Realism- and gameplay-related hardcode fixes for SH3.EXE

[h.sie]

I apologize for any problems my work makes. These problems are not intended. I understand my work as WIP (Mod Workshop!!) and thus problems can occur - they are almost naturally when working with Asm in different OS. So people who fear CTDs should not use my patches - or they should make tests only with an second Installation of SH3 as I do. A CTD is not that worse. It's only that the EIP (instruction pointer of the CPU) doesn't know where to go.......

USE ON YOUR OWN RISK!

Time to update my first post.

[h.sie]

@Stiebler:

No I did not mean a different constant setting. I had random changes of fog in mind.

[Yoriyn]

H.sie: Can you try to upload new EnvSim.act with 12h (or any other) weather change intervals time? Cos last one (with 2h) causing constant weather for me (no change).

[h.sie]

@Yoriyn: Currently not, since work is still in progress. If I'm ready, I'll hopefully upload one working bugfree version.

I am sorry for not being able to fulfill individual wishes. In order to not drive crazy I have to focus on one topic at the time.

And I also apologize for not responding to requests for individual wishes (which normally is not my style).

h.sie

[Yoriyn]

Don't worry about my wishes, I'm very happy of progress you made in the game modification. Just can't wait when you make another big step to fix a weather

[Magic1111]

Originally Posted by makman94

H.Sie , the envsim.act you uploaded is crashing (maybe is an issue again for the win 7 ?)

Originally Posted by Stiebler

@General:
Like Makman, I found that use of H.sie's new envsim.act file caused a crash in second-stage loading (after the main menu appears, and the player starts a patrol). I am also using 64-bit Windows-7. This occurred with an un-modded form of the 4GB_Sh3.exe that I have been using for some time now without problems.
In my case, it was a very serious crash, that apparently corrupted at least one other key file. I had to copy over a clean install from another drive and start again.

Stiebler.

Hi h.sie !

I´ve become the same CTD using the envsim.act (Win 7 64Bit) !

Best regards,
Magic

[h.sie]

Every CTD pains me. And I must say "thank you" and "sorry" to those who anyhow help me with testing. But it shows that I have to restrict on only essential fixes in the future. no eyecandy, no "nice-to-have".

[Stiebler]

@H.sie:

5. Windspeed:
The internal algorithm for windspeed calculates an emphasis factor (depending on season, region and so on) and multipies this emphasis factor with a random number. The result lies between 0 and sometimes more than 20m/s. But later the values are delimited to [0,499.....15m/s]. We should not change that -> Side effects. But that's the reason why windspeed sticks at 15m/s; the algorithm is overdriven in some situations.

This is a very valuable discovery.

Surely this is the first point to fix: emphasis-factor x random-number can give more than 20 m/s.

Redistribute - do NOT create an upper limit! - the random number so that >15 m/s cannot be achieved. For example, if the random numbers are in the range 0-255, but a random number > 200 results in wind speeds > 15 m/s, then multiply all random numbers produced in this part of the code by 200/256.

This will require a jump out of the code to a few lines of assembly at the end of the code, then a jump back to the exit point. Can you show a few lines of the key assembly code in this forum-thread? (with comments about which variables are referred to?).

Stiebler.

[Gammel]

great work you already did h.sie - sorry for interupting the weather talk...

snorkeling with type XXI i have no diesel underwater sound, can anyone using the latest bugfix patch confirm?
Further f12 camera image is blurry 5 knoths and above but thats no biggy to me.

[Yoriyn]

I have the same problems on moded version, but can't reproduce this problems on stock (clean) game.

[CherryHarbey]

Originally Posted by Stiebler

@H.sie:
This is a very valuable discovery.

Surely this is the first point to fix: emphasis-factor x random-number can give more than 20 m/s.

Redistribute - do NOT create an upper limit! - the random number so that >15 m/s cannot be achieved. For example, if the random numbers are in the range 0-255, but a random number > 200 results in wind speeds > 15 m/s, then multiply all random numbers produced in this part of the code by 200/256.

This will require a jump out of the code to a few lines of assembly at the end of the code, then a jump back to the exit point. Can you show a few lines of the key assembly code in this forum-thread? (with comments about which variables are referred to?).

Stiebler.

or alternatively take two random numbers between 0 and 16 and multiply together. Maximum value will still be 256 and answer from the range 0 to 256 but skewed towards the lower end.

[h.sie]

@Gammel, Yoriyn: Thanks for response. I tried to reproduce your sound issue on Win7/64 and WinXP/32. I chose a XXI in 1945. Started snorkelling. In the modded game I could hear the diesel sound for about 20 seconds while going Full ahead, but then engines are limited to 1/3 ahead - diesel sound disappeared almost completely. But the same happened in unmodded sh3.exe when I manually reduced speed to 1/3 ahead: diesel sound disappeared almost completely.

I also cannot reproduce your F12 freecam bug, neither on Win7/64 nor on WinXP/32. F12 always looks good for me. Only explaination could be a modded cameras.dat (maybe someone enabled blur effect for free cam???). I use the original cameras.dat from GWX3. So please disable all mods that change cameras.dat and try again. What GUI do you use?

h.sie

[Gammel]

Originally Posted by Gammel

great work you already did h.sie - sorry for interupting the weather talk...

snorkeling with type XXI i have no diesel underwater sound, can anyone using the latest bugfix patch confirm?
Further f12 camera image is blurry 5 knoths and above but thats no biggy to me.

I have the same problems on moded version, but can't reproduce this problems on stock (clean) game.

dam'n you're right - it's the f**kin mods
but which one?

Clean GWX with h.sie's patch is fine (diesel acoustics snorkeling), sorry for false alarm

note to myself: only reporting here while testing with all mods disabled :-))

gui used was MaGui final...

[h.sie]

@Stiebler:

You can see it by yourself: Take a debugger (I use OllyDbg2) and set a Breakpoint at EnvSim.act+0xDD61. This code section is executed only when new weather is calculated, so use high TC.

Take a look at the variable referenced by [ESI+3d4]. At the end of the routine, it contains the NewWindspeed calculated!!!

At EnvSim.act+0xDD61 you see the routine call for a random number (RND) between -1 and +1. After the call, the RND is in the FPU, st(0).

In EnvSim.act+0xDD66 the random value (RND) is multiplied with 15,0. So that st(0) can contain values between -15 and +15.

Later, after some other code, In EnvSim.act+0xDD99 a value of 15 is added, so that we have values between 0 and 30 in st(0).

Later, at EnvSim.act+0xDDA5 and so on, the value of st(0) is restricted to [0,4499 - 15].

This is what I described earlier. Looking at the windspeed statistics I presented earlier, one can see, that windspeed was never higher than 6 m/s in Koenigsberg. From this I reasoned an (intelligent) emphasis factor (area dependent?), but which I could not locate in the code up to now.

But I made a mistake: The code does not always go this described easy-to-understand path, that means things are more complex than I initially thought. It needs some more time, but I'm pessimistic I'll understand what happens.

So we can only correct the calculated values for windspeed and fog by a little patch at the end of the routine. Maybe some random function or a function that restricts the length of fog periods to a certain value.

h.sie

[Magic1111]

Hello h.sie !

Short question: Is it possible to use your modded File "envsim.act" with your Patch V15C (or without a modded SH3.exe from you as an "Standalone-MOD") ? Or must I use the "envsim.act" ONLY in COMBINATION with your Patch V15D BETA3 ?

Best regards,
Magic