-
Posts
1740 -
Joined
-
Last visited
Content Type
Profiles
Forums
Events
Everything posted by edmuss
-
solved 10-15% Performance hit with blurred rotors/mast in VR
edmuss posted a topic in Bugs and Problems
In VR there is currently a significant performance loss associated with the blurred effect on the rotor blades/mast when they reach operating speed. I see a repeatable 10fps loss that corresponds directly with the rotor switching to the 3D modelled (no blur) to the blurred appearance. Slowing the rotors so that the appearance changes again restores the performance loss. Attached are two tracks comparing a cold start between BS2 and BS3. The blurred effect is not present on BS2 and does not suffer the performance hit at operating rotor speed. The blurred effect is not present on the Hind and similarly doesn't exhibit the loss, I believe that the blurred effect is present on the Apache so it should be easy enough to correlate the typical effect. As the blurred effect is largely invisible in the cockpit, please can it be made a tickbox option (either special option or general) to disable it? Every frame counts in VR and to lose 10-15% for something that is invisible seems silly when the whole range of high to low graphics settings results in the same sort of performance delta. I will try to get a video capture to demonstrate the issue, other people have witnessed the same as well. I have tried all variations of resolution, graphics options, overclocked/stock GPU, updated GPU drivers, disabled mods & cleared shaders. System is as per signature. I run MT exclusively and admittedly I haven't tested it on ST but I suspect it will behave in the same manner. edit: Have captured the video showing the performance drop, you can clearly see the immediate step in the graph when the rotors blur at around 7 seconds, I then stop the autostart and the graph steps up when the rotors unblur at around 32 seconds. Also note that the FPS jumps up when the rotors are not in view as I look down at the fuel shutoffs and then drops down again as I look up at the cockpit. With the engines off, looking up in the cockpit gives around 85-90 fps so it's not the outside scenery that drops it. BS3_high_speed_rotor_blade_performance.trk BS2_high_speed_rotor_blade_performance.trk BS3 Blurred rotor FPS drop.mp4- 1 reply
-
- 1
-
-
From what I've seen, isn't the tech simply rehashing what has been done by ultraleap for the handtracking and varjo for the eyetracking/foveated rendering & stereoscopic passthrough, tied together with good software? Sure they've polished it into the ecosystem and made it very apple, but the tech has been there for a good while in other places. They missed a trick by not having ultrasonic haptic feedback though If the product survives, competitors flourish and tech trickledown happens then that's a positive though
-
How to repair right throttle buttons on TM Warthog throttle?
edmuss replied to ElCuco68's topic in Input Devices
Chances are you have a broken common wire. To the best of my knowledge the common should be daisy chained around all of the switches, grab a multimeter set to continuity and chase the beep around till you get a consistent connection across all the commons. I think that the POV switches are simply tact switches built into a housing so easy to replace an individual switch if it's failed. -
AMD Ryzen and cooling. Never enough?
edmuss replied to Thinder's topic in PC Hardware and Related Software
I don't have a case*, perhaps that's your problem *It's a motherboard tray screwed to the wall with no top or sides so it's open to ambient. The only fan other than the CPU is a 140mm thermalright pulling cooler air up into the PSU and GPU. -
AMD Ryzen and cooling. Never enough?
edmuss replied to Thinder's topic in PC Hardware and Related Software
But he has a 5800x3D doesn't he? -
AMD Ryzen and cooling. Never enough?
edmuss replied to Thinder's topic in PC Hardware and Related Software
My 5800x3D is on air (thermalright truespirit 140 with a bracket that I hacked together to suit AM4) and peaks at about 74°C in DCS with 30°C room ambient and -25mv PBO2tuner. 10 minutes of cinebench gets it up to 76°C, much above that and mine starts to thermally throttle; if it stays mid 70s then it pegs all cores at 4.45ghz constantly. I have used this heatsink since my i5 2500k and it's been superb! -
Mine arrived yesterday morning, scheduled delivery was next Tuesday! 3 days to get from China to UK so credit to fedex for delivering ahead of the long weekend for me Got it hooked up late last night and generally all as expected, clarity is great and with DFR I can pull 60-80fps typically so about the same performance as with the G2@75% but the image is infinitely better. Obviously with only 90Hz refresh available I'm not able to consistently achieve refresh rate as I could with the G2 at 60Hz, however varjo base vsync is much smoother locked to 45fps than WMR and the associated ghosting is absolutely minimal in comparison. The varjo reprojection is pretty good as long as you hold above half refresh rate, I'm keeping it off for now though. Then it was 2am and I thought I ought to go to bed
-
I can see the difference on very brightly lit scenes but the flicker has zero effect on me, never has done from day 1 in WMR headsets. Caveat, I have always used 60Hz displays, pretty much daily for the last 20+ years so that may have something to do with my acclimatisations; I've never made the jump to a higher refresh rate monitor.
-
Varjo base should run fine with WMR left on there Leave it installed till the time that you no longer have the G2.
-
The aero has buttons on the headset so I believe you can manipulate simple stuff in VR without the need for controllers (setup etc.). My aero is a couple of weeks away yet, I'll be going single base station to start with and upgrading to a second when I get the funds together to buy some index controllers
-
I don't think DCS has functionality for the desk mounted leap motion, I dare say it might be possible but I suspect the likelihood of it happening is slim to none.
