I’m Anthony, new Pimax 5K+ user. Nice to know everybody here!
I have noticed something in various games, which feels pretty strange to my eyes while ingame. Looks like I don’t get the same textures/reflections/rendering on both eyes. Well, a picture is worth a thousand words, so here we go:
( This is a photoshop image to explain what’s happening, not a screenshot )
This is exactly what’s happening on my pimax on various games ( noticed in Elite Dangerous and KartKraft at the moment). Well, this is not really an issue, but it looks strange sometimes and breaks the immersion.
Maybe I have a different issue but eventually goes in the same direction, yesterday I played Obduction for the first time and I noticed differences between the both displays by showing other textures/surfaces/ shadows. I dunno if this is Pimax or game software related, anyway I thought I share this information here. AC f.eg. etc does not show any issues. Can you try out a few other games for testing purposes?
Hey, thanks for your answer. I’ve seen this problem with shadows also. It’s pretty eye catching, as your brain immediately knows there’s something wrong with the 3D view. I will do further tests, but as far as I went, it only appeared in both games mentioned above. I believe this is more software than hardware related, but that would be nice to see what’s faulty.
I had the same issue until I disabled the Ambiant Occlusion option in Elite graphical settings. For some reasons it doesnt render the same result on both eyes.
It is in fact normal to have different images for your eyes; the picture you show features a specular highlight, which is highly angle dependent, and the angle to your two eyes differs. You can fairly easily test this example in real life; take a small shiny object such as a knife blade and look at something reasonably bright in its reflection. You’ll find that the reflection is in different places, so it easily appears for only one eye.
That said, some rendering tricks like screen space ambient occlusion depend on other factors, like the field of view rendered, and can produce incorrect result particularly on Pimax headsets as the two views have wider regions that don’t overlap. Other examples include simulated god-rays (e.g. from sunlight through a window) or reflections. You may have to tweak rendering on some titles to avoid these artifacts producing binocular conflict.
Totaly agree, indeed seeing different images on both eyes is what creates 3D, but I don’t know, here it looked a bit weird. Thanks a lot for your messages guys.
It looks weird because it is weird. SSAO estimates the light in a scene by processing the image shown (including Z-buffer you normally don’t see), and that image is widely different for Pimax, as each eye sees a large region the other does not. The effect is that things “pop in” to the lighting model at different angles for the two eyes. It’s worse than the usual pop-in effects because the difference is large and continuously present.
The solution to still apply something like SSAO is to take the image for both eyes into account while calculating the effect. I haven’t seen this done, but it might be interesting. Let’s call that stereoscopic SSAO. It would depend not only on both rendered views but the transformation between them (which is stored in the view matrix).
Yes this is a Parallell Projection issue. I spotted and reported this in a bunch of games already back in the Pimax m1 beta testing phase. Some shadows and reflections renders differently for each eye, most likely because the games were made for two non-canted panels. Unless the game/sim developers fix this for VR headsets with canted displays (Pimax, Valve Index, XTAL etc) then I dont think we will see this fixed. At least Pimax cannot do anything about it (as far as I understand) because the shadow/reflection angles are calculated in the game, not taking your both eyes viewpoint and angle in account.
Look at DCS World developers. They completely removed the need of Parallell Projection. Lets just hope more game developers do the same…
I think it’s more that games assuming parallel projections are more likely to make other assumptions, such as roughly 100% binocular overlap. Doing screen space effects will have this type of issue whether or not the views are parallel, but if the views are not parallel it makes no sense to assume total overlap. In general, it’s easier to not implement the parallel rendering assumption, but we’ve seen it embedded in API levels a few too many times. Basically, the VR API should get you a suitable projection matrix, not something you need to feed into gluPerspective.
