While fixing up underwater, it seemed a bit strange to see things near the horizon at all. Time for a bit of optics, no? The water-air interface should display total internal reflection over the critical angle of 48.6 degrees. It should look water-y. A little experimentation later, and I think it's a good start.
Gallery : http://www.tesnexus.com/downloads/images/26348-4-1280866639.jpg http://www.tesnexus.com/downloads/images/26348-3-1280866639.jpg http://www.tesnexus.com/downloads/images/26348-2-1280866639.jpg http://www.tesnexus.com/downloads/images/26348-1-1280866638.jpg
Meanwhile I'm trying to optimize grass rendering (it doesn't like being optimized), and do something about better interaction with shadow and fog. I would like to see some experimentation and commentary on the issues asked for in my previous posts.
That looks great! I had mentioned way back when Liztail was working on MGE v3.8.0 (when he was incorporating depth-based water effects) about possibly emulating how light refraction behaves from underwater similar to what you have accomplished here. Another behavior that would be terrific to see is how the deeper you go underwater, the smaller that above surface view portal gets.
Also, a much darker ambient light level underwater during nighttime is needed - should be at least the same lighting levels below as above water.
I've been testing build v0.62 a bit, but have been distracted with other projects - I hope to have some more detailed feedback for you soon, including fps benchmarks. At this point, I can confirm what others have reported regarding animated grass stutter and shadows still visible during sundown (albeit faint as if cast from moonlight). The difference in framerate between shadows enabled and disabled is about 18-20%; I still need to do more comprehensive comparison tests against rev-0163+.
A few preliminary observations:
*
I've noticed that there is shadow pop-in while moving which appears to be due to distant statics entering/leaving MW's render distance - this is more easily viewed when traveling thru the vivec cantons.[edit] I double checked this, and haven't been able to reproduce it with v0.62 - I'm pretty sure it was happening with v0.6 tho. Upon further examination, it probably had to do more with the static min size and end distances and those distant statics falling out of range while moving. With v0.62 shadows are fading out as the statics get out of range.
* Once in a while there's also a noticeable shadow wall on the peripheries when rotating the camera view in 1st person - standing still and rotating makes it easier to notice. This usually occurs near cell borders.
[edit] tested some more, and this shadow wall appears to be the projected shadow getting clipped by MW's view range limit. It's similar to the levitating issue below.
* Shadows sometimes pull away from the ground when levitating, as if the calculated position for where the ground plane should be is getting set to the player's z-world position.
[edit] further testing reveals that the projected shadows are getting clipped at the MW view range limit, it's more noticeable when looking almost directly down. Here are some screenshots to demonstrate:
http://www.liquidanvil.com/mw/screenshots/mgexe/mgexe062_levshadowglitch_01.jpg - bottom right of picture 1st tier should be in shadow.
http://www.liquidanvil.com/mw/screenshots/mgexe/mgexe062_levshadowglitch_02.jpg - bottom right of picture 1st tier shadow has mostly disappeared and bottom of Arena canton shadow at Telvanni canton waterline has partially disappeared.
http://www.liquidanvil.com/mw/screenshots/mgexe/mgexe062_levshadowglitch_03.jpg - 1st tier shadows have almost completely disappeared.
All these screenshots were taken at the same time of day, ~4pm
* In someplaces, particularly the north-east corner of the Redoran canton near the 1st tier bridge to FQ, the shadowed ground suddenly brightened within what appeared to be some of the canton model's triangles. I had to be viewing the ground at a specific angle; time of day was afternoon.
[edit] this seems to occur more frequently on flat surfaces. Here are some sample screenshots showing the issue:
http://www.liquidanvil.com/mw/screenshots/mgexe/mgexe062_staticshadowglitch_01.jpg - shadows partially missing.
http://www.liquidanvil.com/mw/screenshots/mgexe/mgexe062_staticshadowglitch_02.jpg - another angle, shadows partially missing.
http://www.liquidanvil.com/mw/screenshots/mgexe/mgexe062_staticshadow_on.jpg - same spot, looking at a slightly higher angle, shadows rendered correctly.
http://www.liquidanvil.com/mw/screenshots/mgexe/mgexe062_staticshadow_off.jpg - same spot and angle with shadows disabled for comparison.
All these screenshots were taken at the same time of day, ~4pm
* Screen repositioning in windowed mode equal to the desktop resolution will cause the game load progress meters to get partially cut-off by the bottom of the windowed frame since the frame doesn't completely fill the screen. The offset appears to be whatever the size was of the window title bar before it's removed.
Some MGEgui settings used during testing:
* view distance of 5 cells
* fog range 0.5 - 5.0
* 'Near', 'Far', and 'Very Far' static end distances of 2, 3.5, and 5 respectively
* Fog mode is set to 'Range vertex'
* primarily 1152x864 windowed mode, other 4:3 ratios as well when testing the screen repositioning feature
* 1280x960 fullscreen mode
Testing with nvidia geforce 8800GT SC 512MB; driver version: 195.62