@@jessekoepp3928 You obviously have no clue how RUclips works. you're a joke. the content always follows up every title and more. the clickbait also isn't nearly as bad as anything like linus or others. at least i know exactly what im clicking on and the videos always deliver. he tries his hardest and its obvious
@@ThrillSeekerVR OMG you responded, that was INSANE! Please keep using every low T puffs favorite word "toxic", it's INSANE! Cmon Commander Clickbait, respond with more INSANE "comebacks". INSANE!!!!
19:30 That was this sounds is that we might be able to get a "seamless" hybrid tracking system, where you *can* have base stations where you want high accuracy, but it also functions without base stations in a lower quality inside out mode. As much as I lament how unportable the index is, the incredible tracking quality is something I can't give up, and this system seems like the perfect compromise.
The 2.0 stations are fascinating, but in a way, it compromises reliability for higher convenience, which I'd argue makes them more inconvenient over 1.0s, but if you have 3 or more stations, that doesn't matter because you never had a choice. From my experience, the 1.0 stations always provided outstanding tracking performance and do have an upper hand on the 2.0s because 1.0s given their greater simplicity in the tracking system, provides better performance against reflections. I have weird tracking quirks with my 2.0s that I never experienced with my 1.0s and it's because of a few small pieces of furniture in my room that messes with the tracking. With a 1.0, you're only worrying with one laser sweep at a time where 2.0s are transmitting, channel data and such, which makes it far more susceptible to tracking issues if you introduce a slightly reflective surface in the play space. The 1.0s in my opinion don't get the credit they deserve because they are outstanding stations if you're looking for a lighthouse system, are on a budget and don't plan to use more than two stations. I see Quest users buying 2.0 stations all the time and I scratch my head over it because 1.0s would have done the job just as well, at a cheaper cost. I rate 1.0s.
This was a great overview! I would love to see more content about things like this. I think it can really help the community not only appreciate the technology we currently have more, but also understand the challenges that developers have to move forward.
The thing at 16:00 is most likely a so called 'PSD' aka. Position Sensitive Device, used in precision laser tracking metrology applications. I have worked with those for the last 15+ years. It most likely boils down to having lighthouses without rotors, but rather fixed wide angle laser cones. Removing moving parts greatly reduces wear and acoustic noise. The PSD would be behind a small wide anlge 'fish eye' lens, being able to 'see' from which direction the light comes in. Several of those dispersed over the exterior like the classical photodiodes would allow for parallax/perspective/triangulation/trilateration, which is necessary for rangefinding. The individual emitters would need precise timing amongst them since a single PSD only works well if hit by one light source.
Oh I figure the parallax would be instead of sending 3 pluses like they do - you would just do one instead and split the light information base on how it hits the individual sensors. It would remove the need for rotors like you said, but also makes it means it doesnt have to be too complex with light bending.. aka sending the light in the cone shape as you said.
That makes a lot of sense! If they can make the laser cone emitter on the base station low power enough, I wonder if they can make a battery powered base station that, like a battery powereed security camera, only needs to be charged every few months! Imagine a small, AA powered box you can put up in several areas of your house that can give you the tracking fidelity of base stations without the need for any wires!
They do mention use of a PSD, bi-cell or quad cell for this to happen. So they can go either way! I ended up focusing on the quad cell diode because the patent images definitely shown that config more. Either way, really enjoyed the flow of comments in this thread!
@@SadlyItsBradley I'd like to know why they use multicell PSDs, the ones I worked with were technically single celled ones. Their readout was analog, basically all 4 sides of the square substrate had one signal output (plus ground) going into the pre-amp and then ADC. But those were struck by a focused laser beam, mirrored back via a retroreflector mounted to the tracked target / measurement probe. I can only guess that they might have the polarisation filters everyone talks about. Those might even discriminate the axial rotation angle (roll^^) from one single solid-state-lighthouse. I kinda like where this might be leading to, heh :D In the end, all hose ideas can be used in multiple ways, it's just a matter figuring out what's the 'best' way.
The worst thing I hear from some people about lighthouses is when they refer to them as "sensors", while they are actually emitters. Technically speaking, the Valve Index is actually inside-out tracked.
Been using lighthouse since days of Vive Pre, still using lighthouse today. Steamworks dev here with several years of experience building prototype steamVR controllers, thoroughly enjoyed your video and collection of historical pieces Keep doing what you are doing it's very welcome 🍰
If they can get rid of the motor and make a solid-state base station system, especially if the base stations themselves can become smaller and lighter in the process (even the 2.0s are still kinda heavy), it will be a game changer. I'm glad to see Valve continuing to push the idea forward instead of giving up and using camera tracking like everyone else.
I've always wondered how those things worked. This was super informative, thank you for going over this. I always love learning more about VR tech from your videos
I assume like everyone else I was watching this thinking - "please, Brad, tell me that my 5 expensive base stations won't be obsolete by the end of the year!??"
That quad photodiode reminded me of how early Philips single-beam CD players achieved tracking by measuring a shift between 4 cells. This is just additional info about focusing for those who care. For focusing CD players use a cylindrical lens just before the photodiodes that distorts the beam when it isn't in focus. If it's out of focus one way, cells A and C will receive more light than cells B and D, and vice versa; this will send either a + or - voltage to the focus coils to move the lens up or down to compensate. But when focus is perfect, a circle will form, and all A, B, C, and D cells will receive the same amount of light. Haven't fixed a CD player in over 10 years, but that's what I remember from the theory lol.
Been using Logitech trackballs for 15+ years. Glad to see someone else realizing their potential. FYI you can get a 3D printed stand for the Ergo+ to raise it to 60 degrees
Some of the arcade games I have repaired have used that same quad photodiode for gun/aim tracking, such as Jurassic Park Lost World. Another arcade game, Rabbids in Hollywood, actually uses the sixense electromagnet tracking system for their guns. I wonder if that tech will ever find a place in VR since the fall through with oculus.
