127 replies to this topic

[Kraftwerkedup]

You realize half your missiles clipped the terrain right?

Its still PGIs fault. But the explanation is clear, and its not hitreg.

[CDLord HHGD]

Christof Romulus, on 14 October 2015 - 11:12 AM, said:

You assume hitreg works in Narnia? =P

LOL, yeah.

Tumnus isn't a mech so I'm assuming the splash damage from missiles on a soft target would hurt. A lot. Probably killed Beaver too... And that pesky fox... LONG LIVE THE QUEEN!

Edited by cdlord, 14 October 2015 - 11:15 AM.

[Kristian Radoulov]

Here's the video from last night. The first salvo seems to register well enough, but the second one half the missiles clearly detonate on the Firestarter's arm and ST, but nothing registers.

Edited by Kristian Radoulov, 14 October 2015 - 11:17 AM.

[Kraftwerkedup]

Grey line is where terrain is visually. Red line is where the "environmental hitbox" is.

Case solved.



Play with your terrain quality in options then go to the testing grounds and find which setting is the closest to the actual hitboxes when you fire near corners. I cant remember what the exact setting is (its not highest or lowest or medium, high but not very high is I think the correct setting to minimize this crap)

Edited by Kraftwerkedup, 14 October 2015 - 11:21 AM.

[Sarlic]

I disagree.

Look at the first page. My second post:

http://mwomercs.com/...ost__p__4762897

Same location in day light.

If that would the case it you should have seen the very same enviromental hitbox. Yet it doesnt. I even painted it out. Look at the crosshair. They're the same position as i hit that Adder. Twist=209 and vertical angle is roughly the same.

[Kraftwerkedup]

Heres your right side 2xSRM4s hitting the terrain:



That white flash is yo missiles exploding. Exploding above the actual visible terrain, because it hit the invisible terrain clip box.

I know it sucks. It shouldnt be like that. But its not a hitreg issue in this case, its a terrain clipping/art issue.

Also the LOD shifts when youre on a live server, that wont happen ever in the testing grounds unless your GPU is terrible, or your settings are set to high.

When youre in a MP game on the live server, and your frame rate drops, even close in terrain will LOD shift.

Thats why you cant recreate it on the Testing Grounds.

Edited by Kraftwerkedup, 14 October 2015 - 11:32 AM.

[Kraftwerkedup]

Here you can see where your missiles hit a little more clearly, the left side missiles have hit the Adder, causing the larger explosion sprite to play.

8 missiles hit the terrain, which causes the small explosion sprite to play.

What is throwing you off is where you watch all the missiles clear the terrain, then the right side ones "warp" to the clipping problem area, and then explode.

Also you might not notice that the sprites are different sizes depending on if you hit a building, mech, or terrain. (dont ask me why).

It doesnt help that the visual flight path of the missiles, is not the same as the ray trace used to apply damage. Thats why the missiles appear to "warp" from being free of the terrain, to hitting the terrain meters to the right of their flight path. The visuals and the "invisible missile entity" arent conjoined.

Nothing has changed internally with the SRM code since spiral missile days. They just have new visuals. The "hit boxes" of the missiles (not accurate but the best way to explain it) remain unchanged.

[Spleenslitta]

I read Paigans post #10 and something occured to me. If i use only a single SRM6 and very few others use missiles/LBX.
Does that mean i will get all 6 SRM's to hit properly or can it only handle 2 missiles at the time?

In short how many missiles can you shoot and still expect them all to register as hits as long as nobody else is constantly bombarding with LRM's? Dumb question i know.
But i really need to know because i got loads of builds that use only a single SRM6/4 launcher along with a bunch of other weapons.
I'm a weirdo who likes mixed weapon loadouts you see.

[PurpleNinja]

My ping is around 200ms, when firing SRMs, some of then explodes after the target is dead or gone.
I always thought this was some HSR double check because of the high ping.

[Sarlic]

Kraftwerkedup, on 14 October 2015 - 11:42 AM, said:

Here you can see where your missiles hit a little more clearly, the left side missiles have hit the Adder, causing the larger explosion sprite to play.

8 missiles hit the terrain, which causes the small explosion sprite to play.

What is throwing you off is where you watch all the missiles clear the terrain, then the right side ones "warp" to the clipping problem area, and then explode.

Also you might not notice that the sprites are different sizes depending on if you hit a building, mech, or terrain. (dont ask me why).

