-ReinForce Alert - worldwide discussion

[Lex15]

Сообщение от CrazyIvan

So far, we are moving slowly but surely. everything will change too much. so special attention for testing. To this we must also add that the database must be radically changed.

Do not want to lay out a raw version with bugs. Moreover, the problem of the equivalence of the player and AI platforms is now being solved. As I said earlier, the sonalysts did not equalize the player and AI subs in the detected abilities.


Let's try to make them equal.

Keep up the good work.

[CrazyIvan]

Сообщение от CubanFoxtrot

Hi CrazyIvan.

Don't know if Google Translate is doing a great job here and if I understood entirely, but the point you're raising is the NSE (AI) simulation models are semplified version of the EXE's (player) ones?
Hence you can get asymmetric detection triggers from sensors?
Performance-wise, it would make sense as NSE has to manager many more entities than the EXE side.
The latter would also need additional details for presentation and user interaction.

Sorry in advance in case I didn't grasp what did you intended to explain and show us (had no time to test the scenario yet).
CF.

You're right.
They made it easier to calculate parameters for AI platforms.
For example, this situation - your sub is in the course of 000.
Behind you,there is an AI hostile sub that is also heading 000. (in your baffle area)
You actively ping and the AI sub behind you detects your ping. %-)

The sonalysts did not recognize WHERE the active sensor radiates.
In this example, they simply check that the emitter of the active signal is not in thebaffle zone of the most active intercept.

At the moment,we have eliminated this bug.
Now it will always be checked whether the target is in the cone of RADIATION of the activesignal - active ping, or radar radiation if the radar does not cover 360degrees. There are aviation radars that have 180 degree coverage, not 360degree coverage.

[CubanFoxtrot]

Great. That was more or less what I had understood then from your previous message.
If I were in your shoes at this point, I'd try to rewrite the whole (exe, not nse) sensor's detect routine in C++ and put it in a DLL.
Takes tons of patience, but maybe it's the safest and most solid solution.
I mean being consistent just through binary patching in such a complex context, umh, I might be wrong, but sounds like a very complicated task to me!
Then try to isolate step by step the entity/sensor/etc. unmatcing input data and feed that reconstructed routine with a shared interface structure that is valid both for AI and player sources (you're gonna fill it before calling the dll code).
Also a common interface to apply the output/results would be needed, probably.
Easier said than done. But just my 2 cents.
Good luck and keep up the good work, as said earlier.

EDIT:
Found some veeery old and never completed notes. Certainly they'll be of no use given your actual understanding, but better sharing for sake of it.

void OsSensor::Detect(int dt)
{
    if (this->_unk_1c8_ == 0) {
        this->V18_NotifySensorEnabled(0, 1);
        return this->_unk_120_;
    }

    if (this->UpdateIntervalInMsecs_1d8 * 0.001f == 0.0f) {
        return this->_unk_120_;
    }

    this->UpdateAccumTime_1dc += 1;
    this->UpdateDeltaTime_1d4 = dt;

    KinematicData var_f8;
    this->entityObj_1bc->_unk_16c_.motion_28->V018_GetCrsState15cAtDeltaTime(dt, &var_f8, true);
    float var_A8 = this->entityObj_1bc->_unk_16c_.motion_28->V11c_GetOwnSpdAtDeltaTime(dt);

    OsSensor::MountPlatformMotionData var_c8{ var_f8, var_A8 };
    this->Update124FromPlatformMotion(&var_c8);

    if (!this->IsActivatedOrDeployed()) {
        return this->_unk_120_;
    }

    std::int8_t var_150 = INT8_C(0);
    if (this->mastData_1c4 != nullptr)
    {
        float mastDeployTime;
        const auto mastDeployState = this->mastData_1c4->V0c_GetDeployEventState(dt, &mastDeployTime);
        if (mastDeployState == 0) {
            if (mastDeployTime * 0.001f <= 0.0f)
                return this->_unk_120_;
        }
        else if (mastDeployState != 1) {
            return this->_unk_120_;
        }

        this->mastData_1c4->V18_UpdateSensorOffsetWithMastMotion(dt, &this->mountPlatformMotionData_124);
        this->_unk_148_ = GetAltitudeToAbsWorldUnits(this->mountPlatformMotionData_124.z); // ftol

        if (this->mastData_1c4->_unk_5c_) {
            return this->_unk_120_;
        }

        if ((this->flags_14 & DA_THERMALLAYER) && (this->mountPlatformMotionData_124.z < 0.0) && (var_c8.z < this->entityObj_1bc->unk_4a8->draft_0c))
        {
            var_150 = static_cast<std::int8_t>(pow((245.f - std::min(this->mastData_1c4->V10_GetDeployLength(dt), 245.f)) * 0.0040816325f, 2.f) * 30.f);
        }
    }