I've had an idea I haven't had time to try prototyping yet, but I think that electromagnet tracking system could potentially scale down nicely for finger tracking. If you were to put the source electromagnet say on the palm or back of the hand, and electromagnetic sensors near each fingertip, that should solve the position of the fingers very accurately. It should be able to capture things like thumb position and finger splay with much less difficulty than something like capacitive sensors or flex sensors. And I don't think it would have issues with occlusion like camera-based tracking has. I'm just not 100% sure yet whether the physics scales down to work with sensors/emitters of the size needed at the distances needed
@@Dazbii I've had literally that exact same idea verbatim haha! Edit: Another Idea that I've had (that may or may not work well) would be to put the electromagnetic transmitter on the headset itself. The headset would use inside out tracking, and the fullbody and finger tracking would be done through electromagnetism. Now you have full tracking wherever you go without occlusion issues! In reality though, I would imagine that it would be difficult to get both technologies working seamlessly together, and to fit the transmitter in the headset. While the receivers I've seen get pretty small, the transmitters generally are much larger.
Would be interesting to see Meta price match Vision Pro and show us what they can do with the same price point. Varifocal lenses, wider fov, battery life, theres a lot of features Vision Pro is still 'missing' they could capitalize on.
logged into google just to praise your video with a thumbs up! this was a great rush through lighthouse history. only complaint: you did not show the internals of the basestation 2. anyway, please make more such more technical videos! THANK YOU!
I was able to use 4 HTC vives on the same space at events, the trick was using just 2 base stations pointing in the same direction synced with the cable.
It is pretty freaking cool man. It's wild the layers and layers that goes into all this. Awesome video man, really top-notch work. This is the kind of thing I love to know about.
Brilliant!! Thank you! I currently use a modded Samsung Odessey +. The next Valve VR helmet is exactly what I am waiting for unless HP comes out with an OLED version of the G2.
Those quad photodiodes are very similar to what's in a CD player (and DVD and Blu-ray) it can be used to sense if the reflected laser off the disc is in focus and centered. So it's possible that Valve has repurposed a diode design that's complex but still extremely cheap because they're been making them since the 80s.
I knew that lighthouses had rotating parts but man is this technology mindblowingly fascinating. Somehow I find it more fascinating than SLaM tracking even though SLaM is more technically difficult to achieve.
I am a little surprised you didn't mention that the most significant advantage of making the controllers mini lighthouses is that you would then need only a single base station for the whole setup to work. Especially if you put diodes all the way around the headset. Another advantage that should be mentioned is that by making the diodes able to sense the directionality of the laser pulse, you make the system way more flexible. For instance, the original lighthouses were limited on where to put them in the room. These would not be like that at all. Where you positioned them would be a lot more flexible. Lastly, I would be shocked if they didn't make this new base station double as a wifi 7 router that must be plugged into your computer. Allowing for a single lighthouse that could act as both your positioning system and your communication hub to stream the games from your PC.
I've known about the Lighthouse technology, probably at least in part because of Brad, since long before upgrading to the Index, but this was still an interesting video to watch.
Good video, I like the detailed chapters so I can skip the old stuff, and lack of annoying sponsored segments. Please don't upload in HDR for videos that don't need it though. On my Pixel at least it forces the brightness/color to follow your video's mastering, meaning if i am in a dark room or have my phone set at low brightness the video is super bright and vibrant compared to everything else.
Honestly, I had been thinking, if there was a way to improve upon tracking as well, came across your video all because of some lingering thoughts I had about "what if some individuals managed to crack an even better or improved tracking solution to the lighthouses". I honestly would be more than hyped to hear if people were creating more reliable 3rd party lighthouses kind of like how we have these headsets and trackers. It's just something I love about what the community and Valve have, this opensource world is just so much more better, and also enables more curious individuals to take a shot at making some new technology... Unlike facebook which I officially will call the IOI of our world from now on XD
I just sent back a Valve basestation 2.0 because it was jittering. Seems like a common problem especially with the 8KX if you have a motion-rig with actuators causing some sort of EMI (electromagnetic interference) and even using four stations doesn't solve it. Two basestation 1.0 work btw. and probably the reason why they are still quite expensive if you buy used ones.
I think the most important thing you may have overlooked is that by using polirized light filter, they would also elimanate the reflection issues with this system. Mirrors, reflective floor and object would be less of an issue. Thats so far the biggest downside for the lighthouse system, something at which visual systems are a lot better actually.
Thanks for the informative video. It is very interesting. I rather see the future in inside-out racking systems. Wouldn't it be effective to use depth sensors for tracking? Is there any development in this direction?
There’s actually a bit of split opinion on what will be used. It depends on how good processing is. My biggest question with structured light though is whether it would improve low light performance.. which is still a huge problem for camera based systems.
I still use LH1.0 with cable. My LH mounting points are not optimal for that IR link because they need to have unrestricted line of sight. Basically I had to mount them too low but cable was enough long to be routed down the floor and I didn't need to make shortcuts. Only one problem was my ceiling light that I needed to change to non reflective material one. Older light did reflection of one lighthouse so I have about cone from that light through play space where headset and controllers go crazy.
I'm gonna go look at my base stations and the room while they're on with my IR camera now. I don't know why I never thought of that before, I knew they were "laser based" and since I can't see them that means they pretty much have to be IR.
