[1.12.x] Parallax - PBR Terrain and Surface Objects [2.0.8]

[Sheppard]

I can confirm a huuuuuge performance improvement without Planet Shine. The game has become playable again.

[HafCoJoe]

Hm, this has me wondering now which creates less lag by itself—PlanetShine, or Parallax...

Edited July 28, 2022 by HafCoJoe

[Benzo Kerman]

23 minutes ago, HafCoJoe said:

Hm, this has me wondering now which creates less lag by itself—PlanetShine, or Parallax...

I believe by itself, planetshine causes less lag. But that may just be me.

[HafCoJoe]

I did a quick FPS comparison test between Parallax and PlanetShine with my full mod suite to compare the FPS hit between them both, and with them combined. All readings taken straight from bootup with QuickStart from the same location on Duna exactly 3 minutes into flight.

PC Specs:

GeForce RTX 3070 Ti
AMD Ryzen 7 5800X
32GB DDR4 3200 CL16-18-18-38

Parallax: ~99-106fps

PlanetShine: ~108fps

Parallax + PlanetShine: ~103-108fps

Parallax: ~99-106fps
PlanetShine: ~108fps
Parallax + PlanetShine: ~103-108fps
 

Curiously, Parallax + PlanetShine seems even faster than just Parallax? Strange. Either way, the FPS hit seems negligible for me—this could be due to having such a powerful machine though. Also worth noting that when I removed Parallax, it lowered the terrain height by about 70m and I had to manually fly the craft down to the surface again.

Edited July 29, 2022 by HafCoJoe

[Benzo Kerman]

2 hours ago, HafCoJoe said:

Curiously, Parallax + PlanetShine seems even faster than just Parallax? Strange.

Try to test this out on a non-atmospheric body. I don't think planetshine applies a light source while in atmosphere, I think it just adjusts the ambient light level. I noticed the biggest difference on the mun. Specifically while landing from orbit. Again, this could just be me.

[HafCoJoe]

5 hours ago, Benzo Kerman said:

Try to test this out on a non-atmospheric body. I don't think planetshine applies a light source while in atmosphere, I think it just adjusts the ambient light level. I noticed the biggest difference on the mun. Specifically while landing from orbit. Again, this could just be me.

Gotcha, will do when I get home tonight 

[Vandest]

9 hours ago, Benzo Kerman said:

Try to test this out on a non-atmospheric body. I don't think planetshine applies a light source while in atmosphere, I think it just adjusts the ambient light level. I noticed the biggest difference on the mun. Specifically while landing from orbit. Again, this could just be me.

Compare the first and last pic, and you can see that reddish light from PlanetShine is visible on ship. (same thing if you are on Kerbin, you will see the bottom of your ship is lit by blue light witch come from ground)

PlanetShine create one light when you are close enough from the celestial body which is supposed to send back a light (on the side lit by the sun). For example, when you are on Kerbin, you receive a blue light from the ground and receive no light from Mun. Conversely, when you are on Mun, you receive a grey light from Mun ground and no blue light from Kerbin.

Also, when you are in orbit, you can see this is not an ambient light only, because you can see on your ship that there is shadow area between area lit by celestial body and area lit by the sun.
When you are on ground it seems that is a ambient light but I'm not sure.

Edited July 29, 2022 by Vandest

[HafCoJoe]

I did a second test, this time on Vall:

Stock: ~84-89fps

Parallax: ~86-90fps

PlanetShine: ~81-86fps

Parallax + PlanetShine: ~80-83fps

Stock: ~84-89fps
Parallax: ~86-90fps
PlanetShine: ~81-86fps
Parallax + PlanetShine: ~80-83fps

Still pretty negligible. And it seems that Parallax runs better than PlanetShine? Again these tests are taken by a very basic set of statistics; rebooting the game with each variation, waiting until the timer is the same, take reading. It's worth noting too that the Vall-henge monolith is underground with Parallax; it seems to mess with the terrain height

Edited August 1, 2022 by HafCoJoe

[Moose]

On 7/30/2022 at 6:18 AM, HafCoJoe said:

Still pretty negligible. And it seems that Parallax runs better than PlanetShine?

PlanetShine is old. Very old. When it comes to KSP modding anyway.

The first commits were made in 2015 and KSP in itself has had a couple major engine upgrades since then. A lot of the options we have now for optimization didn't really exist back then. Ideally @Papa_Joe could find the time to optimize it/change it to an ambient light approach but that's beyond scope I feel.

