32 replies to this topic

[GreyNovember]

MrMadguy, on 27 May 2025 - 10:32 PM, said:

Tie convergence to selected target, not to where crosshair points to. As simple, as that.

This is a halfway decent solution to a related problem, which would also make stealth and ECM significantly more valuable as a side effect.

Because if you can't be targeted, you can't be properly shot, and you gain lots more effective lifespan.

It would also make quick snapshotting much, much worse, now that convergence is no longer tied to your crosshair as you suggest.

Which also leaves the question of WHERE on the enemy mech you're going to converge. Because this sounds like you want all weapons to effectively have auto aim lockon to presumably center of mass.

Stops being so simple once you start asking questions about who it's okay to screw over.

[KursedVixen]

LordNothing, on 27 May 2025 - 01:08 PM, said:

youve never been under the hood have you? all game engines are full of spaghetti code, its the nature of the beast. i remember when they opened the freespace engine, it was full of gotos. every one of them was a necessary optimization to get out of a loop in a hurry to do something critical elsewhere. single threaded engines were fun.

decent freespace? like The great war with Shivans,Vasudans and such?

[Void Angel]

"Simple" solutions posited by random forumites never are.

[MrMadguy]

GreyNovember, on 27 May 2025 - 10:37 PM, said:

This is a halfway decent solution to a related problem, which would also make stealth and ECM significantly more valuable as a side effect.

Because if you can't be targeted, you can't be properly shot, and you gain lots more effective lifespan.

It would also make quick snapshotting much, much worse, now that convergence is no longer tied to your crosshair as you suggest.

Which also leaves the question of WHERE on the enemy mech you're going to converge. Because this sounds like you want all weapons to effectively have auto aim lockon to presumably center of mass.

Stops being so simple once you start asking questions about who it's okay to screw over.

No, ECM blocks radar and communication. This thing is more like simple rangefinder. ECM shouldn't block it.

All Lights already abuse radar derp cheat. If it would be the case, they would also abuse ECM. Most of them already do, but at least not all.

[GreyNovember]

MrMadguy, on 28 May 2025 - 04:52 AM, said:

No, ECM blocks radar and communication. This thing is more like simple rangefinder. ECM shouldn't block it.

All Lights already abuse radar derp cheat. If it would be the case, they would also abuse ECM. Most of them already do, but at least not all.

You cannot select a target without a lock. Something that ECM and stealth armor does, and why people run it in the first place. Have you never played, or played against a Thanatos? A 4L Marauder II? ANY Clan mech that can run an ECM omnipod?

Radar Deprivation is run by everyone who does not want to be tracked behind walls after they leave visual contact. That's EVERYONE who has points to spare, and cares about not telling the enemy where they're going after you get pinged by "Enemy Spotted".

Come on man, at least TRY to be consistent when it comes to your own solutions.

[MrMadguy]

GreyNovember, on 28 May 2025 - 05:13 AM, said:

You cannot select a target without a lock. Something that ECM and stealth armor does, and why people run it in the first place. Have you never played, or played against a Thanatos? A 4L Marauder II? ANY Clan mech that can run an ECM omnipod?

Radar Deprivation is run by everyone who does not want to be tracked behind walls after they leave visual contact. That's EVERYONE who has points to spare, and cares about not telling the enemy where they're going after you get pinged by "Enemy Spotted".

Come on man, at least TRY to be consistent when it comes to your own solutions.

I know, but this mechanic requires target. And there is difference between being unable to lock or transfer data to teammates and just having target for weapons to converge to. Remember? We don't even need to have target itself. We just need to calculate distance to it.

This game already has ton of abusable mechanics, so nobody plays it, and you defend another one? If Lights would know, that their lag/JJ-shield would depend on ECM - we would have 100% Lights abusing ECM on a top of 100% of them already abusing radar derp cheat. Yeah, nice.

Edited by MrMadguy, 28 May 2025 - 06:10 AM.

[GreyNovember]

MrMadguy, on 28 May 2025 - 05:43 AM, said:

I know, but this mechanic requires target. And there is difference between being unable to lock or transfer data to teammates and just having target for weapons to converge to. Remember? We don't even need to have target itself. We just need to calculate distance to it.

Explain.

How do you define a "Target", without a lock, when you want to no longer make convergence depend on your crosshair?

What mechanic are you using to determine the "Target"?

[Quicksilver Aberration]

Convergence on target lock is a much needed feature. Removing convergence on reticle without lock is just a bad idea. Just like all of PGI's ideas, removing nice things we've had to incentivize running certain equipment is just a bad idea, especially when lack of doritos is already a huge boon against a majority of this player base.

Meep Meep, on 27 May 2025 - 10:03 PM, said:

Training grounds and academy are not connected to the server though and the simulation isn't the same. To ensure you get the right results you have to form a private lobby and test there. Easy enough if you have a second pc to load up an alt.