    if (this->nMinOperationAlt_c4 - 0.5f <= this->mountPlatformMotionData_124.z && this->nMaxOperationAlt_c0 + 0.5f >= this->mountPlatformMotionData_124.z) {        
        auto ent;
        if (this->hullID_1e8 != 0 && (ent = g_entityObjCache.find(this->hullID_1e8)) != nullptr) {
            const auto oldConeAngle1F4 = this->coneAngle_1f4;
            KinematicData var_124;
            ent->motion_194->GetCrsState20(&var_124, true);
            const auto v[2] = {
                var_124.x - this->mountPlatformMotionData_124.x,
                var_124.y - this->mountPlatformMotionData_124.y
            };
                this->UpdateConeAngle1F4AndMountPlatformMotionData124heading(CVector2::GetAngleInDeg(v[0], v[1]) , &var_c8);
                if (this->dbSensor->animPartName_8e[0] != '\0') {
                    const theta = fmod(var_c8.heading_14 + this->coneRotAxisZ_0d0 + 360.f, 360.f);
                    CAngle::Diff(theta, this->coneAngle_1f4) - CAngle::Diff(theta, oldConeAngle1F4);
                    msg->unk00 = 48; // type
                    msg->unk04 = 32; // size                   
                    msg->unk0c = this->entityObj_1bc->_unk_16c_._unk_00; // hullID
                    msg->unk10 = { 60 + this->sensorSlotNum_340, -1, -1, 0x00 }; //short4
                    msg->unk18 = dt;
                    msg->unk1c = 1000;
                    msg->unk20 = decimal_t(CAngle::Diff(var_c8.heading_14, oldConeAngle1F4) * DEG2RAD, 0.f);
                    msg->unk28 = decimal_t(CAngle::Diff(theta, this->coneAngle_1f4) - CAngle::Diff(theta, oldConeAngle1F4) * DEG2RAD, 0.f);
                    sub_62A630(msg, true);
                }
            // TODO
        }
        // TODO
    }

    return this->_unk_120_;
}

[seadog0514]

Сообщение от CrazyIvan

Очень давно, еще со времен игры СУБ Комманд, , многие игроки сетовали на то что, нет управления торпедой по глубине и скорости в режиме реального времени.

Многие замороченные вещи тогда были придуманы моддерами того же аддона СКХ-2С для Суб Коммнды - в плане того, как управлять торпедой по проводу. Хитрые перещелкивания, кнопками ЕНАБЛЕ-ПРЕЕНАБЛЕ, в торпедном проводном интерфейсе.

Но только сейчас, эта опция стала действительно, в управлении, легкОй и простой.

Вот так , сейчас выглядит интерфейс, управления торпедой.

Всегда можно поменять скорость и глубину торпеды по проводу (Если только торпеда, не в фазе атаки).

hi
any plan to impose maneuvere-limitation while wire-guiding a torpoto ?

[CrazyIvan]

Сообщение от CubanFoxtrot

Great. That was more or less what I had understood then from your previous message.
If I were in your shoes at this point, I'd try to rewrite the whole (exe, not nse) sensor's detect routine in C++ and put it in a DLL.
Takes tons of patience, but maybe it's the safest and most solid solution.
I mean being consistent just through binary patching in such a complex context, umh, I might be wrong, but sounds like a very complicated task to me!
Then try to isolate step by step the entity/sensor/etc. unmatcing input data and feed that reconstructed routine with a shared interface structure that is valid both for AI and player sources (you're gonna fill it before calling the dll code).
Also a common interface to apply the output/results would be needed, probably.
Easier said than done. But just my 2 cents.
Good luck and keep up the good work, as said earlier.

EDIT:
Found some veeery old and never completed notes. Certainly they'll be of no use given your actual understanding, but better sharing for sake of it.

void OsSensor::Detect(int dt)
{
    if (this->_unk_1c8_ == 0) {
        this->V18_NotifySensorEnabled(0, 1);
        return this->_unk_120_;
    }

    if (this->UpdateIntervalInMsecs_1d8 * 0.001f == 0.0f) {
        return this->_unk_120_;
    }

    this->UpdateAccumTime_1dc += 1;
    this->UpdateDeltaTime_1d4 = dt;

    KinematicData var_f8;
    this->entityObj_1bc->_unk_16c_.motion_28->V018_GetCrsState15cAtDeltaTime(dt, &var_f8, true);
    float var_A8 = this->entityObj_1bc->_unk_16c_.motion_28->V11c_GetOwnSpdAtDeltaTime(dt);