[Papa_Joe]

5 minutes ago, Moose said:

PlanetShine is old. Very old. When it comes to KSP modding anyway.

The first commits were made in 2015 and KSP in itself has had a couple major engine upgrades since then. A lot of the options we have now for optimization didn't really exist back then. Ideally @Papa_Joe could find the time to optimize it/change it to an ambient light approach but that's beyond scope I feel.

Ya a lot has changed since then.  I've also learned a lot about unity since then.  Don't know if I can invest the time, so no promises.

[HafCoJoe]

I realize now that I should have done a stock comparison too, to judge how much baseline performance impact both mods have. Hard to judge how well/poorly optimized either one is without knowing

Edited July 31, 2022 by HafCoJoe

[Othy]

On 7/17/2022 at 1:51 PM, Gameslinx said:

I've been messing around with Duna and decided to go full Mars with it. I'm really happy with how the rocks have turned out - the dust scattered across the top really adds to the environment and helps them blend in

 

This looks awesome so far! Good luck on figuring out the new collisions for these, they made rover driving a lot more immersive when I played with them in the older version

Edited July 31, 2022 by Othy

[Brigadier]

6 hours ago, Papa_Joe said:

Don't know if I can invest the time, so no promises.

None expected.  You do what you feel like doing and we'll be the beneficiaries.

[HafCoJoe]

I did a new performance test, this time with the same visual mods but with stock instead of Parallax or Planetshine. Stock results also added to the post above.

Stock: ~84-89fps
Parallax: ~86-90fps
PlanetShine: ~81-86fps
Parallax + PlanetShine: ~80-83fps

All in all the results are close. Real close. In fact there's no real discernable FPS difference between stock and parallax; they both spike to about as high, and as low as each other. PlanetShine seems to have an impact of around -4fps, and Planetshine + Parallax is about -10fps? Again though these readings are VERY subjective. Literally just one reading, exactly 6:30 into flight. Still I'd call that pretty dang good.

Edited August 1, 2022 by HafCoJoe

[Gameslinx]

Colliders

And here's a video of a Kerbal traversing Tylo's rocky ground

 

[t_v]

Just now, Gameslinx said:

Colliders

And here's a video of a Kerbal traversing Tylo's rocky ground

 

Oh my goodness, I'm going to need a Tylo Rescue Rover Rescue Rover now...

[OJT]

4 hours ago, Gameslinx said:

Colliders

And here's a video of a Kerbal traversing Tylo's rocky ground

 

 

On 7/17/2022 at 7:51 PM, Gameslinx said:

I've been messing around with Duna and decided to go full Mars with it. I'm really happy with how the rocks have turned out - the dust scattered across the top really adds to the environment and helps them blend in

 

 

On 7/11/2022 at 12:58 AM, Gameslinx said:

I've finished Gilly and Eeloo's scatters. Only 4 more celestial bodies to go!

I also added a Fresnel effect to the terrain which is noticeable especially on Eeloo. It makes the ice look shinier and behaves more realistically. I can't believe this wasn't included for Parallax's first release, it's a really nice effect

Finally, a nice picture I took on Kerbin

 

 

I've been following your latest developments so far. All of this looks mad impressive. However I do have some questions about all this

First: My stock planets save is without Parallax. I do have Parallax on my Beyond Home save and I did notice that object spawn and colliders are not always consistent, going from trees spawning in the air and ending with some rocks having way bigger colliders than their visible size would suggest (to the point that I actually lost a plane on Hydrus because it apparently collided with a rock that was nowhere near my plane). I was wondering how consistent the colliders are in these newer versions of Parallax. Or the problems I've encountered have mostly to do with different solar system and in stock solar system it works fine?

Second: I understand that these new Parallax features will make the surfaces of planetary bodies much more closer to reality. However, I also suspect that new colliders will also make landing planes much harder. Big part of what makes KSP fun for me is designing and flying planes that are adapted for different gravitational and atmospheric conditions and, even though making landable plane for a spot that is littered with rocks is an interesting challenge in itself, I fear that new colliders might take some pure "gameplay" fun out of it, if you catch my meaning. I am curious if the collider feature will be adjustable, if it will be biome dependent (say, Lowland plains having less rocks than Midlands) of if there will be an outright different version of new Parallax without any colliders whatsoever?

All in all, fantastic stuff mate. Keep it up! 

[SpaceCube2000]

5 hours ago, Gameslinx said:

Colliders

And here's a video of a Kerbal traversing Tylo's rocky ground

 

Great work! Will the colliders impact performance? If so, how much? 

