True scale, anchored to your floor. Put the headset on first.
Check the build stamp before reporting a bug. The line above and the bottom of the in-VR readout both show it. If it is not the build you were told to test, the headset is showing a cached page — pull down to hard-reload and try again.
Pick any video on this PC — nothing to copy, nothing to restart:
…or just drag a video file onto this box.
Failing that: put a file in the vr folder called monitor.mp4
(or video.mp4 / screen.mp4 / sim.mp4 / .webm)
and it is picked up automatically, or add ?video=myfile.mp4 to the URL.
looking for a video…
A YouTube link cannot work here, and it is not something you typed wrong. YouTube only hands out its video inside its own player window, and a player window cannot be put on a surface inside a 3D scene — the headset has no way to read the picture back out of it. It has to be a file sitting on this PC. Download the clip first and pick it above.
| Controls | |
|---|---|
| Left stick ↑↓ | slide it forward / aft, the way you are facing |
| Left stick ←→ | slide it left / right |
| Right stick ↑ | lower it |
| Right stick ↓ | raise it |
| Right stick ←→ | yaw — spin it about its own centre |
| Right grip + left stick click |
swap between the SEATED and STANDING views. Seated is the default now. It drops the whole machine 145 mm so that sitting in your office chair puts your eyes where they would be sitting in the rig — 470 mm chair minus the 230 mm the rig seat sits above the platform plate = a 240 mm drop. Standing is the physically true 385 mm, which is what you line the model up against on the real platform. The reported heights on the readout board do not change between the two — they are measurements of your machine, not of the view. A vertical ruler out at the right-hand side draws the floor-to-plate distance live, with a green mark where the seat lands, so you can check it from the chair. |
| Right grip + right stick ↑↓ |
trim the seated view: nudges the 230 mm seat-above-the-plate figure the drop is computed from, 200–360 mm, and the board shows the resulting figure rather than a delta. Forward lowers the machine, same as the ungripped stick. Leave it at your measured 230 unless you want to sit differently — it is a comfort adjustment, not a correction. Before 12 Aug 2026 the model was drawn 2 mm high (the placement hung the machine off a 2 mm reference grid line rather than off the platform plate), and setting this to 232 was a valid way to cancel that by hand. That is fixed at source now, so 232 would leave you 2 mm low. Put it back to 230. |
| The height mark (no button — it is always there) |
a bar fixed in the room, under the machine's forward edge, at the height the live view mode says the underside should sit. It follows the rig sideways and in yaw but never up and down — that is the point: the ruler rides the machine, so nothing on it ever looks wrong, whereas this one stays put and the machine moves past it. Green and reading ON TARGET means you are there (within 5 mm / 0.20"); amber gives the miss in mm and inches and says which way to push the stick. The readout board carries the same figure in words. Switching SEATED/STANDING or trimming the seat moves the target with you. |
| Either trigger | reset: 1.2 m in front of you, back to parked height. Ignored while a grip is held — squeezing a grip pulls the trigger too, and that used to teleport the rig mid-pin. Also ignored, by both hands, while the keyboard tray or the yoke handle is being held — see below |
| Point at a joint, then hold the trigger |
move ONE joint and nothing else. A thin blue ray comes out of each
controller. Sweep it over the arm and any joint it crosses lights up with
its own axis drawn through it, so you can see which freedom you are
about to move before you commit. Hold the trigger and that joint follows your
hand along that axis only — a yaw joint tracks your sideways
sweep, a pitch joint tracks up and down, the yoke handle tracks fore and aft. Everything
else stays exactly where it is, and each joint stops at its real limit. Pointable: base yaw, middle-joint yaw, middle-joint lift (that one is the 10.6" of height), tray level, tray tilt, and the yoke handle. The board names the joint and shows its value in degrees, or mm and inches for the handle, so you can read a setting off it and go and set the real part to it. The tray itself is now the target for its own two axes, and the rule is one line: point at the LEFT-hand half of the tray to TILT it up and down, the RIGHT-hand half to SWIVEL it level. Both worked before but each had one small ball somewhere on a 600 mm slab, which in the middle of the rig meant you had to already know where to aim. Almost the whole tray is live now, and the axis is drawn through the real pivot rather than through the spot you happen to be pointing at, so you can see which way it is about to move before you pull. This does not replace grabbing the tray — the two are for different jobs. Grab it to sling the whole arm somewhere roughly right; point at a joint to change one thing by a known amount. Putting your hand on the tray wins if you do both at once. |
| Trigger, hand on the tray (keyboard model and now the rig) |
grab the keyboard tray and move it. Reach out until the tray lights up
blue, squeeze the trigger, and drag — the wall swivel, the middle joint and its
gas spring all solve to follow your hand, stopping dead at the published limits
(15–37.8" of reach, 10.6" of lift, 180° swivel). Twist your wrist to angle the
tray (±45° level) and tip the controller to tilt it (+90 / −20°).
