I'm not sure if you know, you can buy a license from lfs.net once and have access to other updates too.

Why do you want to buy it from Steam rather than lfs.net?

I really don't mind if you buy it or not, I'm just curious about what you think Steam could offer that we can't.

Quote from Gutholz :

As I was watching yesterday's lunar eclipse, I wondered:
Will there be a moon?

I really want to add the moon, it seems quite wrong not to!

But don't worry everyone, this is obviously one of those things that has to wait until after the release.

You could sell a car with some nice improvements and also a leaking fuel tank, some holes in the bodywork, faulty electrics and an uncomfortable seat. Or you could fix the known issues and release it at that point instead. Which is better?

This is an entirely new version. There is no compatibility between tracks in the old and new versions. The old version cannot possibly load the new tracks, and the new version looks terrible with the old tracks. Good news though, they have all been updated, as you can see by reading the previous progress reports I linked to above.

The new version is not a patch for old LFS. It has to be released all at once, and it will be released with flaws, in as a test version. But there is absolutely no point releasing it when it is not ready for testing and we are working full time trying to fix all the issues we already know about.

There is a point to release it before it is perfect (and will never be perfect anyway) but we aren't at that point yet.


EDIT: As mentioned, some important things are simply not done yet. Also, at some point I need to build a public version, actually exporting everything from the development version into a publicly releasable folder. Even that will take me a few days, probably a week by the time I've added some code support to help me build a clean version a few times during the testing stages. To move from a development version to release isn't simply a matter of declaring it ready. It's a careful process of exporting and encrypting the necessary files, possibly stored a bit differently from the current public version. It's not a big deal, I'm just trying to illustrate that there isn't a version to release yet, even if it was ready which it isn't.

Last edited by Scawen, Tue 9 Sep 2025, 20:41.

OK, we get that in Edge too. In Firefox it goes where it's supposed to. So I don't think this test tells us much so far. My son got that result and he is not a moderator.

My hypothesis at the moment is that redirect is browser related and doesn't shed any light on Zero7's mod download issue.

Oh that is strange, for me it stays on the competition pages and I can click around them.

I guess pinging the master is probably not relevant as the mods and skins don't come from the master server, but do come from our web server.

I wonder if BT could be blocking http (rather than https) connections.

Yesterday I came across part of our site that is so old it uses only http. I wonder if BT allows you to connect to that from your browser.

Link: http://competition.lfs.net/

Sorry, I know there are other questions but I can't answer them all as I'm still busy working long days every day.

Quote from nacim :

So there is only baked ambient occlusion during the day, and baked lighting during the night right ?
So no baked Global Illumination that blends during the various time of day for the moment ?

In our system, I call the following ambient lighting:
1) Lighting from the sky
2) Lighting from nearby objects
3) Lighting from artificial lights (excluding headlights which are obviously dynamic)

These are linked together in our system, values baked into vertex lighting and they are in operation at all times, with a varying effect depending on the lighting situation. Some artificial lights are on at all times (e.g. pit garages) and some come on only at night (street lights).

Quote from tankslacno :

Does this new dynamic lightning system force you to release all updated track environments at the same time? Is it even possible to release a patch, where one of the track environments is updated and that particular track environment uses the new dynamic system, while all the other tracks still use the current 2-4 static lightning options for time being?

No, all tracks must be updated. This has all been explained. I know it's a lot of reading but there are links at the bottom of this page:
https://www.lfs.net/patchInfo/report-dec2024-ky.php

That ambient lighting, including from the artificial lights, is baked and vertex based. Where there is an issue like the one you spotted, Eric can improve it by manually increasing the geometry in those areas.


