use a CPU. What happened is that as we started crunching more and more polygons, we started to make specialized hardware to do it. If this caught on then people will want to do the same. Anything that you do a lot of you're going to want to try and offload to a special portion of the CPU, or a separate device all together, until the point that doing said calculations becomes trivial. You want games to run off of floating points or whatever? Then you're going to want a special bit to do it. It's efficient. Also, how does something being fake suggest anything? Even if the Lochness monster is fake, it suggest that leopluradons may still be living today. That's the same logic you just used. If it's fake then it suggests nothing. Besides, why would culling be that different? You don't want to calculate the floating points far afield to the same detail as the ones close to field. That'd be idiocy. The points would be smaller than the pixels on the display. You'd also want to cull rendering any point obscured by another, because that's also a waste of processing power. If anything you'll need to do more culling because you'll have so much information present that you can't possible process it all, if you're going to have 'infinite' detail. Also, the limit on traditional rendering is... processing power and storage. That's the limit of every single thing in computing. How many operations can I perform and how much space do I have to store data. That's all you really care about for any problem (With the exception of a few other more minor limits, like how big is my pipe between my storage medium and my operation performer.).
|
|
#
?
Nov 16, 2014 01:48
|
|
|
Also:EvilGenius posted:As a programmer this never sat right with me. Theses are 'work-arounds'. These are writing extra bits of code around the limitations of your system, where the preference is always to come up with an elegant solution in the first place.
|
|
#
?
Nov 16, 2014 01:59
|
|
|
All other issues aside, can you explain exactly why you seem to consider photo-realistic graphics as the end goal/state for video games? I mean the invention of photography never rendered painting obsolete, why should photographic scanning be any different for 3D modeling?
|
|
#
?
Nov 16, 2014 05:10
|
|
|
I for one am glad polygons are dead. Hopefully we can take on video games as a whole next
|
|
#
?
Nov 16, 2014 05:16
|
|
|
720p? Jesus christ the XBONE is terrible this poo poo looks like complete garbage
|
|
#
?
Nov 16, 2014 05:29
|
|
|
RoadCrewWorker posted:Also:  When you start to write a large complex program, framework, library or whatever, the goal is always 'elegant solution'. The reality is years and years of cobbling together insane fixes and workarounds so you don't have to re-write the whole loving thing. Sophy Wackles fucked around with this message at 05:37 on Nov 16, 2014 |
|
#
?
Nov 16, 2014 05:34
|
|
|
|
| # ? May 4, 2024 23:59 |
|
|
Great Joe posted:This is a scam and everyone taking it seriously after all these years needs to get their brain checked. This. Point Cloud systems like this are bullshit for games because anything approaching the level of detail here DOES NOT SUPPORT TRANSFORMATIONS IN REALTIME. In layman's terms, that means THINGS THAT MOVE. That's why you've never seen a character in one of these loving demos. EvilGenius posted:poo poo, maybe it is a scam, but it's not the point. The point is, polygons are never going to be practical for true photo realism. EvilGenius posted:Vvv once they crack boob physics, they'll crack the industry. They can't crank boob physics, they can't even crack basic rotations, much less any kind of animation or physics Lord Windy posted:How long does it take to make a good polygon asset? Does it take the same time today to do a good polygon asset as it did 5 years ago? This is like asking how long is a piece of string, but the reality is that new methods increase the speed and new quality bars slow it back down. 3D scanning is the wave of the future and is already being used to create base elements that are then turned into final assets. Cutting edge games have been using scans for years to help speed up the process - in the same way that using photos as source material for texturing improved speed and quality, scans do the same thing for 3d. The Euclidean art is only scannable because the authoring methods you would otherwise use would consist of MODELING IN POLYGONS and converting to a point cloud. Which you would do if you needed a loving point cloud, but you don't usually use those for anything outside of fluid sim in film. Ostentatious posted:Will this new engine be able to handle Hooker-Bots? This new engine cannot handle hookers, bots, or hooker-bots, because it cannot make things move at all. EvilGenius posted:Euclidean has a single culling algorithm that doesn't need to change or to be expanded with the level of detail. It runs off CPU, not GPU, so the only bottle neck is RAM and storage. Even if it's fake, it just suggests there might be a far more elegant solution than multiple culling techniques. You dumb fucker, culling is not inherently tied to the format the data is in, and the same poo poo they're using you can use with polygons RoadCrewWorker posted:Also: He is not a graphics programmer in the slightest but I'm sure he made helloworld.cpp once and assumes it gives him absolute authority in the matter because he's a goddamn idiot.
|
|
#
?
Nov 16, 2014 05:49
|
|