Report: They move REALLY fast, so it's not that interesting actually. Kinda just ends up looking like a strobe light with a little bit of a pattern to it on the wall nearby. And they are decently powerful so it blooms. I guess I was hoping to see like, literally the scan lines as the swept like it's a spy movie or something. Maybe if I had a fog machine :D
Anyone who played bow and arrow games, or billiards games (eg Sportbar VR) knows how essential lighthouse tracking is for finding the controller behind your head or next to your head. The only way I see inside out working is if the controllers have cameras to track themselves in 3D space.
Hi Brad, I'm Drischa! Thank you for the video. I'm a big believer in lighthouse and really like the system, I think because I understand how it works more than SLAM lol. The polarization for tracking different LEDs is a neat idea - I'm a little confused as to how they would track the headset though. I suppose you could have some even dumber base stations (even battery powered if the power draw is low enough due to no motors maybe?) that would be easier to set up. Or do you think the headset will be able to emit its own light to track itself-no idea how that would work really but curious to hear your thoughts. thanks again
Yea I am bit confuse as well. I not a big believer in lighthouse, but if they can figure out a way to where you don't need base stations and everything can happen at the headset/controller level that could be a big game changer as then you wouldnt have extra cost in SoC for tracking that SLAM requires. But it also brings up the question, why do a hybrid system? If you are going to do SLAM already, might as well use SLAM all the way through for both the headset and do the same for the controllers to track themselves.
@@adr2t Oh, that actually made me think, and I think the main benefit would be that its much easier to cover all around the headset with photodiodes, so the tracking volume for the controllers would be much better than most SLAM headsets. That is one of the key benefits of lighthouse to me, I hate losing tracking on my controllers
@@Drischa Thats my point though, SLAM it self has unlimited volume tracking. Problem was that the headset was tracking the controllers - but now that the controller track themselves - it removes the limiting volume tracking that Quest 2 had for example that Quest Pro Controllers don't have.
It is really nice to show them running with a camera without IR filter, but it is actually not good for the rotor to be moved while spun up, especially if of rotate the boxes, as the gyroscopic forces can added much more force on the bearings than normal gravity. The manual also warns against moving them while spinning. Alan Yates actually said they even have sensors to detect such disturbance and turn of the laser, so better to transmit not at all than to transmit bad information. You can actually see the laser turning off from your hand movement sometimes.
Yum. More info. You posted this 6 minutes ago and it's Overr 20 minutes long, so I hope to hear you talk about how the basestations used harvested 15,000rpm motors from Seagate cheetah drives!
@SadlyItsBradley was not disappointed! You could fix the early ones by buying the old drives that died for other reasons. Plus they're a great source of high powered magnets lol
I'd really like maybe a wifi tracking thing like Slime, so you can occlude them, but it uses the moments it sees a base station to reconfirm its positioning and calibrate
Had no idea how lighthouses worked, I just knew they worked way better than the cameras on the stand alone headsets. Can't wait for your next video, I have used immersed but using a wired headset has to be better.
Actually, the inside out cameras have much higher accuracy. See the research paper “Comparing the Accuracy and Precision of SteamVR Tracking 2.0 and Oculus Quest 2 in a Room Scale Setup” > The results reveal a significantly higher accuracy for the Oculus Quest 2 compared to SteamVR Tracking in the height of a tracked object. Furthermore, the Oculus Quest 2 tracks its position with substantially higher precision than SteamVR Tracking.
@@float32 I cannot give you a link to a YT video, because it will probably will be taken down, but the controllers and headsets doesn't have a margin of error of 2mm, literally anyone with a Lighthouse tracked device can test that, and I'm telling you, the Vive trackers, due to the small size and number of photodiodes are way less precise, specially the farther away that you are from the base stations.
That was super interesting, I find it realy cool to finaly understand the base station better... I know that Meta have kill there good old traking system that was used with the old Oculus rift, but I am curious to know what was the big difference betwen the 2 method.... I kind of feel this thechnology was having also a lot of potential, specialy on the enegy cost, went a controler could survive month of use on a signe AA battery
Frickin' laser beams! As much as there's cool technology on the receiving end, I'm sort of amazed that there isn't an "electronic", "variable polarizer based" or "solid state" solution (I'm using quotes because I'm just pulling science words and I don't really know) for the base station that would get rid of the spinny mirrors.
In past talks, Alan Yates did say MEMS or electronically variable optics could make a solid state lighthouse. Which truly would be amazing for people who still enjoy and partake in such tech
you could use photo diode to track the controllers with ir lights coming from the like the quest but not with cameras seeing them. also have accelerometers so that atleast it has a rough position of where the controller is when outside the headset field of view. then do head tracking with cameras backed up by an accelerometer and gyroscope.
I really hope there will be an INDEX 2, PCVR-only, lighthouse based headset. I don't wanna pay for extra hardware like chips... and I don't want unnecessary components on my head. Look at what bigscreen is doing! surely there's a more game-oriented higher fov middle ground solution! Something between index and bigscreen. Investors need to realize there are a few different markets to capitalize on... not just what facebook head-dives into.
Very interesting video ! When I got my og vive in 2018, I was impressed by the precision of the tracking, I didn't though you could do precision shooting in VR. The way they simplified the system is very clever. Logically the 2.0 basestations must have been quieter and more reliable since they only have 1 motor and no IR blaster but they are in fact noisier and seems less reliable from the bad experiences I read, I am very curious to know why. I switched from 2x 1.0 basestations to 3x 2.0 to fix a dead spot but I think I will go back to the 1.0 basestations because of the annoying noise of the 2.0 basestations, especially with the (excellent) off-ear speakers of the index. It's a shame I have to go back to an older tracking system to get a better VR experience. Out of curiosity and for personal challenge, I am trying to make a photodiode amplifier to see the IR signal on an oscilloscope. I am able to see the pulses and modulation (1.8MHz) of the 1.0 basestations but the 2.0 is on an other level, I need a faster op-amp.
