Odyssey VR specific bug reports - please contribute!

...
A previous report regarding smoke sprites was also closed with a similar message. Annoyingly, several years ago during a Beta period the cockpit smoke sprites had their twist rotation locked so they only rotated horizontally and vertically, and it looked much better - so we know it can be done. Possibly too much work for it to be justified?
Yeah, this definitely needs to be pushed anew; Maybe with a reframing from "bug report", to "design re-evaluation request", or even "-proposition", if there is somebody among us with the know-how, and ability to formulate a comprehensive suggestion in the proper vernacular and format.

It is not just the Z axis rotation that is the problem, but the two others as well; You'll notice that if you look up/down/left/right, the sprites turn then, too, and it looks not a whole less strange than the roll, so the question that needs to be posed is: "Should these effects be summarily screen-aligned, really?".

Normally, I would set up a sprite object so that it tracks the camera position, and always strives to face it, and on implementation actively choose which axes to match rotation around, and their local coordinate system, so that they stay put in relevant ways in world space, all depending on what sort of object it is (e.g. a glow effect around a beam of light should roll around the spine of the beam, trying to face the camera, but not break free from it with yaw and pitch rotation); But what the ones in Elite: Dangerous does, is that they only track the camera rotation, 1:1 -- not the position, so they are indeed always parallel with the screen, which looks more and more strange the farther out toward the periphery they get; You can really tell that all those steam vent plumes around you is an array of cardboard cutouts, all facing the exact same way on the compass... And that straying anamorphic lens flare? -Judging by the direction it moves when looking up/down, I'd have to presume its billboard, relative to its centre of rotation, is a fair ways behind the star -- this wouldn't have been a problem if it lined up with your position - it would always be on the extension of the line between you and the star... Even the markers on the galaxy map face over your shoulder, as you traverse away sideways, and then over the shoulder of the tenth person next to you...

A mass ejection from a star should stay aligned with the star it belongs to, and not roll around together with your looking around (...or flying around - it's not like the problem does not exist outside of VR); I find it gobsmacking that this needs to pointed out.

One complication I could envision, is if the effect texturing is not applied only with the sprite object UVs, but directly by the screen-space shader -- that could leave you with a projection problem to solve...

It's not like the engine couldn't do it, as indicated by your mentioned cockpit smoke example; Heck... take e.g. the bit in the new video the other day, where the the Thargoids get a splash of water in their faces... That 2D effect is not aligned with the screen, but with the object (possibly being part of it, rather than a separate sprite).

It's always hard to align things, when dealing with 2D surfaces in 3D space, so that the 2D doesn't stand out... Fly anywhere near close to the Proteus Wave site anomaly (depends a bit on the lighting conditions - the bloom and refraction effects hide a lot), and you can see the mesh "shells" of the "pool of light" effect separate them by their parallax. :7
 
Last edited:
New issue report (listed as a Horizons problem): https://issues.frontierstore.net/issue-detail/51888 - VR rendering problem - looks like a more than capable system so I’m not sure what is causing it.
———
@jojon - perhaps a solution would be to lock the sprite movement to whatever vehicle you are in? It would avoid most of the ugly motions we see in VR due to head movement while still stopping us from seeing sprites from edge-on or other angles. A VR compatible flight sim I’ve played did this with their sprite-based clouds, it was only when flying through the clouds that you could notice the sprites start to turn around your plane, though this was mostly disguised by a whispy-whooshing effect layered over the clouds as you passed through them, which I thought worked really well. Much more performance friendly than volumetric clouds too.
 
New issue report (listed as a Horizons problem): https://issues.frontierstore.net/issue-detail/51888 - VR rendering problem - looks like a more than capable system so I’m not sure what is causing it.
———
@jojon - perhaps a solution would be to lock the sprite movement to whatever vehicle you are in? It would avoid most of the ugly motions we see in VR due to head movement while still stopping us from seeing sprites from edge-on or other angles. A VR compatible flight sim I’ve played did this with their sprite-based clouds, it was only when flying through the clouds that you could notice the sprites start to turn around your plane, though this was mostly disguised by a whispy-whooshing effect layered over the clouds as you passed through them, which I thought worked really well. Much more performance friendly than volumetric clouds too.
I see the reporter/user has an SLI rig - SLI used to play very nicely with ED back in the day and I used to run my Rift CV1 with it, but not since circa 2017/2018 iirc - all kinds of rendering weirdness used to happen...
 