This is you not knowing how server/clients work in modern games. Even single player games in things like unreal engine have a server and a client, both are just local. The purpose of bringing up the academy and training grounds is that you are removing latency/netcode as a culprit for hit reg issues. If you can hit in there then the problem isn't collision detection, it's purely in the netcode. You can also simulate the tick rate of the server by limiting your FPS to 30.

Edited by Quicksilver Aberration, 28 May 2025 - 10:25 AM.

[Meep Meep]

Quicksilver Aberration, on 28 May 2025 - 10:22 AM, said:

This is you not knowing how server/clients work in modern games. Even single player games in things like unreal engine have a server and a client, both are just local. The purpose of bringing up the academy and training grounds is that you are removing latency/netcode as a culprit for hit reg issues. If you can hit in there then the problem isn't collision detection, it's purely in the netcode. You can also simulate the tick rate of the server by limiting your FPS to 30.

Then why do you get odd behavior in the training grounds that doesn't match what you get connected to the live servers? Missiles when pushed over a certain velocity will have glitched pathing and heat mechanics don't jive unless you lock to 30 fps and even then you are only ~simulating~ the live server. If you want 100% reliable feedback on your testing you have to do a private lobby.

[LordNothing]

KursedVixen, on 27 May 2025 - 10:53 PM, said:

decent freespace? like The great war with Shivans,Vasudans and such?

yea, freespace 2 was open sourced, fs1 was ported. that big shivan ship at the end of blue planet, that was mine. i made an ungodly number of ships. go by nuke on the hlp forum.

Edited by LordNothing, 28 May 2025 - 03:26 PM.

[LordNothing]

Meep Meep, on 27 May 2025 - 10:03 PM, said:

Training grounds and academy are not connected to the server though and the simulation isn't the same. To ensure you get the right results you have to form a private lobby and test there. Easy enough if you have a second pc to load up an alt.




Just saying that this game has an extra cheese layer of spaghetti and the cryengine base they used wasn't even fully patched. The devs did wonders though and the game mostly works like it should. Mostly..

cryengine's real problem is that it was built on the assumption that 5ghz processors to be just around the corner. instead moore's law plateaued and we got core spam instead. so it was built for hardware that never existed. by the time it did exist cryengine had other issues. the number of common unhandled exceptions is evidence of that (one of them causes the alt-tab bug).

if they knew mwo was going to persist as long as it did, they should have chosen an engine with an upgrade path. its not unusual now for a game to get one or more engine upgrades (not to be confused with an engine swap) during its support life.

Edited by LordNothing, 28 May 2025 - 04:21 PM.

[Quicksilver Aberration]

Meep Meep, on 28 May 2025 - 11:08 AM, said:

Then why do you get odd behavior in the training grounds that doesn't match what you get connected to the live servers? Missiles when pushed over a certain velocity will have glitched pathing and heat mechanics don't jive unless you lock to 30 fps

Missiles breaking beyond 30 fps is likely to do with tracking interpolation code that is unfortunately coupled to server tick rates which is 30. Without being more in the game engine I can only make some guesstimations but regardless none of that has to do with collision detection. Could there be some wonky behavior? Probably, but the biggest difference between a live server and training grounds (assuming you limit to 30 fps) is the lack of netcode being involved.

To be clear though, private lobbies aren't necessarily 100% the same as a QP server either, especially if you do 1v1s where there's been weird behavior noted. Without knowing how they handle the backend servers I wouldn't be shocked if private lobby servers are separate from QP/EQ/FP servers.

tl;dr people are making wild assumptions based on little knowledge background or otherwise about a very complex system (again, it's not unlike a distributed system which are "fun" to design and run "properly")

LordNothing, on 28 May 2025 - 04:19 PM, said:

cryengine's real problem is that it was built on the assumption that 5ghz processors to be just around the corner. instead moore's law plateaued and we got core spam instead. so it was built for hardware that never existed. by the time it did exist cryengine had other issues. the number of common unhandled exceptions is evidence of that (one of them causes the alt-tab bug).

if they knew mwo was going to persist as long as it did, they should have chosen an engine with an upgrade path. its not unusual now for a game to get one or more engine upgrades (not to be confused with an engine swap) during its support life.

I mean I'm pretty sure the decision around cryengine was a pretty shallow decision, on one hand it was considered the pretty engine at the time and on the other the license fees were lower than unreal as well. With soft costs like developer skills, potential tech debt of the engine itself, etc hard to quantify at that point in time I'm not entirely shocked they went with what they did but the market already seemed to be leaning more towards unreal and they just made a choice that would later bite the. Happens way more often than engineers would like.

I thought they hired some MWLL devs as well but not 100% sure on that, that's a decade ago at this point.

Edited by Quicksilver Aberration, 29 May 2025 - 04:06 PM.

[LordNothing]

i always figured it was because mwll was doing well on cryengine 2. at the time cryengine was still a pretty impressive engine. but im sure there were other factors. idk what posessed them to modify an odd mutant fork with no upgrade path though. keeping it stock would have made it easier to port to the next version.