As an Industry 3D-Artist, the amount of false advertising in this video is downright insulting to me. These are very standard point-cloud 3D-Scans of real-world places. These things have been around and used for some time now and they have very limited practical application, especially in the video game industry. What is the great innovation there? Interpolating the gaps? Whoopdie-fucking-doo, that is a totally minor programming effort and doesn't add any detail, it just plugs the holes, just like texture filtering on a textured polygon surface. AND THERE ARE STILL HOLES AND GAPS VISIBLE RIGHT THERE IN THE VIDEO.
And what was that about the 'subtleties of lighting?' So far the second biggest limitation of unlimited detail voxel technology was advanced lighting techniques, and it doesn't even look like they caught up on today's rendering standards. Because guess what? A 3d-scanned scene will capture the lighting from the environment as it was at the point of scanning. So you'd have to somehow edit that out and replace it with real-time lighting. Also, how would you ever get varying material properties into this workflow? So that rock/water/plastic/wood use the right shaders and react to lighting differently? It's all one big chunk of geometry that can't even use standart advanced surface parameters like Normal/Specular/SSS maps. How on EARTH can they make such ridiculous claims? Clearly all they want to do is lull technical illiterate people with their phony bullshit.
I DID like their approach to the per-pixel search algorithm optimized voxel technology, it had some genuine use if developed in the right direction but now they just use ancient tech and make it out to be a fucking cure for cancer
[QUOTE=Buck.;46021687]It was quite easy to tell it was fake. Maybe my eyes are just too trained for that. I hope that they are on to something but will remain sceptical until they show animation and collision detection.[/QUOTE]
Collision planes can be generated with a simple RANSAC or equivalent.
I feel like the lighting in a scene adds more to the realism than the amount of polygon detail, but that's just me.
Also I'm not really into photo-realistic games
If you have the choice to make something artistically stylized I say do it. I play games to escape real life not play a copy of it
[QUOTE=H4ngman;46022349]And what was that about the 'subtleties of lighting?' So far the second biggest limitation of unlimited detail voxel technology was advanced lighting techniques, and it doesn't even look like they caught up on today's rendering standards. Because guess what? A 3d-scanned scene will capture the lighting from the environment as it was at the point of scanning. So you'd have to somehow edit that out and replace it with real-time lighting. Also, how would you ever get varying material properties into this workflow? So that rock/water/plastic/wood use the right shaders and react to lighting differently? It's all one big chunk of geometry that can't even use standart advanced surface parameters like Normal/Specular/SSS maps. How on EARTH can they make such ridiculous claims? Clearly all they want to do is lull technical illiterate people with their phony bullshit.
I DID like their approach to the per-pixel search algorithm optimized voxel technology, it had some genuine use if developed in the right direction but now they just use ancient tech and make it out to be a fucking cure for cancer[/QUOTE]
I imagine techniques similar to deferred rendering could be used for lighting. Varying material is a known problem there but is usually solved by just adding more buffers.
The second I seen the "real world video clips" I seen right through it and knew what was coming up when he started blabbering on about YouTube's time limit. Cringe worthy.
A new Euclidean video? wow
And Black Mesa Zen still isn't out yet
converting laser scan data to voxels isn't revolutionary or practical. Geometric modeling and properly adjusted lighting within a controlled environment that can then be dynamically changed in-engine or baked for a lighter processing load is far, FAR superior in every regard.
[QUOTE=Snickerdoodle;46022343][IMG]http://i.imgur.com/2NMdPKh.png[/IMG][/QUOTE]
I laughed so hard I went, what? Quake graphics mmm
I doubt we will reach perfect graphix any time soon.
"Let's pitch to a bunch of artists that they are useless and replace them with 3D scans"
is this guy both a dick and doing his own PR cause that doesn't go well for anyone
Oh look, this gimmick again.
the Silent Hills PT and the FOX Engine altogether were far more impressive than this thing.
[IMG]http://i.kinja-img.com/gawker-media/image/upload/s--GsJ2-iBQ--/gbmlfvcthmznnbbyjl8w.jpg[/IMG]
Granted, this is a low-resolution screenshot, but it still looks FAR better than Euclideon in full-res.
PT and FOX actually had me fooled for a while. The fact that that was CG was obvious from the first second.
This technology is bullshit, everyone with a working brain can tell it and it's nothing extraordinary, it's been used for ages, all the stuff they did was scan bunch of models from multi angled photos, if there was a better way to achieve graphics then entire industry would have already jumped on that, there is a reason why there isn't anything better than polygons, because there isn't, polygons offer both best functionality and control period.
[QUOTE=Sgt. Lulz;46023786]the Silent Hills PT and the FOX Engine altogether were far more impressive than this thing.