Excellent video. Personally I agree with valve about tracking quality being priority. I'd love a next gen system to provide even faster update rates and higher scalability. To for example turn my entire house into a tracked space. Though the lighthouses will need to be cheaper. It still baffles me that the gen2 lighthouses cost more than the gen1 despite being simpler internally. Ideally I'd ditch tracking pucks entirely for full body tracking and wear a full body suit with photodiode tracking. I've read that in theory you can get a track with as few as 3 diodes. So small groups of diodes in a rigid cluster combined with inertia sensors tiled around the suit should provide an amazing experience.. Need to get better wireless headset tech though. Maybe I'll just have to wear a backpack with batteries.
I've been using my 1.0 Lighthouses since I got them in the OG Vive I bought off a VRC friend last year just for my FBT on my quest 2, I was using the Vive but then the cable died a few months later. And I will still be using them when I get an Index because I can save a bit of money and they still work perfectly well.
You can still use your trackers even with quest 2, will only require you to install space calibrator and then calibrate everytime you want to use them together, better than nothing
been waiting for a Lighthouse 3.0 for a while now. It is cumbersome and expensive, but for millimetrical precision... don't know what else could be better. Guess that, combined with a tracking suit... if you aint a plebe :(
When LH1 come out with original vive I can't remember how many times I had to tell others that they are lighthouses and doing lighthouse things :) Basically others also think those were some kind of cameras like oculus had.
With the Quest 2 passthrough you can see the light emitted from the base stations. They also enable you to play your Quest 2 in the dark (if your using the Quest controllers)
I've heard that the LH 2.0 is limited to 4 at a time. That's not to say you can't have more, however a Vive tracker will not be able to connect to more than any 4 of them at a given time. (Maybe you can correct me on that) I'm about to setup a very large space with 8 and my plan was to isolate the laser light spill of 4 from the area of the other 4 as I'm concerned that if all 8 are overlaping eachother in the same area, that there might be a issue with tracking - either because of too many in the same area or because perhaps in the time it takes to switch from one lighthouse to another, the handoff isn't instanious and thus creating the chance for a bit more tracking jitter. (FYI I need absolute precision because we are doing real-time composting for virtual production and sometimes mocap at the same time). From your understanding of the hardware, do you know if these channel handoffs happen in less than a few seconds? Know that will determine how we flag off the lighthouses.
I readed somewhere that the main problem is that the laser sweeps may end up ocurring at the same time and cancel themselves out. in theory as long as you don't have 2 base stations on the same channel, "it should work"
Trackballs. Are. Cool! Strictly better than any mice, for all possible usecases, IMO. Not sure what so special they do for VR that they don't do for flat screens, though. (Or, rather, what mice fail even harder at in VR than on flat screens.) Will have to wait and see, I guess!
They are. I'd still be using mine had I not scaled up to 4 stations. My 1.0s offered better performance against reflective surfaces and always provided an indiscernible difference in tracking quality over the 2.0s. The 1.0s are great little gadgets and easier to get your hands on, not to mention cheaper.
This is my favorite video of yours.
Lighthouse tracking has always been so fascinating. Thanks for the vid brad :D
Cmon, label it "INSANE" like you label almost EVERYTHING, even if it clearly isn't at all, captain clickbait.
@@jessekoepp3928 nailed it right there lmao
@@jessekoepp3928 Your toxicity is INSANE.
@@jessekoepp3928 You obviously have no clue how RUclips works. you're a joke. the content always follows up every title and more. the clickbait also isn't nearly as bad as anything like linus or others. at least i know exactly what im clicking on and the videos always deliver. he tries his hardest and its obvious
@@ThrillSeekerVR OMG you responded, that was INSANE!
Please keep using every low T puffs favorite word "toxic", it's INSANE!
Cmon Commander Clickbait, respond with more INSANE "comebacks".
INSANE!!!!
19:30 That was this sounds is that we might be able to get a "seamless" hybrid tracking system, where you *can* have base stations where you want high accuracy, but it also functions without base stations in a lower quality inside out mode.
As much as I lament how unportable the index is, the incredible tracking quality is something I can't give up, and this system seems like the perfect compromise.
Also having one as a fallback from the other, such as when you get in one of those weird poses your basestations can't see correctly
The 2.0 stations are fascinating, but in a way, it compromises reliability for higher convenience, which I'd argue makes them more inconvenient over 1.0s, but if you have 3 or more stations, that doesn't matter because you never had a choice. From my experience, the 1.0 stations always provided outstanding tracking performance and do have an upper hand on the 2.0s because 1.0s given their greater simplicity in the tracking system, provides better performance against reflections. I have weird tracking quirks with my 2.0s that I never experienced with my 1.0s and it's because of a few small pieces of furniture in my room that messes with the tracking.
With a 1.0, you're only worrying with one laser sweep at a time where 2.0s are transmitting, channel data and such, which makes it far more susceptible to tracking issues if you introduce a slightly reflective surface in the play space.
The 1.0s in my opinion don't get the credit they deserve because they are outstanding stations if you're looking for a lighthouse system, are on a budget and don't plan to use more than two stations. I see Quest users buying 2.0 stations all the time and I scratch my head over it because 1.0s would have done the job just as well, at a cheaper cost. I rate 1.0s.
This was a great overview! I would love to see more content about things like this. I think it can really help the community not only appreciate the technology we currently have more, but also understand the challenges that developers have to move forward.
The thing at 16:00 is most likely a so called 'PSD' aka. Position Sensitive Device, used in precision laser tracking metrology applications. I have worked with those for the last 15+ years.
It most likely boils down to having lighthouses without rotors, but rather fixed wide angle laser cones. Removing moving parts greatly reduces wear and acoustic noise.