    OsSensor::MountPlatformMotionData var_c8{ var_f8, var_A8 };
    this->Update124FromPlatformMotion(&var_c8);

    if (!this->IsActivatedOrDeployed()) {
        return this->_unk_120_;
    }

    std::int8_t var_150 = INT8_C(0);
    if (this->mastData_1c4 != nullptr)
    {
        float mastDeployTime;
        const auto mastDeployState = this->mastData_1c4->V0c_GetDeployEventState(dt, &mastDeployTime);
        if (mastDeployState == 0) {
            if (mastDeployTime * 0.001f <= 0.0f)
                return this->_unk_120_;
        }
        else if (mastDeployState != 1) {
            return this->_unk_120_;
        }

        this->mastData_1c4->V18_UpdateSensorOffsetWithMastMotion(dt, &this->mountPlatformMotionData_124);
        this->_unk_148_ = GetAltitudeToAbsWorldUnits(this->mountPlatformMotionData_124.z); // ftol

        if (this->mastData_1c4->_unk_5c_) {
            return this->_unk_120_;
        }

        if ((this->flags_14 & DA_THERMALLAYER) && (this->mountPlatformMotionData_124.z < 0.0) && (var_c8.z < this->entityObj_1bc->unk_4a8->draft_0c))
        {
            var_150 = static_cast<std::int8_t>(pow((245.f - std::min(this->mastData_1c4->V10_GetDeployLength(dt), 245.f)) * 0.0040816325f, 2.f) * 30.f);
        }
    }


    if (this->nMinOperationAlt_c4 - 0.5f <= this->mountPlatformMotionData_124.z && this->nMaxOperationAlt_c0 + 0.5f >= this->mountPlatformMotionData_124.z) {        
        auto ent;
        if (this->hullID_1e8 != 0 && (ent = g_entityObjCache.find(this->hullID_1e8)) != nullptr) {
            const auto oldConeAngle1F4 = this->coneAngle_1f4;
            KinematicData var_124;
            ent->motion_194->GetCrsState20(&var_124, true);
            const auto v[2] = {
                var_124.x - this->mountPlatformMotionData_124.x,
                var_124.y - this->mountPlatformMotionData_124.y
            };
                this->UpdateConeAngle1F4AndMountPlatformMotionData124heading(CVector2::GetAngleInDeg(v[0], v[1]) , &var_c8);
                if (this->dbSensor->animPartName_8e[0] != '\0') {
                    const theta = fmod(var_c8.heading_14 + this->coneRotAxisZ_0d0 + 360.f, 360.f);
                    CAngle::Diff(theta, this->coneAngle_1f4) - CAngle::Diff(theta, oldConeAngle1F4);
                    msg->unk00 = 48; // type
                    msg->unk04 = 32; // size                   
                    msg->unk0c = this->entityObj_1bc->_unk_16c_._unk_00; // hullID
                    msg->unk10 = { 60 + this->sensorSlotNum_340, -1, -1, 0x00 }; //short4
                    msg->unk18 = dt;
                    msg->unk1c = 1000;
                    msg->unk20 = decimal_t(CAngle::Diff(var_c8.heading_14, oldConeAngle1F4) * DEG2RAD, 0.f);
                    msg->unk28 = decimal_t(CAngle::Diff(theta, this->coneAngle_1f4) - CAngle::Diff(theta, oldConeAngle1F4) * DEG2RAD, 0.f);
                    sub_62A630(msg, true);
                }
            // TODO
        }
        // TODO
    }

    return this->_unk_120_;
}

Yes, in fact it was difficult, but I almost managed. Now I think that the radar radiation angles should be properly taken into account for the receiving ESM sensor. The horizontal angle of radiation is tracked perfectly. But I had great difficulty with the azimuth angle - when you need to calculate the direction of the beam in the vertical direction.And cut off the signal, if ESM receiver sensor it is not in the emitter cone.


The vertical angle for active interception, was not a big problem. This was resolved very quickly.


I think I'll add a few degrees to the emitter cone, to each side, in the form of spurious emissions.
When the radar of the emitter still does not see the signal from the target platform, but the receiving side will see these stray emissions if the direction of the radar cones, is very close to the bearing to the player's platform.

PS: I'm afraid that this will be a completely different simulator than the simulator from Sonalysts. Since many cheats have been removed (for example, the unequal detection of AI platforms - when a player on his NB sonar sees the 1-st frequency of the target, and the target can only see the 3 - rd frequency of the player, which reduces detection by several times).
Many default scenarios, will simply not run as they did in the default version.


>> if (this->_unk_1c8_ == 0) {

What offset does this code start with? %-(

Your listing, is something unusual.
---------------------------
Along the way, I removed the bug with the contact age update timer. There was a bug where, for example, the radar emitter went beyond the radio-horizon, and the ESM autocommand stopped making marks on the ESM display. When the contact again popup the radio-horizon, the autocommand stopped marking this contact. The age of the contact on the DDI dispay at nav.map was not 0:00, but continued to increase (0:05 ; 0:06 ect) as if the contact was still invisible to the ESM sensor.

[CrazyIvan]

The ESM sensor on AI subs has also been fixed for the new version.
Since in the default version only the periscope (visual sensor) on the AI platform could detect contacts when the sub is at periscope depth.
Now, the ESM sensor will also operate at periscope depth and detect radar emissions.


Simply put - the ESM sensor at AI Sub, got its own working telescopic mast.

However, I'm more and more concerned about realism versus playability.

[GROHOT]

Hello, Crazy Ivan!
Good work! I hope new version be realised really soon.
May you give some news?

[GROHOT]

Hello Crazy Ivan!
Any news about new version of RA?