107 replies to this topic

[Nightbird]

Agreed, Arrow IV and Thumper would have to be new art assets due to size and type of fire. Rest can use existing assets. But... I would still put odds of new hardware at close to 0 because of other concerns like timeline and balance.

Edited by Nightbird, 16 October 2020 - 07:41 AM.

[Nightbird]

Added #6

[LordNothing]

maybe its time for a bit of a jihad.

would you like to buy a wob mech pack?

not exactly low cost.

who am i kidding pgi couldn't even move the rifleman iic.

Edited by LordNothing, 16 October 2020 - 12:50 PM.

[Nightbird]

I would love a time skip with more mechs and tech.

[Cluster Fox]

Belorion, on 14 October 2020 - 05:13 AM, said:

If the methodology is sound yes, but in this case you are comparing the WRL ration with the WRL ration saying that it means that the WRL ratio is a definitive predictor of WRL. You are not allowing for bias in the system itself and only looking for trends.

Sorry I'm a page late.
Belorion, you have to see the MM and PSR as a Control Loop more than a collection of data and statistics.

That's because the stats change dynamically every game and are affected by what you do between input and output.
Sure in stats you can't really correllate a variable to itself, but in control loops it's totally acceptable. Think thermostat... a device that controls the temperature by measuring the temperature.

Edited by Cluster Fox, 16 October 2020 - 05:47 PM.

[Nightbird]

If that explanation works... lol, I'm not sure whether to face-palm or high-five

[LordNothing]

would explain why the wlr graph has a sharp spike at 1 and is not a bell distribution. there is probably some wlr steering going on. also would explain why there are long period oscillations between winning and losing streaks.

Edited by LordNothing, 17 October 2020 - 04:46 PM.

[Nightbird]

They're all "bell distributions" with different standard deviations.

[The Lighthouse]

Have to say only "XML shake-up" is true low-cost game update. I mean, it is not ven "low-cost game update" but more likely "only possible game update" because anything that needs some engineering is probably impossible now for PGI.


I have a strong suspiction that PGI may not even have a single technical staff left for MWO. Especially Tina left means PGI was not willing to spend dime on even basic PR personal for MWO. They probably bet everyone on MW5.


And here comes the fallout.

[Davegt27]

#7 keep doing what your doing, that is keep going with the generous events --its working or was working

the more people that play the better chance they will spend money

[MTier Slayed Up]

Just patch the invisible walls. That's about as low budget as you can get.

It doesn't require new models or mechs. No new content. It's the simplest form of polish any gamer would want in an 8 year old game.

I don't care if it's been around for x amount of years. That's not a good excuse. That's like when my kid doesn't make his bed for 8 years straight and he just says "Well I haven't done it now, why change?!"

Just do it you little puke and quit prancing around it.

[Nightbird]

Sure, as a collaboration between the community for creating a list of invisible walls and PGI knocking them out, it could be a QoL fix. Only for invisible walls visible at short range though, at long range it's cryengine settings.

[The pessimistic optimist]

Nightbird, on 18 October 2020 - 10:22 AM, said:

Sure, as a collaboration between the community for creating a list of invisible walls and PGI knocking them out, it could be a QoL fix. Only for invisible walls visible at short range though, at long range it's cryengine settings.

Grrrrrrr invisible walls hate missing shots because of those

[LordNothing]

invisible walls are either due lod glitches (goes away with higher detail settings), or to bad collision boxes in high detail terrain, in other words a mappers problem. id hate to hear that they no longer have mappers. some would even leave because of that.

Edited by LordNothing, 18 October 2020 - 04:00 PM.

[Nightbird]

I don't know why they don't have a tool that just fits collision boxes over visible art assets. It's apparently all manual work. As for the long distance terrain pop-in, it can be fixed by one of the lod settings, was it e_lod_ratio I forget.

[MW Waldorf Statler]

Nightbird, on 18 October 2020 - 07:43 PM, said:

I don't know why they don't have a tool that just fits collision boxes over visible art assets. It's apparently all manual work. As for the long distance terrain pop-in, it can be fixed by one of the lod settings, was it e_lod_ratio I forget.

you must build each object from scratch and each Collision box ..all handcrafted...a KI have today not the creativity to create good Looking Assets or paint a Mona Lisa and build a frame for it

[VonBruinwald]

Nightbird, on 18 October 2020 - 07:43 PM, said:

I don't know why they don't have a tool that just fits collision boxes over visible art assets. It's apparently all manual work. As for the long distance terrain pop-in, it can be fixed by one of the lod settings, was it e_lod_ratio I forget.

It's a square peg round hole deal.

Collison boxes have a much much lower poly-count than the visible model. Computer generated hitboxes tend to have one of three modes, outer model, inner model, "best fit".

• Outer edge - the hit box surrounds the extremities of the model - Results is invisible walls.
  
• Inner edge - the hit box fills the inner edges - this results in visible walls you can run and shoot through.
  
• Best fit is a combination of the two, tends to give a "better" fit but results in a combination of the two. These often require manual tweaking as the computer often has no concept of what parts need to be covered by the hit boxes and what parts don't.

From a technical perspective it's easier to make the art fit the hitboxes but these days consumers tend to be more concerned with visual fidelity than technical performance.

[LordNothing]

Nightbird, on 18 October 2020 - 07:43 PM, said:

I don't know why they don't have a tool that just fits collision boxes over visible art assets. It's apparently all manual work. As for the long distance terrain pop-in, it can be fixed by one of the lod settings, was it e_lod_ratio I forget.

the low detail boxes should only be used in initial cull passes and not for final collision pass.

i figure they just skippped the detail pass all together because pgi never puts on a second coat of paint on anything.

thats a gross oversimplification, im leaving out the complexity of collision trees and the various algorithms used to check if two objects collide (and you always need a few). or how you work your way down from a lot of aabbs spheres and boxes down to a small number of convex hulls and eventually actual polygons. some of this stuff has to be made by the artist, and its just a low poly abstraction of something complex. how far down the rabbit hole you go depends on a lot of considerations, but i figure pgi came to a point and said 'good enough'.

in all my days of modding i never found an automatic collision mesh generator that worked at a satisfactory level. i always resorted to copying the geometry and start chopping polies. of course that was some years ago.

Edited by LordNothing, 19 October 2020 - 03:26 PM.

[dario03]

I think fixing the hitbox around Omega should be the first invisible wall fix. Its like a giant cube way bigger than the actual gun. Very noticeable at times like if you push in on Hellebore from the north to d2/d3 and are shooting at enemies on the other side of the gun.

[Cluster Fox]

https://youtu.be/kg6nuaJgf6Y?t=597

Cry engine collision geometry, posted elsewhere here... can't remember.
Notice how the collision geometry has to be simplifed and is then scaled up to avoid clipping. Explains a lot.