Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Herra Tohtori]

Due to the release thread becoming larger every day, I'm opening this thread by request to continue discussing the ongoing development of benitomuso's Visual Mod.

Here we can discuss the present and future functions and features of the mod. Things like features suggestions, or major changes in functionality should be here. Let's keep the release thread for the actual releases and possibly bug reports; that should make it easier to actually find the relevant information there - namely, download links, installation help, and troubleshooting.

I would suggest, for clarity, that if you have a feature request or improvement suggestion to make, you would make it with the following format or something close to it:

FEATURE REQUEST - CALL IT SOMETHING SENSIBLE

Problem:

Identify the problem with existing feature, or an aspect of simulation that you want to address with new feature here.

Solution suggestion:

If you have a suggestion as to how the problem could be solved, put it here. It can be anything from a general description of how you would wish the feature to work, or you can go into technicalities if you have the know-how.


I'll start by pasting my earlier post here to kick off the discussion:

_____________________________________________________________________________________

NEW FEATURE REQUEST - 3D IMPLEMENTATION FOR GUNSIGHT VIEW

I had a brilliant new feature request idea that I simply must bother you with.

The 3D view is awesome, but there are two situations where it doesn't work quite correctly or seamlessly with the gameplay experience:


1. Gunner position aiming

 Problem:

For gunner positions, the 3D view causes a small problem. Instead of having the view perfectly "centre-lined" with the gunsight, you end up with two view points, both sides of the gunsight line. This is especially problematic with iron sighted weapons, as you don't really get any sights at all but have to rather aim "along the barrel" instead. For turrets with reflector sights, this is not as much of an issue; however with sufficient eye separation values, you end up with getting both eyes not seeing the centre of the gunsight reticle.

Solution suggestion:

The feature I would like to suggest for the gunner positions, and in fact it could be useful for the Shift+F1 "gunsight view" cockpit mode as well:

-When view is in gunner cockpit or Shift+F1 mode, instead of placing the cameras symmetrically exactly the separation values beside the gunsight line, place the right eye ON the gunsight line (with separation value 0) and the left eye would be positioned two times its separation value to the left!

So, for example: If you're flying with internal separation value of 6.5 centimetres, then normally your eyes would be at distance (-6.5 cm) for left eye and (6.5 cm) for right eye, measured from the centreline. This adjustment would put the RIGHT eye on the distance (0 cm) from centre line (gunsight line), and left eye would be moved (-13 cm) from the centreline.*

This would, in effect, make it seem like the pilot is actually using their right eye to peer through the gunsight, or to line up with the iron sights, if such are applicable on the position. Of course, it would be possible to add a [MOD] section switch to select your dominant eye: If your left eye is dominant, the system could switch to using left eye on the gunsight line and right eye would be displaced 2x separation value to the right instead. That's something personal, and considering it's quite similar to being left- or right-handed, I think it would be remiss to force everyone to use their right eye in these situations.

The OTHER solution would be to actually allow 6DOF movement in gunner cockpits, which would allow one to MANUALLY shift the viewpoint so that their dominant eye (the one they prefer to aim with) is lined with the gunsights. But I'm not sure if this is doable.



2. Optical Instrumentation

Problem:

Telescope gunsights and bombsights that you are supposed to operate with one eye alone, instead of like binoculars where you can use both eyes.

When peering through optical instruments, like for example the gunsights in early Ki-43, Ki-27, SBD-3... or using any other optical instrumentation like the periscope sight in Arado Ar-234, dive bombing sight, Norden bombsight... in all of these instruments, you are only using one eye, so the 3D mode should be disabled for these situations.

Solution suggestion:

Currently, 3D mode stays active with all these devices, which basically makes you see two gunsights or bomb sights and you have to sort of aim somewhere in the middle. This is rather distracting and, as said, when you peer through these things with one eye alone, you should not get any 3D at all! So, it would be ideal to simply have the 3D system automatically disabled when using any optical instrumentation.

However, I'm not sure if this is feasibly done; considering the controls to peer through gunsight or bomb sight is the same Shift+F1 as moving the gunsight mode, I don't know if the optical modes have any sort of identifier that could help the mod to recognize it as a situation where it should switch 3D off. But, it is worth looking at regardless, as the bombsight 3D mode feels very distracting and unreal.





FEATURE REQUEST - FOLLOW CAM FUNCTIONALITY


Problem:

What's happening with the Follow Camera camera mode is that when the TrackIR is constantly trying to enforce a new camera position on the external view, the inertial system doesn't get time to properly function. This causes the image to "jump" from one position to another, as the inertial follow cam system has moved the camera position slightly, and then TrackIR immediately transposes it again to a new position and the inertial mode tries to kick in from that position. This is what causes the discontinuity in the camera view. I noticed the same thing happens when you are moving the mouse slowly while maneuvering; the mouse is shifting the camera reference point, and inertial view is shifting the camera around the reference point. When the reference point is constantly on the move - whether by mouse or TrackIR, the inertial mode causes problems.

Solution suggestion:

Disabling TrackIR for F2 and F8 camera modes would also probably be better at this time.

With mouse it's probably not nearly as apparent as it's easier to have the mouse stay put where you placed it, but it definitely gets disruptive with the TrackIR.


_____________________________________________________



Please tell me what you think about these two feature ideas and if they're feasible to execute!

These suggestions are strictly based on my experiences flying with the mod both online and offline. I don't see any flaws in them but I'm biased by nature; if anyone has any opinions or spots a fatal flaw that makes these a really Bad Idea, go ahead and say it. I came up with these trying to think of how to make the 3D mod feel more real in different situations, to figure out how the pilots would act in different situations... gunners would line their dominant eye along the gunsights, for example. And the bombsight problem is obvious when you think about it, yes?