[Gameslinx]

13 minutes ago, OJT said:

I was wondering how consistent the colliders are in these newer versions of Parallax

Parallax aims to build upon or improve the systems Squad implemented into the game. The current terrain scatter system is pretty barebones and, in combination with Kopernicus' mishandling of it for so long, it leads to unreliable placement. So, this new scatter system was born . To answer your question there are two types of scatters: Persistent and Non-Persistent. Persistent scatters are generally rocks, trees, and other large-scale objects. Non persistent scatters come in the form of grass or objects that appear very frequently and require a greater level of precision in terms of placement. Only persistent scatters have colliders, and they really are persistent. You can shut down the game and load it back up and they will still be there in exactly the same place, even if you were landed on top of one your craft will be fine.

17 minutes ago, OJT said:

I am curious if the collider feature will be adjustable, if it will be biome dependent (say, Lowland plains having less rocks than Midlands) of if there will be an outright different version of new Parallax without any colliders whatsoever

I want the gameplay to be positively impacted. People getting frustrated and feeling like they're too limited by the rock placements isn't what I want, so I'm going to be experimenting with which scatters have colliders and which ones do not. Small rocks won't be collideable - only the medium to larger ones. The colliders will be completely optional and off by default. You can also use the in-game GUI to toggle colliders on or off for a particular scatter, or turn them off permenantly for a scatter in the configs

7 minutes ago, SpaceCube2000 said:

Will the colliders impact performance? If so, how much? 

In this test, performance was not noticeably impacted. However, I have some optimization to do so I can't really say at the moment  

[OJT]

12 minutes ago, Gameslinx said:

Parallax aims to build upon or improve the systems Squad implemented into the game. The current terrain scatter system is pretty barebones and, in combination with Kopernicus' mishandling of it for so long, it leads to unreliable placement. So, this new scatter system was born . To answer your question there are two types of scatters: Persistent and Non-Persistent. Persistent scatters are generally rocks, trees, and other large-scale objects. Non persistent scatters come in the form of grass or objects that appear very frequently and require a greater level of precision in terms of placement. Only persistent scatters have colliders, and they really are persistent. You can shut down the game and load it back up and they will still be there in exactly the same place, even if you were landed on top of one your craft will be fine.

My question was more in the direction of "Will that rock's collider be the same size as its visual model/texture consistently?". Because I was occasionally having some issues with exactly that in Beyond Home

[Gameslinx]

Just now, OJT said:

My question was more in the direction of "Will that rock's collider be the same size as its visual model/texture consistently?". Because I was occasionally having some issues with exactly that in Beyond Home

Ah! Yes, they will. I pull the exact same transform matrix that the visual scatter uses from the GPU and use it to create the collision version CPU sided. There is no way the scatter or the mesh can ever become out of sync or inconsistent 

 

..or so my testing shows so far  

[t_v]

1 hour ago, Gameslinx said:

In this test, performance was not noticeably impacted. However, I have some optimization to do so I can't really say at the moment  

Wait wait wait, does the update perform comparably to previous parallax? I was ready to take a 25% hit to my fps, but are you saying that I'll be getting the same amount?

[HafCoJoe]

Holy Kerb does this mean I'll actually have to pick my landing sites now?! 

I'm curious how it performs—it'll be interesting to compare the fps hit with current release

Edited August 1, 2022 by HafCoJoe

[Gameslinx]

2 hours ago, t_v said:

Wait wait wait, does the update perform comparably to previous parallax? I was ready to take a 25% hit to my fps, but are you saying that I'll be getting the same amount?

My system is getting the same FPS (maybe a tad lower) than the current Parallax version. The scatters are really not that difficult to render at all thanks to GPU instancing

I am also going to be making optimizations to the current parallax version as well as fixing some noticeable visual bugs, including EVE city light support. I've not started on those optimizations yet, but I have a good idea for what I will be doing

Edited August 2, 2022 by Gameslinx

[HafCoJoe]

19 minutes ago, Gameslinx said:

My system is getting the same FPS (maybe a tad lower) than the current Parallax version. The scatters are really not that difficult to render at all thanks to GPU instancing

I am also going to be making optimizations to the current parallax version as well as fixing some noticeable visual bugs, including EVE city light support. I've not started on those optimizations yet, but I have a good idea for what I will be doing

Another bug I noticed is terrain height changes with Parallax—is it possible to fix that or just a consequence we'll have to accept? It only bothers me because it makes the easter eggs spawn underground 

Edited August 2, 2022 by HafCoJoe