I see the reporter/user has an SLI rig - SLI used to play very nicely with ED back in the day and I used to run my Rift CV1 with it, but not since circa 2017/2018 iirc - all kinds of rendering weirdness used to happen...
Nicely spotted - I hadn’t registered the SLI bit after their graphic card 👍

I’ve added a comment on the report in the hope that it may help them.
 
@jojon - perhaps a solution would be to lock the sprite movement to whatever vehicle you are in? It would avoid most of the ugly motions we see in VR due to head movement while still stopping us from seeing sprites from edge-on or other angles.
Can't say I see how that would help, even after reading your account of that other game; To my mind that should only exacerbate the problem in some ways -- you wouldn't even be able to get a nearby sprite to face you (...which is the whole point), by turning to look straight at it.

At first glance it would bring us to exactly where non-VR players are, where the sprites stay put relative to headlook, but roll together with the ship (...which is not where we want to be, either), but only as long as you never look-, or move around.

One thing others have done, is to add a slight "inertia" to the sprite's turning, which masks the noticeable immediate response to head motion (there is even a developer commentary bit about it in Half Life: Alyx, which describes this), but that is not at all about the same issue... :7
 
Can't say I see how that would help, even after reading your account of that other game; To my mind that should only exacerbate the problem in some ways -- you wouldn't even be able to get a nearby sprite to face you (...which is the whole point), by turning to look straight at it.

At first glance it would bring us to exactly where non-VR players are, where the sprites stay put relative to headlook, but roll together with the ship (...which is not where we want to be, either), but only as long as you never look-, or move around.

One thing others have done, is to add a slight "inertia" to the sprite's turning, which masks the noticeable immediate response to head motion (there is even a developer commentary bit about it in Half Life: Alyx, which describes this), but that is not at all about the same issue... :7
I’m probably not explaining it very well 😅

At the moment when we move our heads in Elite, the sprites move with our HMD motions - be it massive stellar ejecta or the wisps of smoke around landing pads - and it looks fake and awful. If the horizontal and vertical facing of the sprites was locked to our vehicle instead, they wouldn’t move when we’re looking around. There shouldn’t be any roll movement of the sprites at all. Only when flying or driving close to these sprites would we be able to notice them turning to keep facing our vehicle, which might be masked by our vehicle’s structure or some graphical effect as per the flight sim I mentioned.

I’d like to show the clouds in the game I’ve mentioned* but I’m away from my computer for a few more days yet. Thanks for the HL:A commentary heads-up, I’ve been meaning to replay it and those info snippets sound like something I should listen to. Perhaps after the HL2 VR mod comes out next month 👍

*Edit: this old video (not mine) shows what I’m on about - the sprite based clouds:
Source: https://youtu.be/70I2WcIp6IM?t=288
 
Last edited:
I believe this new case is a cmdr I tried helping on Reddit. Surprised they opened a bug report and not a support case since it seems an issue to his particular ED setup and not widely experienced.

To me, if they're not getting a VR mode prompt when launching from Steam, that's an issue right there that needs to be addressed before going any further. They're probably seeing two flat screens and not VR at all. They did confirm all other VR games are working fine so it's not their rig.

I'm not sure they'll get much help from the opened ticket. Let me try to find the Reddit thread and send the support ticket link again. Unless, of course, FDev instructed them to raise this new case.

Edit: NVM, FDev Support had already responded to their post.
 
Last edited:
...If the horizontal and vertical facing of the sprites was locked to our vehicle instead, they wouldn’t move when we’re looking around. There shouldn’t be any roll movement of the sprites at all. Only when flying or driving close to these sprites would we be able to notice them turning to keep facing our vehicle, which might be masked by our vehicle’s structure or some graphical effect as per the flight sim I mentioned...
I am afraid I am still too slow to understand the precise manner in which the sprite is locked to the vehicle here...

If (if !) it is "absolute", there would be no roll when looking around - no, but I can't see how there wouldn't be when rolling and turning the vehicle, no matter how far away...

I find it hard to tell anything from that particular video, because our grumpy pilot is not only mostly still as a statue, but his manouvering is all very gentle, which is understandable, of course: It is a plane, not a spaceship, having a "down" to consider, so no entirely arbitrary rolling.