The tray goes orange while you have hold of it. Reach and tray height are on
the in-VR readout board in mm and inches while you move it, so you can read
off a position you like. Let go of the trigger to drop it. It is the trigger and not a
grip because every grip is already spoken for — and the trigger only grabs when
your hand is actually on the tray, so it still resets the rig everywhere else. The arm is bolted into the full rig, on the rear face of the right-hand monitor post, and its height is solved, as high as it will go. The screen leans back over that very face — its bottom edge is under a millimetre aft of it — so the mount hardware is genuinely underneath the glass. Every corner of the plate, pole, collar, lower arm and middle-joint post is swept through the whole swivel, everything outside the screen's plan-view footprint is thrown away, and the tallest thing left is set 10 mm below the screen's bottom edge. That puts the plate top edge 10 mm under the monitor and the tray between 838 and 1309 mm (33.0 and 51.5") off the floor. You can still drive the tray into the screen, and that is deliberate. Raise it fully and swing it inboard and it goes up to 437 mm past the glass. The gas lever and the tilt handle are hand controls you can see; solving the plate against them would have dropped it 447 mm and made the whole mounting useless. The gantry post and the yoke-rest crossbeam are both still in the file as alternatives to compare against. |
| How the arm actually moves | The wall bracket is fixed: it takes a vertical pin and the arm clamps
to it, so its height is chosen once at installation and never moves again. The first arm
only swings left and right about that pin. All of the height lives at
the middle joint, on the gas spring, which also swings laterally. The
tray stays level through all of it. The gas arm drops well below horizontal — about 30° down, read off your video frame. The model used to stop it dead at level, which was my clamp and not the hardware, and it is the direction that matters most for getting the tray down to a seated height. End to end the modelled swing moves the tray 471 mm, against a published "10.6 inches of height adjustment" (269 mm) — those are not the same measurement, so the readout shows both. The first arm RISES, and the model is built at +5°. That is your call and it is what the model draws — a deliberately conservative version of what you read off the video frame by frame. Every clearance below is quoted at +5° first, as the working answer. The other angles are still printed after it (−8 / 0 / +8 / +17°) purely as a sensitivity check, so you can see how much a protractor reading would actually change — they are not four equal guesses any more. Where an answer changes sign across that spread, that instability is worth knowing, but +5° is the number the model is built at. The published 37.8" of reach is only available at the bottom of the height travel — raising the tray shortens it, exactly as it does on the real arm. |
| Trigger, hand on the yoke handle (full rig) |
pull the yoke handle in and out. The shaft comes straight out
of the yoke's rear face, horizontally, and turns 90° upwards
at its aft end — that upward leg is the part you grab. It lights up blue when your
hand is on it; squeeze the trigger and pull. It slides fore and aft only, and stops dead
at 105 mm (pushed fully forward, the shaft swallowed back into the yoke)
and 260 mm (pulled fully aft), both measured the way you measured them:
from the yoke's rear face to the rearmost edge of the handle. It goes orange while you
have hold of it. The distance is on the in-VR readout board in mm and inches,
along with the height the grab leg ends up at off the floor. It rides the crossbar, so
X / Y move it too. The shaft sits where you measured it on that face: 55 mm up from the bottom edge and 55 mm in from the left-hand edge. Your 35 mm down from the top edge agrees with the 55 up exactly — the face measures 100 mm, so the two figures close on a 20 mm shaft, which is the diameter drawn. The old 135 mm takeoff-trim mark is gone. It was not a stop and there is no new trim figure, so the bar now just shows the plain 105–260 range rather than a mark that could be planned off. Only the travel and those three offsets are measured. Tube diameters and how far the grab leg rises are an envelope so there is something to hold; do not scale a part off them. |
