Donington Park Grand Prix Circuit 1938

Tracks Donington Park Grand Prix Circuit 1938 1.0

Login or Register an account to download this content
I get some dropped frames in the wood area, but it's pretty much limited to these two places.

2017-09-21 22_11_05-Donington Park Grand Prix Circuit 1938 _ RaceDepartment.jpg


2017-09-21 22_14_17-Donington Park Grand Prix Circuit 1938 _ RaceDepartment.jpg


I use Vsync though and unlimited it's 70-80fps usually, though it might be worse in other places too, the two marked ones are were I notice them, frametime otherwise is rock-solid in the render stats app.
 
I get some dropped frames in the wood area, but it's pretty much limited to these two places.

View attachment 212767

View attachment 212768

I use Vsync though and unlimited it's 70-80fps usually, though it might be worse in other places too, the two marked ones are were I notice them, frametime otherwise is rock-solid in the render stats app.
Yeah I think it is a combination of things between the 3D people, trees, and grass. But if I remove the trees I get much more stable frames. Grass seems a bit dense especially since it is using X objects not rectangle 2D objects. The tree issue might be too many shadows. But as I said I don't think it is one single thing. As with Okayama it was a combination of multiple little problems. My machine will drop as low as 49 and with most tracks, including my own, rarely go below 70.
 
Yeah I think it is a combination of things between the 3D people, trees, and grass. But if I remove the trees I get much more stable frames. Grass seems a bit dense especially since it is using X objects not rectangle 2D objects. The tree issue might be too many shadows. But as I said I don't think it is one single thing. As with Okayama it was a combination of multiple little problems. My machine will drop as low as 49 and with most tracks, including my own, rarely go below 70.

I get similar frame rates on this track and yours. Im running a 1060 gfx with a ryzen 1700. 16 core. Its all i have to test on
 
force feedback on my side is acting very weird, and your method for producing the physics layer which ive taken the liberty of looking at, its not good practice to make it like that, needs consistency over the poly spread, the patches of subdivided areas imo not nice man. but a nice track visually for sure !
 
force feedback on my side is acting very weird, and your method for producing the physics layer which ive taken the liberty of looking at, its not good practice to make it like that, needs consistency over the poly spread, the patches of subdivided areas imo not nice man. but a nice track visually for sure !

Please elaborate. Doesnt feel at all horrible to me. Irs bumpy for sure intentionally. What whell are you useing. To help me understand
 
force feedback on my side is acting very weird, and your method for producing the physics layer which ive taken the liberty of looking at, its not good practice to make it like that, needs consistency over the poly spread, the patches of subdivided areas imo not nice man. but a nice track visually for sure !
What's the physics layer?
I thought physics would only concern cars, sort of.
 
Though if you run with lower world detail. The crowds will be omitted
The crowd isn't the problem. I gain only a frame or two with them out. I think the biggest problem is the trees and grass. For one the grass needs a much more aggressive LOD system. Here is a paste from a conversation I had with Mitja about Okayama and his grass.

"Number per section I don't really care about. It's the length of the section that concerns me.

I do something like this over the grass. In this case each section is about 500' long.

grass4.JPG


I then cut out sections of the grass according to the pattern.

grass3.JPG


The key thing is that the origin or pivot point must be at the center of all the new separate grass objects.

For Bridgehampton I have 5 layers of grass all cut into the same sized pieces. Here you can see all 5 layers selected for one section. Notice how the origin points are all within close proximity of each other.

grass5.JPG


Some base LOD #s for 5 layers is:

KSLAYER1 = LOD_OUT=350
KSLAYER2 = LOD_OUT=300
KSLAYER3 = LOD_OUT=260
KSLAYER4 = LOD_OUT=160
KSLAYER5 = LOD_OUT=80

For a 3 layers system like Riverside and some Kunos tracks:
KSLAYER3 = LOD_OUT=350
KSLAYER4 = LOD_OUT=170
KSLAYER5 = LOD_OUT=80

A 5 layer system has less noticeable pop-in or 'crawl' than a 3 layer system.

Here is a final overall shot of Bridgehampton grass. You can see the origin points of all the objects.
grass6.JPG


As for LOD#s for other objects I don't really bother with them much. Only with the grass have I found the most FPS gain. In either case if I do set LOD #s for an object I just look for blatant pop-in and adjust the numbers to avoid it."