EDIT: To be clear, in the older versions of LFS, we had automatic geometry splitting to try to allow more detailed shadows (that were also for sun shadows). It worked OK in some places, but sometimes it creates long shards of undesirable lighting. In the new version of LFS, direct lighting shadows are done using shadow maps. Vertex lighting is now only for the ambient lighting (including sky lighting and street lights). Automatic splitting then seemed to cause more problems than it could solve, so we decided that it would be better if Eric has full control of the geometry instead of leaving anything to a flawed algorithm.

The design compromise of storing lighting in vertices has pros and cons, it's very efficient and helps LFS keep running on less powerful GPUs that is important for many of our customers. In my opinion it usually works quite well but there can be problems in some places, e.g. sometimes near street lights or occasionally there are excessive ambient shadows. I think it's not worth trying to make it too perfect but it can be improved a bit in some places where it might be seen a lot or has obvious issues.

Last edited by Scawen, Mon 8 Sep 2025, 08:29.

Thanks for the updates!

I had a fun race against 3 AI in VR at Fern Bay Club.

The climb was a struggle for that car and the internal rear view mirror is not very useful!

Nice mod and I learned some more about 2CV history.

Yeah I know that's hard, and some people are good at working with LFS strange sound engine.

Anyway, I had a quick go and it improved just by setting it to flat twin and max inertia mul.

I'm not familiar with this old 425cc version, but I found a short video of it.
https://www.youtube.com/watch?v=1S-IYaQ4MIU

Anyway in the SHIFT+A editor I think I got a better result by reducing the "Collector MIX" and turning up "Header MIX" and I think maybe "Unevenness" isn't needed for this engine?

Last edited by Scawen, Tue 9 Sep 2025, 10:36. Reason : removed off-topic

I love the Citroen 2CV and drove my mother's 2CV6 a bit around 1990.

I thought I'd have a go on this nice looking mod.

I noticed it is too quick revving and doesn't sound right.

My suggestion: it should be a boxer twin, not inline 4. And the other thing I remember was the relatively heavy flywheel compared with the engine size, so it took longer than most engines to rev up but you could get quite a bit of acceleration on letting out the clutch from a standstill. Top speed was slow but could reach about 70mph if I remember correctly.

I found a video here that shows the engine revving: https://www.youtube.com/watch?v=b1mSMEeruNo

Quote from Byrontr :

...but if we pay once, there will be no problem.

That's how it works, you pay once. It's not a subscription or rental business model.

It's my understanding that the S3 price in Turkey (£18) is approximately 1000 Lira. I read on Wikipedia the minimum wage is around 26,000 Lira per month. So this leads me to believe it is around 1 day's work to buy an S3 license.

That is for a product we have worked on for around 25 years, and you keep that license for life, with unlimited access to our services (up to the S3 license level, as long as we continue to work on LFS).

To purchase 25 years of work and unknown future years of work, for one day of your own hard earned cash, doesn't seem really unreasonable to me. I understand that in some other countries, you can work less than half a day on minimum wage to purchase an S3 license. But it's not really off the scale. We do provide services to you that cost money and huge amounts of time.

The developers don't earn that much these days, and we also need to buy food, heat our homes, etc, and spend money on servers to keep our services running.

Nearly all the other products I see on the internet, you have to pay monthly or pay for upgrades in order to keep using them. I think we have one of the best deals available for an ongoing product in continual development.

As part of the privacy and security updates I have removed unnecessary data from our system, including birthday, so that's why the age graph has disappeared. We don't know anyone's age.

I've fixed a bug about friends lists in another window but don't think this would be related, as it was just a bug in that window.

I've just tested this now and it seemed to be working. I was looking at the live alert window and my friend appeared and disappeared as expected in the live alert window (I have a 1-minute polling interval in my LFSW settings).

Please can you tell me if it seems to be working now, or there is a different way to see the bug?

Thanks for the report.

Please can you tell me if it appears as expected now?

Thanks for the report.

Please can you check if it looks as expected now?

Quote from loopingz :

That looks great as always! Fine looking job...

Yes, it's all with the 1000 Hz physics and multithreading which seems stable and solid.

