Spotting dot bugs in VR

[Goofy12]

It looks absolutely awful in VR, its basically a massive square box. If you have a formation far away in front of you it looks like Santa and his reindeer are flying past. Unfortunately spoils the immersion for me personally completely. The autoexec.cfg workaround doesnt work btw.

[YoYo]

10 minutes ago, Goofy12 said:

It looks absolutely awful in VR, its basically a massive square box. If you have a formation far away in front of you it looks like Santa and his reindeer are flying past. Unfortunately spoils the immersion for me personally completely. The autoexec.cfg workaround doesnt work btw.

Welcome in DCS: Minecraft DLC !

[TAIPAN_]

On 10/21/2023 at 4:03 AM, BIGNEWY said:

Hi the team are looking at the feedback and we will be tweaking for future updates. 

thank you

Please don't reduce dots for high res though (Varjo Aero and Pimax Crystal), using DFR eye tracking supersampling already shrinks and fades them they are hard to see. 

Options menu would be ideal. 

[MoleUK]

3 hours ago, TAIPAN_ said:

Please don't reduce dots for high res though (Varjo Aero and Pimax Crystal), using DFR eye tracking supersampling already shrinks and fades them they are hard to see. 

Options menu would be ideal. 

Agreed there. Moving towards a slider or list of various pre-sets we could choose from would be ideal.

Some will be more bothered by bigger dots, some by dots that are too small. So some flexibility would be good for users across various different resolutions/headsets and between flatscreen and VR.

Edited October 22, 2023 by MoleUK

[YoYo]

37 minutes ago, MoleUK said:

Agreed there. Moving towards a slider or list of various pre-sets we could choose from would be ideal.

Some will be more bothered by bigger dots, some by dots that are too small. So some flexibility would be good for users across various different resolutions/headsets and between flatscreen and VR.

 

But it is very important that it can also be blocked from the online game position (settings of server, not user only). Otherwise, some will have a large square, others will not be able to see the targets. 

It's not that simple. I have quite a high resolution myself, set to about 3200x3200 pxt per eye and I certainly think that the square is currently too large as well. Another thing is that as the object approaches - its last LOD is loaded, so the square disappears and the target becomes invisible. The transition between the point and the last LOD is as important as the rest . This requires a deeper analysis, not just "reduce the dot" btw.

[Nedum]

Hi,

 

it would be nice if we all would post our DCS settings. Without that, it will be hard to generate a picture of how "well" the new spotting system works for each of us.

Personally, I am a Fan of it. I am playing only in VR and with a resolution of 4312x5102 pixel.

As YoYo already said, the transition from the Dot to the model seems a bit off. There is a distance at which the Dot disappears, and the object is invisible.

And at some view angles (peripheral view) I can see a Dot, which I can't see if I am looking directly in the direction where the object/dot is.

Can't say if that is only a VR thing.

nullnull

DxDiag.txt

[MoleUK]

8 hours ago, YoYo said:

But it is very important that it can also be blocked from the online game position (settings of server, not user only). Otherwise, some will have a large square, others will not be able to see the targets. 

It's not that simple. I have quite a high resolution myself, set to about 3200x3200 pxt per eye and I certainly think that the square is currently too large as well. Another thing is that as the object approaches - its last LOD is loaded, so the square disappears and the target becomes invisible. The transition between the point and the last LOD is as important as the rest . This requires a deeper analysis, not just "reduce the dot" btw.

It will require some time/tweaking from ED for sure.

I would very much prefer that it NOT be controlled on the server side however. Just give us the options to work from and go from there, as there are too many hardware combinations for a one size fits all approach.

If worried about some players using it to seek an advantage: That is literally how it has been for years already. They played at 1080p to get that advantage.

Edited October 22, 2023 by MoleUK

[Brother_Leb]

On 10/20/2023 at 5:33 PM, MoleUK said:

 

 

Thanks for posting @MoleUK These two videos are good examples of the pop in/out depending on viewing angle, which seems to me to be a slightly different issue than dot size but definitely needs addressing.

As for dot size, I play on a potato PC (by DCS standards at least, see my signature if ur curious) and the dots are massive until about 3NM out and then, poof! I'd love to have a better PC and run crazy settings, but that's not going to be a reality for me in the near future. Whatever solution/adjustment is made, please don't discount us low-budget/low-end folks. We're not down here because we want to be. Thanks!