It doesnt help that the visual flight path of the missiles, is not the same as the ray trace used to apply damage. Thats why the missiles appear to "warp" from being free of the terrain, to hitting the terrain meters to the right of their flight path. The visuals and the "invisible missile entity" arent conjoined.

Nothing has changed internally with the SRM code since spiral missile days. They just have new visuals. The "hit boxes" of the missiles (not accurate but the best way to explain it) remain unchanged.

But that's one of the exact reason i am also theorying about, Kraft.

Its not only hitreg. The loss still looks like to be minimal.

Here's your theory: Look carefully at my position, it's practically the same. The 2 difference doenst matter, the initial loss would be still minimal.

The only thing that could make a slight different is that i got my crosshair a little bit aimed down because i naturally follow the targets movement.



As shown in this picture:



Do you notice that some of the missiles dissappear? Atleast it looks like it: as shown below. Only roughly one got hit. The only reason i can think of is the slight arm movement to the right could be one of the issues hitting a few more missiles the terrain.


Fill in the rest: video illustrating it below: (watch it to the very end of SRMs salvo-ing. Even in night vision)


My opinion: lost=minimal. It looks bigger because of the night vision.

All debatable if terrain got inflicted, but as shown above the lost would be still minimal at the pointed target. Atleast in my opinion.

We all knows HPG has terrible hitboxes. I wouldnt be suprised if Caustic valley has it over the place. Still doenst remove the fact SRMs are still in a terrible state.

Good discussion folks.

Thanks for your support.

Edited by Sarlic, 14 October 2015 - 12:50 PM.

[Mazzyplz]

terrain also doesn't explain why the missiles clip through and explode behind

[Kraftwerkedup]

Sarlic, on 14 October 2015 - 12:33 PM, said:

I love that you do this much work. If everyone else did this....geez what a world we'd have.

See here youre just reading the picture wrong, you can see it in the gif of the the firing. Youre not losing 2 missiles, youre losing 7.

You should have 16 right? And you can count them all across the first picture. However there are 2 that are doubled up in the screen shot. The right 7, hit the mountain, even though it shows a small explosion sprite (which it does for enviro hits), its actually grabbed 7 of your missiles, as you can count the 9 remaining still in the air.

In the original, you lose all 8.

Mazzyplz, on 14 October 2015 - 12:35 PM, said:

terrain also doesn't explain why the missiles clip through and explode behind

It does. Because the missile is "dead" as far as the hitscan code goes, but your client is still displaying the missile in flight. Theres no correction code, just like theres only the basic Cryengine prediction code.

All these can happen:

Missile hits code wise, graphical missile still flying.

Graphical missile hits something, still flying code wise.

Missile hits code wise, graphical missile changes to sprite mid air.

Graphical missile hits something, hits something code wise, doesnt play a sprite.

Graphical missile hits something, hits something code wise, does play a sprite.

Graphical missile hits something, doesnt hit something code wise, does play a sprite

Edited by Kraftwerkedup, 14 October 2015 - 01:18 PM.

[Sarlic]

Kristian Radoulov, on 14 October 2015 - 11:16 AM, said:

Here's the video from last night. The first salvo seems to register well enough, but the second one half the missiles clearly detonate on the Firestarter's arm and ST, but nothing registers.

First salvo is as you described a full hit with a few loss on the left side as far i see it..but hey i am no expert.
Some of them goes through the small gap at the shoulder joint. Yes it's possible, i had it plenty of times with the ACH. That's why the first one is messy!

Look carefully at your slowmotion: some are going through the gap where i am talking about.



The second one should have atleast taken his arm, yet it didnt.


Good stuff.

Thanks for your video!

Edited by Sarlic, 14 October 2015 - 01:10 PM.

[Kraftwerkedup]

Of course SRMs ARE in a bad place.

I just dont believe you "captured the ghost on film" here. It IS displaying a serious flaw with MWO, and thats on PGI, but as far as hitreg goes, that appears to have worked.

SRMs dont need hitreg help, they need other help. McGral has some fantastic ideas for SRMs that I support.

And of course better netcode, predictor and correction code, better handling of packets, will help everything. Not just SRMs, but general quality of life of playing MWO.

I agree with what youre saying 100%. I just think the explanation here is the problem with whats happening thats invisible (hitscan, hitreg, clipping, enviro hitboxes) isnt matching up with whats happening visually (missiles in flight, explosion sprites, etc). I dont think where the invisible part, the backend, the code, sees your missile hit, and plays a sprite, is where your missile actually "hits" if you get what im saying. Like the missiles that go past the terrain, then vanish, because they actually code wise, hit the invisible wall.