As for the trees I don't really know what is going on. I have a feeling they are just too dense and the amount of shadows the sim is trying to render is too much. It sounds like you have a decent rig so I can see you not having issues. They thing to pay attention to is when you might have a 5 FPS swing on your rig it might result in a 10+ FPS swing on a lower end machine. Going into the first tree section I see 22+ FPS drop between the first bridge and the end of the woods section which is huge and that is without the grass and 3d people being on. Also a good practice is anything under the trees should not be trying to make shadows. This would be the bushes and small trees that are under there.
 
The crowd isn't the problem. I gain only a frame or two with them out. I think the biggest problem is the trees and grass. For one the grass needs a much more aggressive LOD system. Here is a paste from a conversation I had with Mitja about Okayama and his grass.

"Number per section I don't really care about. It's the length of the section that concerns me.

I do something like this over the grass. In this case each section is about 500' long.

View attachment 212783

I then cut out sections of the grass according to the pattern.

View attachment 212784

The key thing is that the origin or pivot point must be at the center of all the new separate grass objects.

For Bridgehampton I have 5 layers of grass all cut into the same sized pieces. Here you can see all 5 layers selected for one section. Notice how the origin points are all within close proximity of each other.

View attachment 212785

Some base LOD #s for 5 layers is:

KSLAYER1 = LOD_OUT=350
KSLAYER2 = LOD_OUT=300
KSLAYER3 = LOD_OUT=260
KSLAYER4 = LOD_OUT=160
KSLAYER5 = LOD_OUT=80

For a 3 layers system like Riverside and some Kunos tracks:
KSLAYER3 = LOD_OUT=350
KSLAYER4 = LOD_OUT=170
KSLAYER5 = LOD_OUT=80

A 5 layer system has less noticeable pop-in or 'crawl' than a 3 layer system.

Here is a final overall shot of Bridgehampton grass. You can see the origin points of all the objects.
View attachment 212786

As for LOD#s for other objects I don't really bother with them much. Only with the grass have I found the most FPS gain. In either case if I do set LOD #s for an object I just look for blatant pop-in and adjust the numbers to avoid it."

As for the trees I don't really know what is going on. I have a feeling they are just too dense and the amount of shadows the sim is trying to render is too much. It sounds like you have a decent rig so I can see you not having issues. They thing to pay attention to is when you might have a 5 FPS swing on your rig it might result in a 10+ FPS swing on a lower end machine. Going into the first tree section I see 22+ FPS drop between the first bridge and the end of the woods section which is huge and that is without the grass and 3d people being on. Also a good practice is anything under the trees should not be trying to make shadows. This would be the bushes and small trees that are under there.
The crowd isn't the problem. I gain only a frame or two with them out. I think the biggest problem is the trees and grass. For one the grass needs a much more aggressive LOD system. Here is a paste from a conversation I had with Mitja about Okayama and his grass.

"Number per section I don't really care about. It's the length of the section that concerns me.

I do something like this over the grass. In this case each section is about 500' long.

View attachment 212783

I then cut out sections of the grass according to the pattern.

View attachment 212784

The key thing is that the origin or pivot point must be at the center of all the new separate grass objects.

For Bridgehampton I have 5 layers of grass all cut into the same sized pieces. Here you can see all 5 layers selected for one section. Notice how the origin points are all within close proximity of each other.

View attachment 212785

Some base LOD #s for 5 layers is:

KSLAYER1 = LOD_OUT=350
KSLAYER2 = LOD_OUT=300
KSLAYER3 = LOD_OUT=260
KSLAYER4 = LOD_OUT=160
KSLAYER5 = LOD_OUT=80

For a 3 layers system like Riverside and some Kunos tracks:
KSLAYER3 = LOD_OUT=350
KSLAYER4 = LOD_OUT=170
KSLAYER5 = LOD_OUT=80

A 5 layer system has less noticeable pop-in or 'crawl' than a 3 layer system.

Here is a final overall shot of Bridgehampton grass. You can see the origin points of all the objects.
View attachment 212786

As for LOD#s for other objects I don't really bother with them much. Only with the grass have I found the most FPS gain. In either case if I do set LOD #s for an object I just look for blatant pop-in and adjust the numbers to avoid it."

As for the trees I don't really know what is going on. I have a feeling they are just too dense and the amount of shadows the sim is trying to render is too much. It sounds like you have a decent rig so I can see you not having issues. They thing to pay attention to is when you might have a 5 FPS swing on your rig it might result in a 10+ FPS swing on a lower end machine. Going into the first tree section I see 22+ FPS drop between the first bridge and the end of the woods section which is huge and that is without the grass and 3d people being on. Also a good practice is anything under the trees should not be trying to make shadows. This would be the bushes and small trees that are under there.
Ok thanks man. All good info. Looks like this may take an age.i will do my best. Did you drive it though. Its fun no!
 
Back
Top