Tracks Sveg, Sweden

mhh thanks, but I think it would be more interesting to get some lower spec'd systems to report performance. After all, our 1070s won't differ that much.
Fully agree! Just wanted to make that offer for whenever needed. Would only differ from different settings in AC and not be worth the effort to sending it over etc :p
 
In researching the big fps loss due to tree walls I did some investigations.
So far I can report that a large object (same texture, same mesh, just scaled up) is segnificantly more taxing than a smaller one. I was able to generate a 2-3 fps drop with a single object (8 polygons, 1 texture) when I scaled it up really large.

Another hunch that I wasn't able to prove yet is that an object that covers the sky is more taxing than one that doesn't.

Waiting for LilSki to report his findings... ;-)
 
In researching the big fps loss due to tree walls I did some investigations.
So far I can report that a large object (same texture, same mesh, just scaled up) is segnificantly more taxing than a smaller one. I was able to generate a 2-3 fps drop with a single object (8 polygons, 1 texture) when I scaled it up really large.

Another hunch that I wasn't able to prove yet is that an object that covers the sky is more taxing than one that doesn't.

Waiting for LilSki to report his findings... ;-)

Do you think it's the scale of the object only, or does increasing the size allow the engine/gpu to select a higher resolution mip?

Would be interesting to know if doubling the size of the object but also reducing the resolution of the texture would give parity.
 
Do you think it's the scale of the object only, or does increasing the size allow the engine/gpu to select a higher resolution mip?

Would be interesting to know if doubling the size of the object but also reducing the resolution of the texture would give parity.

I don't know, but large objects in general seem to have a large cost in terms of fps, larger than one would typically expect from a simple object with a simple texture.

I'd like to invite everybody to help investigate this. I personally have spent so many hours optimizing now, I just have to move on with the actual completion of the track in order to make it in time for the release.
 
Just to give you guys an update: The physical meshes are almost done. All the walls are in place, and I have quite a sophisticated 6-layer system for the bumps in the road in place. Maybe some final tweaks today and tomorrow, but I'll soon get back to completing the visuals in a rough first version. After that I can finally work on replay cameras and making things a little prettier.

Can't wait to have all the annoying stuff done and get back to the visuals =)
 
Good news everyone,

three people whos opinions I value highly approve of the road surface: LilSki, Nick Moxley and myself :)
I think it's actually quite exciting with a realistic distribution of bumps and irregularities.

In other words, road surface is final for 1.0.

I will probably widen the circle of beta testers in about two weeks time.
 
It's AI line & camera weekend. I've done both before, but it's been awhile. Can you guys help me out a little?

1) Cameras: I remember cameras being fairly easy and all done in the editor. I think I can figure it out. Note to self, here's some advanced camera stuff: http://www.assettocorsa.net/forum/index.php?threads/track-cameras-a-tutorial.30931/#post-644403

2) AI line can be complicated. I've done this before, but I'm not sure when to hit the rocord AI button (and others), how long to drive, etc. - links please? :)

3) I found this thread about track limits: http://www.assettocorsa.net/forum/i...define-track-limits-for-ai.45913/#post-891839 Can somebody tell me how to get the coordinates of a path from blender into a csv?
 
Hi Leon,

first make sure to enable the developer apps in AC, so that you can choose the AI app on the right menu when in the car. Also before recording a line switch IS_VALID_TRACK=1 to 0 for the PIT surface (or whatever it is called for the track) in the surfaces.ini. When in the car click "Start recording" in the AI app and drive until you think you have driven a good, clean lap. It's better to drive moderate and precise (runder Fahrstil auf deutsch) than superfast and hectical. If you are happy stop the car and save the AI in the AI app (bottom right button). You now have a fast_lane_candidate in the ai folder of your track. Rename it to fast_lane in order for it to work. Now you'll probably want also a pit_lane. The procedure is similar with the difference that you do not need to drive a whole lap. Just get to the point where you want the pit_lane to start - normally several hundred meters before the pit entry - and click "Record pit lane" (or similiar). Drive through the pits, leave them and come to a stop shortly after the pit exit on your fast_lane. Save your recorded line in the AI app, leave the sim and rename pit_lane_candidate to pit_lane. You should now have a working AI. You can now switch IS_VALID_TRACK=0 to 1 again (for PIT surface) in the surfaces.ini and save.

To take it a step further you might want to add ai_hints in the data folder of your track.
 
You start recording ai just after the start line and it automatically finishes when you have crossed it again. The small gap between where you have started and the finish line gets connected automatically by magic.
I would consider the AI-line helper app too
 
its probably easier to record those cvs files ingame with this (http://www.racedepartment.com/threads/ai-line-helper.137572/)
Also because you get smoother realistic borders this way that will result in smoother AI driving.

mh. driving along the outline seems really imprecise. I can easily derive a path of the outline in blender, it feels like it should be possible to convert that to a csv.

edit: https://blender.stackexchange.com/questions/51706/exporting-coordinates-of-vertices-to-csv maybe I can use this somehow.

You start recording ai just after the start line and it automatically finishes when you have crossed it again. The small gap between where you have started and the finish line gets connected automatically by magic.
I would consider the AI-line helper app too
Thanks. How do I do it with the pitlane?


edit: Nevermind, I overlooked the first answer to my question first. The pitlane quesiton has already been answered :)
 
Last edited:
You start recording ai just after the start line and it automatically finishes when you have crossed it again. The small gap between where you have started and the finish line gets connected automatically by magic.
I would consider the AI-line helper app too
This is wrong. Never ever start recording after the line. Always start recording before the line. If you start after the line the 0 point of the line will not line up with the actual start finish line. When that happens timing apps can get messed up and also time attack mode will not work right.
 
This is wrong. Never ever start recording after the line. Always start recording before the line. If you start after the line the 0 point of the line will not line up with the actual start finish line. When that happens timing apps can get messed up and also time attack mode will not work right.
Ah thank you! Seems like I made it wrong for the last year or so. I tried it before the line but then the line was just the bit from the starting point to the start line
 
mh. driving along the outline seems really imprecise.
but the sole reason we have those AI borders is that the AI will not drive over them when overtaking. If you export it in some way from Blender you will (probably) with something with jagged edges and crevices which may result in erratic AI driving. But if you drive the lines yourself you can very precisely set those limits, and incorporate things like pit exit lines and natural flow on curbs.

the linked app works really well, it records the position of either the front left or right wheel. Just take a slow car.
 
Exporting from blender is more precise than driving it with the tool. Plus you can fine tune the edges very easily. The AI line tool is great but it is meant for non track builders.
 
Ah thank you! Seems like I made it wrong for the last year or so. I tried it before the line but then the line was just the bit from the starting point to the start line
The first lap you make is always a throw away. It will usually have a little funky loop at the line. But on the second lap it will go away.
 

Latest News

Do you prefer licensed hardware?

  • Yes for me it is vital

  • Yes, but only if it's a manufacturer I like

  • Yes, but only if the price is right

  • No, a generic wheel is fine

  • No, I would be ok with a replica


Results are only viewable after voting.
Back
Top