The PSD would be behind a small wide anlge 'fish eye' lens, being able to 'see' from which direction the light comes in.
Several of those dispersed over the exterior like the classical photodiodes would allow for parallax/perspective/triangulation/trilateration, which is necessary for rangefinding.
The individual emitters would need precise timing amongst them since a single PSD only works well if hit by one light source.
Oh I figure the parallax would be instead of sending 3 pluses like they do - you would just do one instead and split the light information base on how it hits the individual sensors. It would remove the need for rotors like you said, but also makes it means it doesnt have to be too complex with light bending.. aka sending the light in the cone shape as you said.
May work by just using different pulse widths... If light is detected from one angle, now how long was the light on for in ms...etc
That makes a lot of sense! If they can make the laser cone emitter on the base station low power enough, I wonder if they can make a battery powered base station that, like a battery powereed security camera, only needs to be charged every few months! Imagine a small, AA powered box you can put up in several areas of your house that can give you the tracking fidelity of base stations without the need for any wires!
They do mention use of a PSD, bi-cell or quad cell for this to happen. So they can go either way! I ended up focusing on the quad cell diode because the patent images definitely shown that config more.
Either way, really enjoyed the flow of comments in this thread!
@@SadlyItsBradley I'd like to know why they use multicell PSDs, the ones I worked with were technically single celled ones. Their readout was analog, basically all 4 sides of the square substrate had one signal output (plus ground) going into the pre-amp and then ADC. But those were struck by a focused laser beam, mirrored back via a retroreflector mounted to the tracked target / measurement probe.
I can only guess that they might have the polarisation filters everyone talks about. Those might even discriminate the axial rotation angle (roll^^) from one single solid-state-lighthouse.
I kinda like where this might be leading to, heh :D
In the end, all hose ideas can be used in multiple ways, it's just a matter figuring out what's the 'best' way.
Can't wait for the Lighthouse 2.0: Episode 1, to release!
The worst thing I hear from some people about lighthouses is when they refer to them as "sensors", while they are actually emitters. Technically speaking, the Valve Index is actually inside-out tracked.
Been using lighthouse since days of Vive Pre, still using lighthouse today.
Steamworks dev here with several years of experience building prototype steamVR controllers, thoroughly enjoyed your video and collection of historical pieces
Keep doing what you are doing it's very welcome 🍰
If they can get rid of the motor and make a solid-state base station system, especially if the base stations themselves can become smaller and lighter in the process (even the 2.0s are still kinda heavy), it will be a game changer. I'm glad to see Valve continuing to push the idea forward instead of giving up and using camera tracking like everyone else.
I've always wondered how those things worked. This was super informative, thank you for going over this. I always love learning more about VR tech from your videos
I assume like everyone else I was watching this thinking - "please, Brad, tell me that my 5 expensive base stations won't be obsolete by the end of the year!??"
very important for understanding the Valve Lighthouses - many many thanks - Brad + all the best the next months + more
Thank you for that detailed explanation of how light houses work. Very interesting.
I do always wonder the tech behind that little black box, now I know and appreciate it more. Thanks
That quad photodiode reminded me of how early Philips single-beam CD players achieved tracking by measuring a shift between 4 cells.
This is just additional info about focusing for those who care.
For focusing CD players use a cylindrical lens just before the photodiodes that distorts the beam when it isn't in focus.
If it's out of focus one way, cells A and C will receive more light than cells B and D, and vice versa; this will send either a + or - voltage to the focus coils to move the lens up or down to compensate.
But when focus is perfect, a circle will form, and all A, B, C, and D cells will receive the same amount of light.
Haven't fixed a CD player in over 10 years, but that's what I remember from the theory lol.
This video is criminally underrated!!
Great stuff!
Been using Logitech trackballs for 15+ years. Glad to see someone else realizing their potential. FYI you can get a 3D printed stand for the Ergo+ to raise it to 60 degrees
super cool
Some of the arcade games I have repaired have used that same quad photodiode for gun/aim tracking, such as Jurassic Park Lost World. Another arcade game, Rabbids in Hollywood, actually uses the sixense electromagnet tracking system for their guns. I wonder if that tech will ever find a place in VR since the fall through with oculus.
Oh wow that’s pretty cool!
I've had an idea I haven't had time to try prototyping yet, but I think that electromagnet tracking system could potentially scale down nicely for finger tracking. If you were to put the source electromagnet say on the palm or back of the hand, and electromagnetic sensors near each fingertip, that should solve the position of the fingers very accurately. It should be able to capture things like thumb position and finger splay with much less difficulty than something like capacitive sensors or flex sensors. And I don't think it would have issues with occlusion like camera-based tracking has.
I'm just not 100% sure yet whether the physics scales down to work with sensors/emitters of the size needed at the distances needed
@@Dazbii I've had literally that exact same idea verbatim haha!
Edit: Another Idea that I've had (that may or may not work well) would be to put the electromagnetic transmitter on the headset itself. The headset would use inside out tracking, and the fullbody and finger tracking would be done through electromagnetism. Now you have full tracking wherever you go without occlusion issues! In reality though, I would imagine that it would be difficult to get both technologies working seamlessly together, and to fit the transmitter in the headset. While the receivers I've seen get pretty small, the transmitters generally are much larger.
It's nice to hear that Alan Yates is still at Valve. I was wondering what his current projects are
Oh nice! been waiting for this video to drop 👍
Would be interesting to see Meta price match Vision Pro and show us what they can do with the same price point.
Varifocal lenses, wider fov, battery life, theres a lot of features Vision Pro is still 'missing' they could capitalize on.