[IMG]http://i.kinja-img.com/gawker-media/image/upload/s--GsJ2-iBQ--/gbmlfvcthmznnbbyjl8w.jpg[/IMG]
Granted, this is a low-resolution screenshot, but it still looks FAR better than Euclideon in full-res.
PT and FOX actually had me fooled for a while. The fact that that was CG was obvious from the first second.[/QUOTE]
The FOX Engine looks great, because it also has really great lighting while the Euclideon video shows scenes with nearly no shadows, it just looks full-bright to me.
So... yeah, you can scan environments and make 3D models out of it. Good job, I guess ? I was under the impression that most video games out there were based on non realistic, made up environments, not exact reproductions of real places.
The only thing that was even remotely convincing to me before he "revealed" that they were all renders was that close-up of the stairs.
You would still need tons of artists to create models, animations, and stuff that simply doesn't exist in reallife.
Where did all this procedurally generated stuff go from the older videos, showing infinite detail with voxels?
The absolute lack of lighting and material definition is really offputting. Everything looks and feels fake, in a bad way. I think the goal of an artist working with real time engines is not to make things perfectly realistic, but to find creative solutions to the limitations you have. 3D scanning an entire environment feels - and looks - like a copout.
[QUOTE=Sgt. Lulz;46023786]the Silent Hills PT and the FOX Engine altogether were far more impressive than this thing.
[IMG]http://i.kinja-img.com/gawker-media/image/upload/s--GsJ2-iBQ--/gbmlfvcthmznnbbyjl8w.jpg[/IMG]
Granted, this is a low-resolution screenshot, but it still looks FAR better than Euclideon in full-res.
PT and FOX actually had me fooled for a while. The fact that that was CG was obvious from the first second.[/QUOTE]
Before I read the text, I did actually think that was a photograph at first.
I fucking despise the intonation he uses in every sentence, a lot of British people on telly seem to use it. Like, at the end of [B][I]every[/I][/B] sentence he goes in pitch from low to high and finishing on low. His voice was the reason I stopped watching. If the whole thing was done in Microsoft Sam I would have gotten further.
The reason the "IRL" video didn't look real, mainly because they were dumb enough to use a flycam in the scene at a speed that is impossible for a camera unless it was on a zipline.
Not to mention the frame skipping and the sudden changes of speed. That doesn't happen in real life. They should have tried to replicate the movements of a camera slowly tracking on a slider or looking hand held. You don't fool people when you don't even bother trying to make it look convincing.
It's a great tech for making a Google-maps style tours for use with VR or something. However for most games it's practically useless. The world is static, textures and lighting are baked and there's no surface information.
[editline]19th September 2014[/editline]
I believe there was a trailer for a PS4 shooter that used (small scale) laser-scanned environments. Don't remember what it was called.
I really want to punch this guy. He reminds me of the avid video game alchemist
[QUOTE=Snickerdoodle;46022343][IMG]http://i.imgur.com/2NMdPKh.png[/IMG][/QUOTE]
So are they basically saying that everything looks really shitty up close?
[QUOTE=Sharker;46025349]So are they basically saying that everything looks really shitty up close?[/QUOTE]
In their first videos when showing that weird island thing, one of the neat features were that things looked really detailed up close, not sure what happened here.
Oh, it's [i][b]these[/b][/i] guys again.
Even though this is all bloated up advertisement, I'm still very interested in voxel technology. If you look at atomontage engine site for example, the developer there shows some cool things and give more insight into the matter. Too bad he stopped updating, I wonder what he's up to.
[QUOTE=Marik Bentusi;46021775]I'd say rendering pretty environments is one of the least important aspects when it comes to improving realistic graphics right now.
I think animations are a much bigger hindrance, it's difficult to make a face "behave human" with just motion capture or to make mocap animations interact with non-mocapped objects and scenarios. Stuff like not making characters' fingers glitch through objects they're supposed to be holding, turning around to face the moving player they're talking to, pupils following objects that follow unscripted paths, etc.
No wonder so many games (still) use masked characters and briefing cutscenes.[/QUOTE]
I've read that the most difficult thing in games is getting two different entities to interact via animation in a believable way. If you're off by just a smidge it looks really off. For example, in a lot of games where it shows a person handing something to someone else, it never looks quite right, or they do a cut-away at the hand-off.
[editline]19th September 2014[/editline]
Also this is reminding me of that one absurd engine that made everything out of points instead of vertices.
They used a remix of the Temple of Time. Wow.
They need to drop the fucking pretentious and sensationalist bullshit they pull, id be taking them a lot more seriously if they were.
The technology seems really nice if it actually works and worthwhile for video games, but jesus fuck, they act like they reinvented the wheel already.
Sorry, you need to Log In to post a reply to this thread.