Unfortunately I don't have time at the moment to try the new version - I need to get to school - but I will definitely give it a spin in the evening!



*Yes, I use 6.5 as both internal and external separation value. It may sound very high, but with the convergence values I can avoid having everything come right into my face from the screen, and since about 13 cm is the average distance between human eyes, this gives me a rather nicely realistic sense of depth and scale. It also has added benefit of being actually able to see past most cockpit canopy struts, as well as lining the right eye with the gunsight in many fighters with offset gunsight (Bf-109, FW-190, Ki-84, P-47 Razorback variants, and many others)...

However this feature would work for any other separation values just as well!


Interestingly, I have also noticed a bug with the gunsights in the game: Depending on your head position, they don't always point to the "forward" direction! There can be quite a bit of difference in the gunsight vector even when you just check with the horizontal eye separation. This basically causes you to see the gunsight as a "duplicate" when you focus into the distance... it's perhaps most apparent in the Spitfires. This can actually cause you to miss targets quite a bit...

This is nothing related to the 3D mod at all, it's a retail game bug, but it just becomes most apparent with the 3D mod active since you see the gunsight from two camera positions, effectively.

I should make a thread about this issue to find out if it can be fixed.

[benitomuso]

Herra,
                 regarding your comments:

  1-Gunsights: it could be done what you suggest. Just a setting saying PredominantEye=R or L could lead to asymetry in gunsights view, and an empty value would mean symetry (as today). Consider that done.
  2-Optical Instrumentation: what you say is true in reality, but I don't know if you have experienced the MOD in the bombsight for example (if you have it properly adjusted). It really gives you a huge sense of spatiality. It is true, optical instruments of the sort, only produced monoscopic views, but we in the game don't see through a small hole, we see with both eyes to a big screen, so the situation is not the same. But for sure it is impossible to use the separation and the convergence angles of the other views, so previously to your comment I had started to make a third level of Stereo Adjustment, so we will have separate settings for External / Cockpit / Aims . Any of the tubed sights, with a small stereo separation and an almost null convergence settings, perform beautifully (check them). And those views are quite the same of what you see in a mono-microscope, where there is a "phantom distance feeling".
3-OK, I will try to avoid TrackIR movements. The problem is that they come exactly the same way as reset view and other commands (even the mouse).

  Regards,
                          Pablo

[Herra Tohtori]

Herra,
                 regarding your comments:

  1-Gunsights: it could be done what you suggest. Just a setting saying PredominantEye=R or L could lead to asymetry in gunsights view, and an empty value would mean symetry (as today). Consider that done.

Excellent!

  2-Optical Instrumentation: what you say is true in reality, but I don't know if you have experienced the MOD in the bombsight for example (if you have it properly adjusted). It really gives you a huge sense of spatiality. It is true, optical instruments of the sort, only produced monoscopic views, but we in the game don't see through a small hole, we see with both eyes to a big screen, so the situation is not the same. But for sure it is impossible to use the separation and the convergence angles of the other views, so previously to your comment I had started to make a third level of Stereo Adjustment, so we will have separate settings for External / Cockpit / Aims . Any of the tubed sights, with a small stereo separation and an almost null convergence settings, perform beautifully (check them). And those views are quite the same of what you see in a mono-microscope, where there is a "phantom distance feeling".

This is even better than my suggestion, as now this will be easily customizeable by everyone.

3-OK, I will try to avoid TrackIR movements. The problem is that they come exactly the same way as reset view and other commands (even the mouse).

  Regards,
                          Pablo

I was afraid of that.

The problem seems to be rooted on the fact that the inertial mode does not actually move the camera reference point, but instead moves the camera around a point determined by TrackIR or Mouse-guided camera controls. The problem, then, happens when you have both inertial adjustments AND mouse/TrackIR camera view movements happening at the same time, where the mouse/TrackIR changes the reference point and inertial view tries to adjust the camera location around that reference point. This duality of different camera movements might be what's causing the problem.

Hmm. The problem of stuttering view seems to happen because the inertial mode resets for every tiny camera movement, bringing the camera back to the reference point and then applying the inertial view adjustments to camera location, but as soon as it does this, there's already a new camera location adjustment from the TrackIR which polls the position constantly.

Perhaps it would be possible to make the inertial view adjustment hold as the camera is moved to new location?

Here's how I imagine it's working now:


-Camera position is set to point P1
->inertial system applies translation XYZ and rotation R to camera position and vector, shifting the camera away from point P1
-Camera position is set to point P2 by TrackIR interface - not taking into account the distance and direction change that the inertial system has applied
->inertial system applies translation XYZ and rotation R to camera position and vector, shifting the camera away from point P2 again

...and this goes on ad infinitum.

Here's what I suggest:

-Camera position is set to point P1
->inertial system applies translation XYZ and rotation R to camera position and vector, shifting the camera away from point P1
->the changes made by the inertial system are kept in program memory

-Camera position is set to point P2 by TrackIR interface
->previous frame's inertial adjustment values for translation (XYZ) and rotation (R) are added to the position and direction vector given by the TrackIR system
->inertial system applies translation XYZ and rotation R to camera position and vector, shifting the camera away from point P2 again
-the changes made by the inertial system are kept in program memory.

Some kind of system like this could remove the difference between two subsequent frames, which I believe is what causes the flickering.

Does this sound sane? ???

[santobr]

It would be good if the trackir was ignored at external views, the mouse with some inertia would be the best for external views.



santobr.

[PhoenixOrion]

It would be good if the trackir was ignored at external views, the mouse with some inertia would be the best for external views.

I agree.

santobr.