[MoleUK]

52 minutes ago, Brother_Leb said:

Thanks for posting @MoleUK These two videos are good examples of the pop in/out depending on viewing angle, which seems to me to be a slightly different issue than dot size but definitely needs addressing.

As for dot size, I play on a potato PC (by DCS standards at least, see my signature if ur curious) and the dots are massive until about 3NM out and then, poof! I'd love to have a better PC and run crazy settings, but that's not going to be a reality for me in the near future. Whatever solution/adjustment is made, please don't discount us low-budget/low-end folks. We're not down here because we want to be. Thanks!

This problem alongside potentially some LOD issues has been suggested as the root cause.

Edited October 22, 2023 by MoleUK

[Why485]

1 hour ago, MoleUK said:

This problem alongside potentially some LOD issues has been suggested as the root cause.

What it looks like to me is that for whatever reason the aircraft models are getting culled too closely to the camera. When in 2D I haven't seen this issue of the aircraft disappearing when I zoom out, but something about VR is causing the distance at which planes are drawn to be much shorter than they would be in 2D.

Edited October 23, 2023 by Why485

[YoYo]

What is good its on the list of known issues: 

so fingers crossed for solution!

[MoleUK]

7 hours ago, Why485 said:

What it looks like to me is that for whatever reason the aircraft models are getting culled too closely to the camera. When in 2D I haven't seen this issue of the aircraft disappearing when I zoom out, but something about VR is causing the distance at which planes are drawn to be much shorter than they would be in 2D.

 

To further complicate matters, lots of VR users use techniques that render a slightly higher resolution image towards the centre of the headset and slightly lower res towards the edges.

Not sure if that could be playing a part as well. But the zooming in creating a sudden appearance of big dots would suggest something else I guess.

And i'm also not sure if all headsets are affected equally or if it varies from one to another.

Edited October 23, 2023 by MoleUK

[TAIPAN_]

21 hours ago, YoYo said:

 

It's not that simple. I have quite a high resolution myself, set to about 3200x3200 pxt per eye and I certainly think that the square is currently too large as well. 

That's actually quite low resolution compared to this. Crystal at 1.0 quality is 5100x4350 per eye then we run 200% supersampling on the DFR quadviews portion which pushes the effective resolution up even higher. 

If the dot is same as the mod, there's a cap on dot size that probably means my dot is the same number of pixels as yours (initially), and is then reduced down from the supersampling. So it starts at 2x3 pixels or something similar then when it gets mapped to the 2880x2880 panel it loses size and becomes hard to see. 

Currently we have to turn quadviews off to even see the dot, and disabling quadviews tanks FPS by 30-40%. Ideally for quadviews we need it even bigger to make up for the deficit. 

Server side option I totally disagree - that would lock people into running low resolution all over again which is the legacy we are trying to avoid. The whole point of the 2.9 change is to level the playing field, a server option only makes it favour certain setups like it used to in 2.8 before the mod. 

Edited October 23, 2023 by TAIPAN_

[YoYo]

2 minutes ago, TAIPAN_ said:

That's actually quite low resolution compared to this. Crystal at 1.0 quality is 5100x4800 per eye then we run 200% supersampling on the DFR quadviews portion which pushes the effective resolution up even higher. 

If the dot is same as the mod, there's a cap on dot size that probably means my dot is the same number of pixels as yours (initially), and is then reduced down from the supersampling. So it starts at 2x3 pixels or something similar then when it gets mapped to the 2880x2880 panel it loses size and becomes hard to see. 

Currently we have to turn quadviews off to even see the dot, and disabling quadviews tanks FPS by 30-40%. Ideally for quadviews we need it even bigger to make up for the deficit. 

Server side option I totally disagree - that would lock people into running low resolution all over again which is the legacy we are trying to avoid. The whole point of the 2.9 change is to level the playing field, a server option only makes it favour certain setups like it used to in 2.8 before the mod. 

 

I prefer to have full spectrum of FPS than the qulity (I dont use any ASW/MR and have stable 72fps), natively Crystal has 2880 x 2880 pxt per eye so I dont complain ;).

[TAIPAN_]

5 minutes ago, YoYo said:

I prefer to have full spectrum of FPS than the qulity (I dont use any ASW/MR and have stable 72fps), natively Crystal has 2880 x 2880 pxt per eye so I dont complain ;).

Exactly same here, that's why I want to run quadviews because it gives me stable 90fps even with high res. Quadviews is a miracle technology. 

Unfortunately it shrinks & fades the dots. 