I don't know if this helps, but I just ran 8 BF1 AI at KY Oval, in daytime, on my old PC, with everything turned up, I got 80 to 100 FPS.
EDIT: Internal view, mirrors on, shadow maps in mirrors, 4 shadow map cascades at 2048x2048 each, 1920x1080.

Spec:
Intel Core 2 DUO (dual core) E8500 3.17 GHz
8 GB RAM
NVIDIA GeForce GTX 1050 Ti

You can save by reducing textures to half res and reducing shadow maps and environment maps if needed.

I think it runs well on a lot of older computers. I think we tested and it ran OK (with some things turned down) on a really slow laptop with onboard graphics. But we don't have the resources to run tests on all sorts of configurations, so can't promise anything. I guess we'll find out in public testing.

Graphics and physics are both more demanding than in the old version but they are split over two CPU cores instead of just one.

Last edited by Scawen, Thu 4 Sep 2025, 17:52.

Quote from tankslacno :

Thanks for the explanation! Haven't seen that issue happening on any track configuration (that has pits) ever before. Strange that it has happened on that particular track configuration

I think the same will happen on "Chicanes" and "Oval" in the new version as Chicanes is just the Oval with a couple of chicanes.

The attached image shows how the issue came up - basically the pit lane is so long compared with the track. It would be dangerous for them to swerve off into that pit lane after passing the last split.

Quote from evandroPRO123 :

pc specs ?

Guess it's the same PC he had 3 hours ago.

https://www.lfs.net/forum/post/2133827#post2133827

Quote from tankslacno :

I just have that concern that making strategic pit stops for AI may become unnecessarily harder. Is there a reason for this change?

Of course there is a reason, I'm way too busy to waste my time coding random changes pointlessly.

Eric found that on the "Kyoto Chicanes" configuration (that uses a lot of the Oval) AIs in some cars would run out of fuel, because they decided to pit too late, given their starting fuel load.

At first I thought it must be an error in their fuel calculation but all seemed OK, they were just heading into the pits too late. It turned out that they made their decision at the last split, but on at least this configuration, the AI drivers' imaginary line to head for the pits starts before that final split point.

So when they would decide to pit, it was too late, I guess you see what I mean. So I made them make the decision to pit or not, at the last moment they could still enter the pits, which of course makes more sense and solves the bug.

Thanks for the feedback and discussions!

I'll now close this thread as we have a new September Progress Report

Hello Racers,

Eric has recorded two unedited videos of AI drivers racing around the updated South City and Kyoto Ring, the two most time-consuming of our updated tracks. The videos are taken from a high viewpoint, using a time multiplier of 200, so you can see the day to night transitions.


Video 1: LFS AI drivers at Kyoto Endurance




Video 2: LFS AI drivers at South City Long




Program development:

I've had to work a lot since 25th July on security updates.
Here is a short list of a few notable updates from before that time.

May:

Daylight saving time is now accounted for (on the UK tracks)
Specified start time is now given in track local time (not UTC)
Accurate sun direction using astronomical calculations
- previously used a rough model based on circular orbit
- apparent sun direction now adjusted for atmospheric refraction
Leap years are now supported (can set 29 February on a leap year)
Fixed time renamed to "set" and includes a time multiplier option

June:

Multithreading update for sun position
- sun position is updated as part of the game update
- means that sun height can now affect physics (e.g. temperature)
- initial test increases air temperature for higher sun direction
Removed jaggies from sky texture around dome edge (at horizon)
Display AI calculated wing and final drive settings in setup screen
AI decide to pit at pit lane transition point rather than last split

July:

Front of vehicles now consistently aligned with start grid slots
Splits and finish line activate when front of vehicle crosses line
Improved AI pit garage entry and exit

I didn't know I had changed a setting, maybe I did when testing something. My profile should be visible now.

You know I can't answer that until it's ready.

I have learned that I can't predict the future.