| Trigger, hand on the avionics console (full rig) |
slide the RealSimGear console up and down. Reach out until it lights up
blue, squeeze the trigger, and move your hand up or down — it follows, and
only up and down. Sideways and fore/aft hand drift are ignored, exactly
as they are on the yoke handle. It goes orange while you have hold of it, and the board
shows the top edge of the console off the floor in mm and inches along
with how far off centre it is. Travel is 150 mm — 5.9" — quoted as
±75 mm about where the CAD draws it, which is mid travel. The page
no longer holds that figure: it reads it out of the model file it just loaded, so
rebuilding the CAD is enough to change it here. It slides sideways too, and the part you grab is what picks the axis. The console rides vertical slots cut in its bracket; the bracket rides the crossbar's two top slots. So grab the console body and it goes up and down; put your hand on the flat steel bracket lying on TOP of the crossbar and it goes left and right (±30 mm), taking the console with it. That is not a convention invented for this page — it is which part actually moves in which direction on the real mechanism. The split is the underside of that flat bracket: hand below it → up/down, hand on or above it → sideways, and the readout names the axis while you hold it, so there is no guessing which one you got. (It used to be the crossbar's top face, which was the same plane while the console's top was flush with it. The rebuilt bracket hangs the console 98 mm lower, so the boundary is now taken off the part that does not move.) The ±30 mm sideways figure is not a measured limit and should not be planned off. It came from a study that has since been retired, of a bracket whose top leg had 69 mm slots. The bracket actually being built has plain round M6 clearance holes on T-nuts, and a T-nut slides the whole length of the crossbar's slot — so what really stops it sideways is hitting the centre stick or running off the end of the crossbar, and nothing has worked that out yet. Decision needed. Both axes are also pointable from the seat — sweep the ray over the console and its bracket and the marker draws the direction of travel through them before you pull the trigger. The 150 mm is the real bracket's, not a study's: one continuous 211.5 mm channel per side, so both of the console's bolt rows ride the same slot and their 55 mm pitch stops being a limit. Both figures are offsets about the position the CAD draws, and the CAD draws the console at mid travel. |
| X / Y (left hand) | the keyboard plate down / up its post — slides the wall plate,
and the whole arm with it, up and down the rear face of the right-hand monitor post,
continuously while held. Its height off the floor is on the in-VR readout in
mm and inches, with the trim beside it, so you can find a position by eye and
then go and set the real bracket to that number. The range is measured off the post every time the page loads, never typed: about −455 mm down and +368 mm up about the solved height. It will go well above the monitor's bottom edge on purpose — you asked to judge that by eye rather than have the model refuse it. The only stop is the plate running off the end of the post it is bolted to, and if it ever gets clamped the board says so in red rather than the arm quietly vanishing. In a model with no keyboard arm (the slide study) X / Y still drive the control crossbar, so the binding is never dead. |
| Right grip + X / Y | control crossbar down / up — slides the crossbar, the side stick,
the GCU stack and the whole yoke assembly (yoke, clamp plates, its
forward-foot shelf and the yoke-rest crossbeam that carries it) as one, without moving the rig.
This is where the old bare X / Y went. Travel is
measured off this model every time the page loads,
never typed here: up 0 mm — there is none left, down roughly
−152 mm on the current CAD. DOWN is set by the
carriage meeting the structure over the pedals, not by the slider
plates — the plates alone would allow −370 mm, which is the figure this
page used to quote and the one that reached Paul's planning.
Do not plan off −370.