Not that its malfunctioning, that its simply not matching up because theres been so many changes to graphical content without the backend work being done because...."reasons at PGI".

Weve all experienced it. We've all shot mechs in the open for no damage, shot around a corner and watched lasers hit invisible walls, be caught on them, warped through them, warped through other mechs, shot mechs behind other mechs, shot mechs in the chest and hit them in the back. Theres problems assuredly.

Everytime we catch anything this definitively, its good.

Edited by Kraftwerkedup, 14 October 2015 - 01:15 PM.

[Sarlic]

Quote

I love that you do this much work. If everyone else did this....geez what a world we'd have.

It's a discussion for a reason, no?

Quote

See here youre just reading the picture wrong, you can see it in the gif of the the firing. Youre not losing 2 missiles, youre losing 7.

Hmm.. i dont think so. Look carefully again at the first salvo: they disppear. The terrain only takes two.

It's the first salvo of that video.

I still count 14 missiles on 0.01 seconds by your number.

Edited by Sarlic, 14 October 2015 - 01:16 PM.

[Mazzyplz]

Quote

It does. Because the missile is "dead" as far as the hitscan code goes,

alright but that doesn't explain why this seems to happen a lot more to SRM than other weapons; i can count with the fingers of one hand the times i have seen erppc going through a mech from that range

terrain being nearby or not.


ALSO - in the video sarlic posted; his orion is moving slowly or barely moving.
and his ping was good according to him

so if the terrain was in fact eating his missiles - it woulda shown clearly in his own client the terrain eating the srms. the mech that was moving fast was the target, not the shooter.

if you yourself are moving slowly that means the desync from your own machine to the server isn't huge, your target might be rubberbanding due to it's speed but an orion coming out of a 180 turn?

Edited by Mazzyplz, 14 October 2015 - 01:32 PM.

[Kraftwerkedup]

Sarlic, on 14 October 2015 - 01:16 PM, said:

It's a discussion for a reason, no?



Hmm.. i dont think so. Look carefully again at the first salvo: they disppear. The terrain only takes two.

It's the first salvo of that video.

I still count 14 missiles on 0.01 seconds by your number.

Fair enough.

I count all 16 in the first picture, on the right middle two are on top of another one.

In this picture I count 9 still in flight:



Thats just subjective unless we had logs unfortunately.

I know that 7 missiles can look like 2 small explosion sprites, so thats nothing to go on, and really even given how the game is coded (used to be able to open all this stuff n cry editor and load it up on the test server *sigh*) even if there were 9 models and 9 engine sprites, that doesnt mean 14 missiles didnt make it past the terrain code wise, just not visually.

Super frustrating I know.

Id just like to say thanks for doing this, at least I appreciate it. This is how discussions are supposed to work. Evidence, debate, conclusions...rare in these parts.

Edited by Kraftwerkedup, 14 October 2015 - 01:22 PM.

[Sarlic]

If you don't mind i am going to grab a beer and get drunk for the night.

I think we both agree on the fact that SRMs (and SSRMs) need work badly. Either a fix, or weapon value adjustments.

Thank you too!

Much appreciated for the support and the discussion folks.

Edited by Sarlic, 14 October 2015 - 01:25 PM.

[Alistair Winter]

Tried to pilot my good old GRF-2N today. Previously my best mech, statwise. 2SRM6, 2SRM4, 2ML.

What a piece of junk that mech is today.

Caught an enemy Enforcer standing still at 100 meter range, his back turned to me. Standing still, shouldn't be any issues. Full alpha strike at his back. Didn't even manage to get past his rear armour.

ggclose. Back to laservomit.

[Kmieciu]

I don't have access to current XML files right now, but SRMs used to have a 5 cm (0,05meter) explosion radius. ( radius="0.05") Not the visual explosion - the actual damage sphere has 5 centimeters.

Damage spheres made sense back when they were 5 meter large. You could hit the ground and still take out both legs on a light mech.But since the "splash damage" nerf, MWO simulates 5-centimeter explosions and that has to take a lot of server resources for nothing.

In my opinion it's high time PGI scrapped the explosion code and used ballistic code for SRM hit reg. Hell, they could even use "volleydelay", like Clan UACs if it improves hitreg. I would take "streaming" SRMs over "not-registering" SRMs.
Once we get them to work reliably, they can be fine-tuned using projectile speed, spread, ammo count and damage.

Edited by Kmieciu, 15 October 2015 - 02:05 AM.