[BIGNEWY]

Hi all, 

we continue to look at the feedback and the team are working on tweaks. With so many combinations of resolutions and settings out there it is a difficult task to get it right for everyone in every situation but we will try. 

It seems (especially in VR) with different resolutions with the combination of upscaling the larger dots are more obvious and it will need to be tweaked. 

thank you

[MoleUK]

6 minutes ago, BIGNEWY said:

Hi all, 

we continue to look at the feedback and the team are working on tweaks. With so many combinations of resolutions and settings out there it is a difficult task to get it right for everyone in every situation but we will try. 

It seems (especially in VR) with different resolutions with the combination of upscaling the larger dots are more obvious and it will need to be tweaked. 

thank you

Yep, a one size fits all approach might be extremely tricky given the variation in headsets, upscaling and sharpening.

No idea how doable sliders would be, with caps at the higher end to prevent too much cheesing but allowing those with undersized dots to increase them a bit, and those with overscaled dots to tone it down manually.

Regardless it's already been a net positive overall to say the least imo.

Edited October 23, 2023 by MoleUK

[Roostyla]

I'd just like to add that the issue of looking directly at aircraft and them being invisible , then looking to the side in peripheral vision and they pop in happens on 2D as well. I'm having to look out the side of my eyes to maintain visual sometimes. Within 2 miles or so. Longer range the dots are fine.

[YoYo]

43 minutes ago, TAIPAN_ said:

Quadviews is a miracle technology.

So I have to try this finally . Btw. do you use DLSS in the same time?

[YoYo]

42 minutes ago, BIGNEWY said:

Hi all, 

we continue to look at the feedback and the team are working on tweaks. With so many combinations of resolutions and settings out there it is a difficult task to get it right for everyone in every situation but we will try. 

It seems (especially in VR) with different resolutions with the combination of upscaling the larger dots are more obvious and it will need to be tweaked. 

thank you

Thanks! A good option would probably be a slider, but with the option of locking it. Everyone could then adapt it to themselves. I think that a good solution would also be two separate settings, for objects/targets on the ground and for objects in the air. Also, the spotting dot itself should not be larger than the lowest LOD, because now it is so that when the target gets closer and the farthest LOD of the object is loaded, it becomes invisible, while a second earlier it was a larger box The effect is that it gives the impression that it is moving away from us rather than getting closer.

[MoleUK]

48 minutes ago, Roostyla said:

I'd just like to add that the issue of looking directly at aircraft and them being invisible , then looking to the side in peripheral vision and they pop in happens on 2D as well. I'm having to look out the side of my eyes to maintain visual sometimes. Within 2 miles or so. Longer range the dots are fine.

Ah, interesting.

[Why485]

3 hours ago, YoYo said:

Also, the spotting dot itself should not be larger than the lowest LOD

If this were true, then spotting dots would do literally nothing. Keep in mind that the dots don't replace the 3D models. The lowest detail LOD is always being rendered out to the max object rendering distance (~50 miles, but this range is adjusted in real time up and down based on FOV). 

The problem they are trying to solve in the first place is that aircraft are difficult to impossible to see at distances which according to various research papers and studies should be visible to the pilots. As I've said many, many times in the past decade on this subject. The main problem that should be being solved here is making WVR visibility match the data. I'm less interested in "distant spotting" (which is what a dot adjacent approaches are best suited for) than I am in making WVR actually mean visual range. This is especially important to WWII, Korea, and early Cold War era fighter jets which can't rely on radar to do spotting for them.

There is also a difference between spotting and identifying, that I think is glossed over in all these conversations. "Spotting" is simply the earliest you know something is out there. It can, in the best of cases, mean something like being able to see the glints from other fighter sized aircraft tens of miles away under certain lighting conditions. E.g. Chuck Yeager talking about seeing fighters in Korea from 30 miles away. This is the situation where dot-like solutions can be best, since you're mainly interested in specks of shadow or light at large distances. "Identifying" is when you are within visual range (~5 miles, but depends on the aircraft and conditions) and close enough to tell what the target is, and more importantly, what its aspect is. This is where solutions like smart scaling work best, and despite what so many people seem to insist, this can be done subtly and well.

Edited October 23, 2023 by Why485

[YoYo]

1 hour ago, Why485 said:

If this were true, then spotting dots would do literally nothing. Keep in mind that the dots don't replace the 3D models. The lowest detail LOD is always being rendered out to the max object rendering distance (~50 miles, but this range is adjusted in real time up and down based on FOV). 