Run w3s.check() in the console for the live figure and what is limiting it. |
| A / B (right hand) | next / previous monitor size — cycles four real panels |
| Left grip | aspect mode — cycles FIT / FILL / 16:9 / 21:9 / 32:9. FIT shows your clip at its true shape with bars; the rest draw a game window at that aspect and fill it, so you can see how much of the glass a game would actually cover |
| Left stick press | aspect guides on / off — draws all four footprints on the screen at once, labelled by colour. Works with no video loaded |
| Right grip + right stick ←→ | skip ±30 s, one jump per push. Yaw is suppressed while the grip is held so the rig does not spin while you scrub |
| Right stick press | play / pause |
| BOTH grips together | pin it to the room — drops a spatial anchor so the rig is in the exact same physical place next time you put the headset on. Nudging it with a stick releases the pin; press both grips again when it is where you want it |
| Right grip + A / B | next / previous video — cycles every clip in the
vr folder, no reload |
| Walk | just walk. It stays put — it is pinned to the room, not to your head |
| What each aspect covers on the 40" panel (929 × 392 mm) | |
|---|---|
| 16:9 | 697 mm (27.4") wide — 75% of the width, pillarboxed. This is your footage, and it is why the clip looks narrow on this panel |
| 21:9 | 929 mm (36.6") — 100%. It fills the panel exactly. Not approximately: 2560×1080 is 2.3704:1 and 5120×2160 is the same number — 5120×2160 is literally 2560×1080 doubled. A "21:9" game and a native render of this panel are the same shape |
| 32:9 | fills the width but letterboxes — 261 mm (10.3") tall, 67% of the height |
| The label "21:9" is marketing, not arithmetic. 21÷9 = 2.333, but no monitor is that shape — the real panels are 2560×1080 (2.370) and 3440×1440 (2.389). Using the literal 21÷9 gives 915 mm and 98%, which is wrong by 14 mm. | |
| Monitor presets — these are LIT AREA, not chassis | |
|---|---|
| 40" ultrawide | 929 × 392 mm (36.6 × 15.4"), 5120×2160, 2.370:1 — Samsung Odyssey G7 G75F. Not 929 × 422. 929 × 421 is the chassis (Tom's Hardware measured it), which is what the CAD box is. The glass is 5120 and 2160 pixels at the 39.7" 5K2K panel's 0.1815 mm pitch = 929.3 × 392.0 mm. The body is still drawn at 929 × 422, so the model envelope is unchanged. |
| 32" 16:9 | 697 × 392 mm (27.5 × 15.4"), 3840×2160 — a "32 inch" panel is really 31.5". LG Display LM315WR-class active area 697.31 × 392.23 mm |
| 49" 32:9 | 1193 × 336 mm (47.0 × 13.2"), 5120×1440 — Samsung LSM490YP02 panel (Odyssey G9 / CRG9 / AOC AG493UCX), published active area 1193.472 × 335.664 mm. Actual diagonal 48.8", not 49" |
| 55" 16:9 | 1210 × 680 mm (47.6 × 26.8"), 3840×2160 — the industry-standard 54.6" panel, 1209.6 × 680.4 mm (Panasonic, Elo, Barco and AUO all publish the same figure) |
| Bezels | Only the 40" has a real chassis size. The other three are drawn as the panel plus a nominal 10 mm bezel per edge — cosmetic, and the only invented number here. The orange outline is always the true lit area. |
It starts at 385 mm off the floor — measured off the actual machine,
not the data sheet, which claims 360 — so you can walk over and drop it straight onto
the real platform. Nudge it up ~25 mm for mid-stroke, ~75 mm for full extension.
Sanity check: 1162 mm wide, 1114 mm tall above the plate. Nothing scales it;
the model is in metres and so is the headset.
The screen is drawn flat. The G75F and the 49" G9 are both 1000R curved in
real life. Flat is what the CAD says and it is the honest way to compare four panels against
each other, but a 1000R panel pulls its far edges roughly 30–50 mm closer to you than
this shows.
No video? You get a test pattern instead — a true 100 mm (3.94")
grid drawn at the panel's real size, so you can still check scale by eye.
The monitor only appears in the full rig. It used to draw in every model,
studies included, because it was made a permanent fixture to stop it vanishing when the model
picker shipped — and a fixture draws everywhere by definition. It is now switched on per
model and off is the default, so a study added later is clean without anyone having to
remember. A second fault went with it: the panel was being rebuilt on every load without the
previous one being thrown away, so after two model switches three monitors were stacked on top
of each other. Both are fixed; nothing about the video, the presets or the aspect guides
changed in the rig.