Thank you for another video brad ;3
logged into google just to praise your video with a thumbs up! this was a great rush through lighthouse history. only complaint: you did not show the internals of the basestation 2.
anyway, please make more such more technical videos! THANK YOU!
I wanted to open one of my own. But once I found out I’d need some advanced things to do I noped out… I should have just grabbed some pics off the net
@@SadlyItsBradley Bet you could steal our deconstructed one, next time you're around...
The angle sensitive photodiodes + the light scatter detector is a big improvement. More accuracy, and less reflection effect from shiny objects.
Thanks for this. So helpful to get some insights into how the tech works.
Great video Brad, I thought I knew how the light house system functioned, but you are on another level 🙇♀️
Appreciate the long video. Your passion really shows through and it gets me excited too! 😃
I was able to use 4 HTC vives on the same space at events, the trick was using just 2 base stations pointing in the same direction synced with the cable.
It is pretty freaking cool man.
It's wild the layers and layers that goes into all this. Awesome video man, really top-notch work. This is the kind of thing I love to know about.
Nice one
Top work. Very informative
This imply there's a heavymansion tracking somewhere out there
I’m more looking forward to DarkTent tracking, I hear it’s more mobile and doesn’t depend on lasers
Still sitting on my hands for the Foghorn tracking drop.
@@EvKem nolo uses ultrasound, that could be called foghorn
Massive castle tracking
@@SoggyMicrowaveNugget I say sir you’ve gone too far!
Brad, that was a great video! I learned so much! Thank you!
good one Brad, good work, good description, good job....
Stellar video, fascinating content!!
Brilliant!! Thank you! I currently use a modded Samsung Odessey +. The next Valve VR helmet is exactly what I am waiting for unless HP comes out with an OLED version of the G2.
This explains the Lighthouse revisions really well. I understand now- my brain is slightly more wrinkled.
This was a good explanation of how lighthouse tracking works, i learned a good bit, thanks
I enjoyed this video very much. Color me a fan of lighthouse tracking.
Great video! Very informative
Those quad photodiodes are very similar to what's in a CD player (and DVD and Blu-ray) it can be used to sense if the reflected laser off the disc is in focus and centered. So it's possible that Valve has repurposed a diode design that's complex but still extremely cheap because they're been making them since the 80s.
I knew that lighthouses had rotating parts but man is this technology mindblowingly fascinating. Somehow I find it more fascinating than SLaM tracking even though SLaM is more technically difficult to achieve.
Super interesting stuff, thanks for sharing. I always found the lighthouse technology fascinating!
I am a little surprised you didn't mention that the most significant advantage of making the controllers mini lighthouses is that you would then need only a single base station for the whole setup to work. Especially if you put diodes all the way around the headset. Another advantage that should be mentioned is that by making the diodes able to sense the directionality of the laser pulse, you make the system way more flexible. For instance, the original lighthouses were limited on where to put them in the room. These would not be like that at all. Where you positioned them would be a lot more flexible. Lastly, I would be shocked if they didn't make this new base station double as a wifi 7 router that must be plugged into your computer. Allowing for a single lighthouse that could act as both your positioning system and your communication hub to stream the games from your PC.
As a life time track ball user I agree they are amazing .
LH tracking is my favorite high speed and accurate tracking method.
I also love my LH v1 ones since they are the most high tech I own.
💚Fascinating work Brad! Thanks for breaking down this highly technical discussion, you are a master at it!
I want more Pappy Bradley vr history fun time!
I've been using trackballs for over 13 years. I even use the the trackball mouse on laptops!!!
I've known about the Lighthouse technology, probably at least in part because of Brad, since long before upgrading to the Index, but this was still an interesting video to watch.
Good video, I like the detailed chapters so I can skip the old stuff, and lack of annoying sponsored segments.
Please don't upload in HDR for videos that don't need it though. On my Pixel at least it forces the brightness/color to follow your video's mastering, meaning if i am in a dark room or have my phone set at low brightness the video is super bright and vibrant compared to everything else.
I understand a bit better why I want to upgrade from my 1.0s on my walls now haha
Great video, awesome to understand what's going on under the hood.
Honestly, I had been thinking, if there was a way to improve upon tracking as well, came across your video all because of some lingering thoughts I had about "what if some individuals managed to crack an even better or improved tracking solution to the lighthouses". I honestly would be more than hyped to hear if people were creating more reliable 3rd party lighthouses kind of like how we have these headsets and trackers. It's just something I love about what the community and Valve have, this opensource world is just so much more better, and also enables more curious individuals to take a shot at making some new technology... Unlike facebook which I officially will call the IOI of our world from now on XD
Lighthouse Alyx
I just sent back a Valve basestation 2.0 because it was jittering. Seems like a common problem especially with the 8KX if you have a motion-rig with actuators causing some sort of EMI (electromagnetic interference) and even using four stations doesn't solve it. Two basestation 1.0 work btw. and probably the reason why they are still quite expensive if you buy used ones.
I think the most important thing you may have overlooked is that by using polirized light filter, they would also elimanate the reflection issues with this system. Mirrors, reflective floor and object would be less of an issue. Thats so far the biggest downside for the lighthouse system, something at which visual systems are a lot better actually.
Thanks for the informative video. It is very interesting. I rather see the future in inside-out racking systems. Wouldn't it be effective to use depth sensors for tracking? Is there any development in this direction?
There’s actually a bit of split opinion on what will be used. It depends on how good processing is. My biggest question with structured light though is whether it would improve low light performance.. which is still a huge problem for camera based systems.
I still use LH1.0 with cable. My LH mounting points are not optimal for that IR link because they need to have unrestricted line of sight. Basically I had to mount them too low but cable was enough long to be routed down the floor and I didn't need to make shortcuts.