The problem they are trying to solve in the first place is that aircraft are difficult to impossible to see at distances which according to various research papers and studies should be visible to the pilots. As I've said many, many times in the past decade on this subject. The main problem that should be being solved here is making WVR visibility match the data. I'm less interested in "distant spotting" (which is what a dot adjacent approaches are best suited for) than I am in making WVR actually mean visual range. This is especially important to WWII, Korea, and early Cold War era fighter jets which can't rely on radar to do spotting for them.

There is also a difference between spotting and identifying, that I think is glossed over in all these conversations. "Spotting" is simply the earliest you know something is out there. It can, in the best of cases, mean something like being able to see the glints from other fighter sized aircraft tens of miles away under certain lighting conditions. E.g. Chuck Yeager talking about seeing fighters in Korea from 30 miles away. This is the situation where dot-like solutions can be best, since you're mainly interested in specks of shadow or light at large distances. "Identifying" is when you are within visual range (~5 miles, but depends on the aircraft and conditions) and close enough to tell what the target is, and more importantly, what its aspect is. This is where solutions like smart scaling work best, and despite what so many people seem to insist, this can be done subtly and well.

 

Currently, you have such a situation that the trace, i.e. the visible square (spotting dot), is larger than the lowest (poorest) LOD of model, which proves the funny situation that after loading the LOD of the model, the trace, simply disappears, and practically nothing is visible in this place. As for spotting and identification, currently I think it's good (I write about 2.8), in fact I've never had a problem knowing who I'm fighting with within a few nm, but it's not visible from a distance, even with the zoom turned on in VR. 

It is certainly easier to spot a plane in real life according me (unless its against the ground), I think that in good weather I can see for at least several dozen miles. It also happens that you can't see the plane, but your eyes are drawn to reflections from the sun (btw. it would be nice to have the same thing in DCS as in IL-2 Blitz by Team Fusion). This means that you do not see the silhouette of the plane, but you do see a flare and flashes of the windshield or glossy paint, e.g.

[Why485]

24 minutes ago, YoYo said:

Currently, you have such a situation that the trace, i.e. the visible square (spotting dot), is larger than the lowest (poorest) LOD of model, which proves the funny situation that after loading the LOD of the model, the trace, simply disappears, and practically nothing is visible in this place.

The trace, as you call it, must be larger than 3D model of the plane or else it would essentially do nothing at any range because it would be same size as the aircraft. I.e. if the plane is impossible to see at 5 miles without the trace, it'll be impossible to see at 5 miles with the trace.

The problem you're talking about is how the trace simply disappears completely and abruptly at close range, which I agree is a problem. I have some thoughts on how I would fix this if it were me (I tried to fix this in the mod but it wasn't possible without source), but I think it's more important to just say that the aircraft abruptly going going from highly visible to almost invisible feels bad.

I don't want to get too in the weeds either on "how to fix dots" because that's too limiting for what we should be trying to achieve. I'd rather frame the discussion as "fixing aircraft visibility" instead, since that doesn't preclude smarter and more robust systems than drawing 4 pixel squares.

I think ED understands this based on how Wags specifically phrased the known issue: 

• Units are visible in VR at unrealistically far ranges and are less visible at closer ranges.

Edited October 23, 2023 by Why485

[YoYo]

1 hour ago, Why485 said:

The trace, as you call it, must be larger than 3D model of the plane or else it would essentially do nothing at any range because it would be same size as the aircraft. I.e. if the plane is impossible to see at 5 miles without the trace, it'll be impossible to see at 5 miles with the trace.

The problem you're talking about is how the trace simply disappears completely and abruptly at close range, which I agree is a problem. I have some thoughts on how I would fix this if it were me (I tried to fix this in the mod but it wasn't possible without source), but I think it's more important to just say that the aircraft abruptly going going from highly visible to almost invisible feels bad.

I don't want to get too in the weeds either on "how to fix dots" because that's too limiting for what we should be trying to achieve. I'd rather frame the discussion as "fixing aircraft visibility" instead, since that doesn't preclude smarter and more robust systems than drawing 4 pixel squares.

I think ED understands this based on how Wags specifically phrased the known issue: 

• Units are visible in VR at unrealistically far ranges and are less visible at closer ranges.

 

What you write is exactly the same as what I write, only from the other side of the stick. The most important thing here is the effect that is in the middle of the road.

"Units are visible in VR at unrealistically far ranges and are less visible at closer ranges." - 100% of the true.