4 ms·
And that my friends, is why Virtual Reality will never ever be real, or even look real. I know it's truism, but the sheer realization of that makes me empty and
by bibelo 3y ago
And that my friends, is why Virtual Reality will never ever be real, or even look real. I know it's truism, but the sheer realization of that makes me empty and desperate (jk).
Just trying to simulate realism would mean you take an object in your hand in VR, and are able to examine it closely, seeing scratches or others details.
Of course, these details must not be made using procedural generation. That would mean going back to square 1, a superficial world of shallow details. We want details that have a history.
So imagine you're able to find a way to scan these details and reproduce them in VR. You'd have the same problem at a deeper level, say using a magnifier, only revealing the void of your digital life and nothing around has meaning.
Say now you're able to scan details at deeper levels, even at an infinitely deep level. You'd have to store them, and that would probably mean using more storage that they are atoms in this world, so it would be impossible, or would it? Would it be possible to compress details, or vectorize them, so that they'd take less weight that their equivalent?
Another problem would be details for objects that exist only in VR. Here again, these details would need to have a history. That scratch would be here or there and in that shape for a specific reason. That would require a deep understanding of physical laws, and being able to reproduce them at infinitely smaller levels. And store them here again.
So I guess that makes the whole task impossible. I guess it'll never be possible to transfer your soul into VR, and even if it were, you'd become crazy realizing your world is limited, not by the geographical limits you're bound to, but by the depth of details you'd be able to explore.
- imadierich 3y ago[dead]
- immibis 3y agoJust don't put magnifiers in your VR game. Problem solved.
- persnickety 3y ago> We want details that have a history. For the VR to be real, sure. But for the VR to look good? I don't think so. People don't care about history when they care about looks. Case in point: pre-damaged jeans. Procedural generation is also not failing when it's lacking details, but when the large scale stuff looks off. Take a basic example of grass patterns in games: if they are made non-repeating, then at the level of detail the game is about, they have all the realism you could ask for.
- Tyr42 3y agoSpeaking of grass distribution. https://caseymuratori.com/blog_0009 https://caseymuratori.com/blog_0009 There is a lot of detail that goes into it sometimes.
- looping8 3y agoI don't think virtual reality should feel real. It's the same as animation or painting - if you can make it look real, of course it takes skill and looks impressive, but these things let you create absolutely unreal things. Space monsters, landscapes like Van Gogh did, talking fish, whatever one can imagine. Of course it will not have details, but when you know something isn't real, the details will not matter as much.
- jebarker 3y agoThere's a similar point to be made about simulation. When philosophers suggest we might live in a a simulation I always wonder where the simulation would be running such that all this detail could be coherently generated.
- r2_pilot 3y agoThe simulation only has to simulate one mind with enough details to keep the one mind occupied.
- butlike 3y agoIt can be lonely at the top, so maybe go for two.
- butlike 3y agoYou could probably snapshot the universe in a hologram-esque structure when it's in a state of high entropy/low kinetic energy
- BlueTemplar 3y agoWho said that it had to be ? Paintings / Photographs / Movies / Video Games aren't real either, and still we appreciate them nonetheless. Some of the issues that you raise up can still probably dealt with. It's a shame that TFA stops at the molecular level, because lower than that things are different. To start with, details aren't infinite. You literally cannot tell elementary particles apart, because they do NOT have detail, and this has wide-ranging implications. One of the laws of physics that is deemed to be the most inviolable is the law of paraconservation of information : our ability to transform energy into useful work depends on the information we have about the system in question ! Related to this, the number of involved particles is typically so huge that you can assume fluctuations to be negligible. So for instance what one could do, is to only simulate the reality deep enough depending on the momentary attention the user is paying to a specific part of it, and rely on statistics otherwise. And only store the history of those parts, and only for as long as they are expected to persist. You could also provide intricate and this time literally infinite amount of detail using fractals - if exact reproduction isn't the goal. «Transferring your soul» is a way harder question, since you would have to first figure out what you even mean by «soul».
- ramses0 3y agoOnce upon a time, I was furious with my 4th grade science teachers for teaching me LIES... the Niels Bohr Model of the Atom, bringing coathangers and styrofoam balls to school on the bus when TRULY the atom was an electron cloud of probabilities, etc. etc. All that time wasted, all those coathangers and styrofoam ball USELESS because that wasn't what was really going on. Cue several years later and I'm trying to model my kitchen in sketchup to figure out if something is going to fit, or if some remodeling project is worthwhile or even feasible. I start measuring and modeling the countertops, and question: do I need to account for the countertop overhang, or can I just treat them as cubes? I needed the overhang. Do I need to account for the "toe-kick" (cutout where your feet go instead of banging the cabinets)? Maybe? What about the cabinet doors? Further: what about the inlays? The handles? What about the insides? The supporting structures, the screws and studs? I then thought back to the humble "Niels Bohr Model of the Atom" and forgave my teachers, because I had just then realized: "A model must be just good enough to be useful, and sometimes too much detail isn't helpful (or useful)" It's an extremely useful perspective to have: First decide the utility (the questions you'd like to answer) and then choose the most effective ROI to get there. Further: having varying levels of "zoom" or detail with different acquisition, maintenance, or usage costs isn't out of the question!
- belugacat 3y agoWalls and floor/ceiling not being parallel (old house) tripped me up when I modeled my perfect kitchen in abstract 3D software land and then tried to make it happen in real life.
- manx 3y agoIt might be possible with generative AI at some point...
- datameta 3y agoIn due time, I believe in-situ AI rendering will allow us to generate new zoom layers upon further inspection of an object. Do I think a generation of kids growing up with AI-enhanced VR to be dystopian in a way? Yeah.