Only one problem was my ceiling light that I needed to change to non reflective material one. Older light did reflection of one lighthouse so I have about cone from that light through play space where headset and controllers go crazy.
I'm gonna go look at my base stations and the room while they're on with my IR camera now. I don't know why I never thought of that before, I knew they were "laser based" and since I can't see them that means they pretty much have to be IR.
Report: They move REALLY fast, so it's not that interesting actually. Kinda just ends up looking like a strobe light with a little bit of a pattern to it on the wall nearby. And they are decently powerful so it blooms. I guess I was hoping to see like, literally the scan lines as the swept like it's a spy movie or something. Maybe if I had a fog machine :D
Anyone who played bow and arrow games, or billiards games (eg Sportbar VR) knows how essential lighthouse tracking is for finding the controller behind your head or next to your head. The only way I see inside out working is if the controllers have cameras to track themselves in 3D space.
i tried 1.0 and 2.0. i love it both ❤
Hi Brad, I'm Drischa!
Thank you for the video. I'm a big believer in lighthouse and really like the system, I think because I understand how it works more than SLAM lol. The polarization for tracking different LEDs is a neat idea - I'm a little confused as to how they would track the headset though. I suppose you could have some even dumber base stations (even battery powered if the power draw is low enough due to no motors maybe?) that would be easier to set up. Or do you think the headset will be able to emit its own light to track itself-no idea how that would work really but curious to hear your thoughts. thanks again
Yea I am bit confuse as well. I not a big believer in lighthouse, but if they can figure out a way to where you don't need base stations and everything can happen at the headset/controller level that could be a big game changer as then you wouldnt have extra cost in SoC for tracking that SLAM requires. But it also brings up the question, why do a hybrid system? If you are going to do SLAM already, might as well use SLAM all the way through for both the headset and do the same for the controllers to track themselves.
@@adr2t Oh, that actually made me think, and I think the main benefit would be that its much easier to cover all around the headset with photodiodes, so the tracking volume for the controllers would be much better than most SLAM headsets. That is one of the key benefits of lighthouse to me, I hate losing tracking on my controllers
@@Drischa Thats my point though, SLAM it self has unlimited volume tracking. Problem was that the headset was tracking the controllers - but now that the controller track themselves - it removes the limiting volume tracking that Quest 2 had for example that Quest Pro Controllers don't have.
I love lighthouses. I hope they never die.
It is really nice to show them running with a camera without IR filter, but it is actually not good for the rotor to be moved while spun up, especially if of rotate the boxes, as the gyroscopic forces can added much more force on the bearings than normal gravity. The manual also warns against moving them while spinning. Alan Yates actually said they even have sensors to detect such disturbance and turn of the laser, so better to transmit not at all than to transmit bad information. You can actually see the laser turning off from your hand movement sometimes.
Wow video is shot using HDR and it looks so good on mini led iPad! Bright colors are almost blinding you!
That giant cable between the base stations was still some of the best tracking for me
Yum. More info.
You posted this 6 minutes ago and it's Overr 20 minutes long, so I hope to hear you talk about how the basestations used harvested 15,000rpm motors from Seagate cheetah drives!
I mentioned they used hard drives in the early ones. But didn't mention this exact tidbit
@SadlyItsBradley was not disappointed! You could fix the early ones by buying the old drives that died for other reasons.
Plus they're a great source of high powered magnets lol
I see "Valve" and i insta click. Thanks for keeping us informed :)
I'd really like maybe a wifi tracking thing like Slime, so you can occlude them, but it uses the moments it sees a base station to reconfirm its positioning and calibrate
Had no idea how lighthouses worked, I just knew they worked way better than the cameras on the stand alone headsets. Can't wait for your next video, I have used immersed but using a wired headset has to be better.
Actually, the inside out cameras have much higher accuracy. See the research paper “Comparing the Accuracy and Precision of SteamVR Tracking 2.0 and Oculus Quest 2 in a Room Scale Setup”
> The results reveal a significantly higher accuracy for the Oculus Quest 2 compared to SteamVR Tracking in the height of a tracked object. Furthermore, the Oculus Quest 2 tracks its position with substantially higher precision than SteamVR Tracking.
@@Steam_VR that’s not my opinion. It’s a literal research paper where they measure it. Any reference?
@@float32 I cannot give you a link to a YT video, because it will probably will be taken down, but the controllers and headsets doesn't have a margin of error of 2mm, literally anyone with a Lighthouse tracked device can test that, and I'm telling you, the Vive trackers, due to the small size and number of photodiodes are way less precise, specially the farther away that you are from the base stations.
@@Steam_VR just put the video id, not the full link.
@@float32 htt ps: // youtu. be/Uzv2H3PDPDg
That was super interesting, I find it realy cool to finaly understand the base station better... I know that Meta have kill there good old traking system that was used with the old Oculus rift, but I am curious to know what was the big difference betwen the 2 method.... I kind of feel this thechnology was having also a lot of potential, specialy on the enegy cost, went a controler could survive month of use on a signe AA battery
Frickin' laser beams! As much as there's cool technology on the receiving end, I'm sort of amazed that there isn't an "electronic", "variable polarizer based" or "solid state" solution (I'm using quotes because I'm just pulling science words and I don't really know) for the base station that would get rid of the spinny mirrors.
In past talks, Alan Yates did say MEMS or electronically variable optics could make a solid state lighthouse. Which truly would be amazing for people who still enjoy and partake in such tech
Love it, thanks Brad!
Always love looking at my basestations through the quest 2. You can see it spinning.
you could use photo diode to track the controllers with ir lights coming from the like the quest but not with cameras seeing them. also have accelerometers so that atleast it has a rough position of where the controller is when outside the headset field of view. then do head tracking with cameras backed up by an accelerometer and gyroscope.