Hard to tell, as I said: Maaaybe the cloud particles roll when the plane rolls? ...but I could imagine it more likely that they could possibly be fixed in world space on the X, and Z axis, so that they always stand upright in world space, and only swivel on Y, to either face the airplane, or clone its yaw in world space.

Transposed to Elite:
For the dashboard smoke: Perfect! This smoke needs to have its emitter orientation fixed to the ship, since it is what it is part of, and here the particle system should rotate together with the ship, although each particle's normal still needs to face the camera, and nothing else, even with its "up" being toward the top of the canopy (animation-imposed transformations excepted).

For objects outside the ship, other than the ship's hull itself, however: Not so much -- If our star protuberance was a big pointy arrow instead of a flung mass of searing hot plasma, it needs to always point away from the star, even when turning the ship in any direction.
 
Last edited:
I am afraid I am still too slow to understand the precise manner in which the sprite is locked to the vehicle here...

If (if !) it is "absolute", there would be no roll when looking around - no, but I can't see how there wouldn't be when rolling and turning the vehicle, no matter how far away...

I find it hard to tell anything from that particular video, because our grumpy pilot is not only mostly still as a statue, but his manouvering is all very gentle, which is understandable, of course: It is a plane, not a spaceship, having a "down" to consider, so no entirely arbitrary rolling.

Hard to tell, as I said: Maaaybe the cloud particles roll when the plane rolls? ...but I could imagine it more likely that they could possibly be fixed in world space on the X, and Z axis, so that they always stand upright in world space, and only swivel on Y, to either face the airplane, or clone its yaw in world space.

Transposed to Elite:
For the dashboard smoke: Perfect! This smoke needs to have its emitter orientation fixed to the ship, since it is what it is part of, and here the particle system should rotate together with the ship, although each particle's normal still needs to face the camera, and nothing else, even with its "up" being toward the top of the canopy, animation-imposed transformations excepted.

For objects outside the ship, other than the ship's hull itself, however: Not so much -- If our star protuberance was a big pointy arrow instead of a flung mass of searing hot plasma, it needs to always point away from the star, even when turning the ship in any direction.
The sprite clouds stay fixed - aligned to the ground. As the plane passes them you can sometimes see the edge of the cloud rapidly move as the sprite keeps its face to the plane as it passes.

If we had a similar system in Elite then stellar ejecta would be aligned to its parent star, geyser smoke to its planet surface, smoke to its landing platform etc. We should, hopefully, only see the sprite move when passing it. The pitch and yaw of our vehicles would still cause the sprite to move but hopefully it would be imperceptible, especially when compared to how we can currently move them with our heads.
 
This is after 15 minutes in the same game mode, and has only manifested in the last week. It gets even worse if game mode changes, such as landing at a station, landing on a planet etc.
I don’t have the Steam version of Elite and cannot check my game until later this coming week, but I will see if I can duplicate the issue. Prior to the latest Update the only performance drops I got in Odyssey were due to https://issues.frontierstore.net/issue-detail/36142
 
...aligned to the ground...
Ok, confusion solved.

Not that I don't still think it is naturally the camera that needs to be matched, mind - not the vehicle, but I see your argument.

It is just that the sprites need to be matched to the camera in the correct manner, and oriented relative to their parent object, as appropriate on a case-by-case basis.

I really wish that somebody technically adept, and well spoken, among us, could have a one-on-one with the relevant lead at Frontier.

It is not like there is any possibility the same people who have the chops to make the game, could be unaware of the problem, or its potential solutions, but somewhere priorities have to be set, I have to concede, with a game that still has occasionally game-loop-breaking issues... :7
 
Last edited:
Is it ok to put the errors I notice with Ody in VR in this thread or is it too old ?
I check the thread and Issue Tracker daily and update as necessary - if you have issue reports then I’ll happily add them to the OP 👍

Please check the current issues in the OP and perhaps the reports at the bottom of the Expired spoiler list - they might save you a bit of typing 😁
 
I check the thread and Issue Tracker daily and update as necessary - if you have issue reports then I’ll happily add them to the OP 👍

Please check the current issues in the OP and perhaps the reports at the bottom of the Expired spoiler list - they might save you a bit of typing 😁
Ok will do thanks.
 
White Enviroment when looking at Codex :-


Kamadhenu()_00001.png


Kamadhenu()_00002.png
 
Back
Top Bottom