- 1 reply
-
- 1
-
-
Query regarding resolution settings in performance toolkit
edmuss replied to markturner1960's topic in Virtual Reality
Specifically with the G2, when running 90Hz with FPS below refresh rate, locking to 55fps seems to give a fractionally smoother image. It's not significant but noticeable. May not translate to other headsets though -
Carried out some tweaks to improve the feel of the lever. Replaced the rubber brake dome with a printed version, this has a far more accurate profile (the rubber dome was machined by spinning a rubber disk in a pistol drill and using a metal file to lathe it to shape) so will give a more consistent gap to the brake plunger. The natural layer steps from the 3D printer gives a good friction surface for the plunger to grip against. Additionally the rubber on rubber interface has a tendency to stick together after a while ago this releases much more freely. Reshaped the lever and printed it off. This version has a bellied gap at the back so that the trim can be pulled with the middle finger whilst leaving enough space for the ring and pinky fingers to get the the NWS button and pinky paddle on the warthog grip. It's also a lot stiffer so that it will pull cable more directly rather than flexing and making the action feel mushy. One further immediate mod to carry out tomorrow will be to move the upper cable barrel adjuster to a tapped hole through the aluminium extension, currently it's fitted to the underside of the lever bracket which is flexing a fair amount against the spring tension. Eliminating the flex will give a much more positive lever feel and will hopefully be the last tweak needed
-
Note that the xr3 is also a subscription headset, if you don't pay the subs for the software you can't use it. This is on top of the initial purchase cost but then it's aimed at enterprise solutions not people pretending to fly planes at home
-
The Acer and HP units are pure VR, not true mixed reality as the varjo xr3 does; they take the MR moniker from the Windows Mixed Reality ecosystem which they work in. The first gen WMR headsets were actually pretty good at the time, I had a Lenovo explorer that was better than the rift S equivalent from oculus. They have been superceded on all fronts by all modern VR headsets though.
-
VR Distortion Effect and Worse FPS with MT
edmuss replied to Highwayman8948's topic in Virtual Reality
If you're using openxr toolkit and reprojection, ensure that movement prediction (I think it's called these days) on the input tab is disabled. In conjunction with reprojection it causes the watery effect you describe. Performance wise it could just be that openxr is running reprojection down to 1/4 refresh rate (22.5fps assuming 90Hz). You don't mention what headset you have but the 3060 is on the weak side for DCS vr unfortunately. -
Finally getting round to updating this! Whilst the original trimmer mechanism worked really quite well, I was finding that the slight drag from the rods and the reasonably heavy spring pull in conjunction with the extension only being 75mm was causing really fine control to be pretty difficult at times, particularly when trying to hover the hind which needs far more tiny cylic inputs compared the to hokum. Time to make a bigger extension I say! Found some 1 1/4" x 1/4" aluminium bar stock and bent it into a nominal gooseneck shape, then jumped into solidworks again and set about redesigning a trimmer that could work with the much larger extension. I had the idea of a single point brake acting upon a dome that was attached to the top of the gimbal and fleshed it out, the brake plunger is tipped with, and the dome is machined from EPDM. The plunger is spring loaded and a single cable pull from the grip lever pulls it off the dome, essentially making it a dead stick that will flop all over whilst the trim lever is held. To try to replicate the stick deflecting from it's trimmed position I designed a translational spring that can move in two axes, this sat between the dome and the gimbal to all the stick some sprung movement when the trim lever isn't being pulled. The total monstrosity looks like this The brake dome, translational spring and spider framework to hold the brake. A really bad video of the brake plunger engaging with the brake dome, also shows the translational spring movement Another video showing operation of the stick and trimmer It's not perfect as there is a little slop in the translational spring, I may print a heavier spring off to see if it stabilises better but as long as you have the stick stationary when you release the trim lever it generally stays where you want (a lighter grip would make this a non-issue but the warthog grip weighs at least a million kg!). I may explore replacing the rubber dome with a printed one as it may give a more positive interface with the plunger and it's also suffering from a bit of wear and I keep getting black rubber crumbs all over the place In addition to being far superior for fine control (I can hover the hind now without much hassle at all) the ability to be able to give cyclic input without having to hold the trim lever is a big bonus, the translational spring is sufficient enough to resist fingertip inputs and gives about +/- 1" of movement at the top of the grip. Total height of the stick now is 28" and it sits on the floor with the grip just above my thighs, I have to reduce the saturation of the axes in DCS to 75% in order to achieve full input at the limits of physical range but that doesn't cause any issues. Because there is only a single spring to pull now, the lever pull tension is halved which makes it easier to give the smaller cyclic inputs. Super happy with it so far, more tweaking to do but it works nicely
-
This has all changed! I now have an 18" extension on the warthog with a new trimmer mechanism that works even better for rotary wing Watch this space, will put an update in soon.
-
MASSIVE ARTEFACTS in VR after upgrade OpenXR/PC
edmuss replied to Dovivan's topic in Virtual Reality
If you have motion prediction enabled on the input tab of openxr toolkit then that can cause the reprojection jelly, check it's set to zero and retest, it might allow you to re-enable NVOF again. -
You need to update openxr tools for wmr from the MSFT app store
-
It's in the works, it will come when Microsoft release the update. All being well it should improve the reprojection artifacts, especially when looking though rotor blades.
-
It's nothing to do with DCS, it's the WMR runtime update that will improve things
-
Doesn't clearing the metashaders and fxo cause a rebuild of all the shaders on the fly which then slows down DCS?
-
Rotor wobble will be gone soon