That was such a good engineering candy. Thank you very much!
Awwww thanks for the shout-out to both the companies I'm involved with. Tundra and Triad. TnT. Lol
Thanks for sharing. I am still patiently waiting for Valve Deckard even though I probably should just get Index and start playing...
I really hope there will be an INDEX 2, PCVR-only, lighthouse based headset.
I don't wanna pay for extra hardware like chips... and I don't want unnecessary components on my head.
Look at what bigscreen is doing! surely there's a more game-oriented higher fov middle ground solution!
Something between index and bigscreen.
Investors need to realize there are a few different markets to capitalize on... not just what facebook head-dives into.
Very interesting video !
When I got my og vive in 2018, I was impressed by the precision of the tracking, I didn't though you could do precision shooting in VR.
The way they simplified the system is very clever. Logically the 2.0 basestations must have been quieter and more reliable since they only have 1 motor and no IR blaster but they are in fact noisier and seems less reliable from the bad experiences I read, I am very curious to know why.
I switched from 2x 1.0 basestations to 3x 2.0 to fix a dead spot but I think I will go back to the 1.0 basestations because of the annoying noise of the 2.0 basestations, especially with the (excellent) off-ear speakers of the index. It's a shame I have to go back to an older tracking system to get a better VR experience.
Out of curiosity and for personal challenge, I am trying to make a photodiode amplifier to see the IR signal on an oscilloscope. I am able to see the pulses and modulation (1.8MHz) of the 1.0 basestations but the 2.0 is on an other level, I need a faster op-amp.
22:42 Yep, thanks for explaining.
Interesting, cool to hear about this tech
Excellent video.
Personally I agree with valve about tracking quality being priority.
I'd love a next gen system to provide even faster update rates and higher scalability. To for example turn my entire house into a tracked space. Though the lighthouses will need to be cheaper.
It still baffles me that the gen2 lighthouses cost more than the gen1 despite being simpler internally.
Ideally I'd ditch tracking pucks entirely for full body tracking and wear a full body suit with photodiode tracking.
I've read that in theory you can get a track with as few as 3 diodes. So small groups of diodes in a rigid cluster combined with inertia sensors tiled around the suit should provide an amazing experience..
Need to get better wireless headset tech though. Maybe I'll just have to wear a backpack with batteries.
i cant belive the first tracking system for vr invented is literally still the best one
Love this video, very facilitating
I've been a trackball fan boy for decades.
I've been using my 1.0 Lighthouses since I got them in the OG Vive I bought off a VRC friend last year just for my FBT on my quest 2, I was using the Vive but then the cable died a few months later. And I will still be using them when I get an Index because I can save a bit of money and they still work perfectly well.
You can still use your trackers even with quest 2, will only require you to install space calibrator and then calibrate everytime you want to use them together, better than nothing
@@exdorms I know, and I've been doing they, but recently I've been getting crummy results
Would love a video on what you think AI means for VR Brad. Love your work mate.
been waiting for a Lighthouse 3.0 for a while now. It is cumbersome and expensive, but for millimetrical precision... don't know what else could be better. Guess that, combined with a tracking suit... if you aint a plebe :(
I have the Pimax Crystal and the inside out tracking is pretty good at least for Sim Racing... idk about full body tracking.
Damn finally know wtf those things are actually doing. The term "lighthouse" makes a lot more sense now.
When LH1 come out with original vive I can't remember how many times I had to tell others that they are lighthouses and doing lighthouse things :)
Basically others also think those were some kind of cameras like oculus had.
With the Quest 2 passthrough you can see the light emitted from the base stations. They also enable you to play your Quest 2 in the dark (if your using the Quest controllers)
I've heard that the LH 2.0 is limited to 4 at a time. That's not to say you can't have more, however a Vive tracker will not be able to connect to more than any 4 of them at a given time. (Maybe you can correct me on that)
I'm about to setup a very large space with 8 and my plan was to isolate the laser light spill of 4 from the area of the other 4 as I'm concerned that if all 8 are overlaping eachother in the same area, that there might be a issue with tracking - either because of too many in the same area or because perhaps in the time it takes to switch from one lighthouse to another, the handoff isn't instanious and thus creating the chance for a bit more tracking jitter.
(FYI I need absolute precision because we are doing real-time composting for virtual production and sometimes mocap at the same time).
From your understanding of the hardware, do you know if these channel handoffs happen in less than a few seconds? Know that will determine how we flag off the lighthouses.
I readed somewhere that the main problem is that the laser sweeps may end up ocurring at the same time and cancel themselves out.
in theory as long as you don't have 2 base stations on the same channel, "it should work"
@nyae Good point. Will have to check how many channels are available.
I just hope the 1.0s are still gonna be supported in the near future.
Their range and having only 2 is totally fine for me.
I always wondered how the valve's lighthouse works... now I know, and it is freakin' cool, laser beams! :)
Trackballs. Are. Cool! Strictly better than any mice, for all possible usecases, IMO.
Not sure what so special they do for VR that they don't do for flat screens, though. (Or, rather, what mice fail even harder at in VR than on flat screens.) Will have to wait and see, I guess!
Im a huge fan of 1.0 lighthouses
They are so freaking reliable but might not be optimal if i ever did full body tracking..
They are. I'd still be using mine had I not scaled up to 4 stations. My 1.0s offered better performance against reflective surfaces and always provided an indiscernible difference in tracking quality over the 2.0s. The 1.0s are great little gadgets and easier to get your hands on, not to mention cheaper.
Cool! Dev of the kennect (Xbox) would be interesting too!😊
If I had to choice to hang up or stand up fewer tracking systems in a room verses putting on a tracking suit, I would choose the lighthouse style.