639 replies to this topic

[Morang]

Koniving, on 07 September 2013 - 09:12 AM, said:

So if you're 800,000 polygons, the network's hitbox is 100,000 polygons.

Far as grass goes, with a larger environment, you'd have to have more panels of grass. For a smaller environment but giving the gigantic illusion, you'd have to shorten the length of the grass but those same size panels would cover more.

Think of it this way. If your panels cover 2 meters, tax you 400 polygons, and you have 2000 meters to cover.. That's 400,000 polygons just in grass and 1,000 panels.

But if you scaled it to be 1 meter = 5 meters, then 2000 meters = 400 meters, then 2 real meters covers 10 fake meters, then you need 40 panels and 16,000. I'm sure I botched that math somewhere but I have to take a shower for work. Ask SCJazz about his slowing time idea, but be ready to spend hours listening to him loop back and forth.

I once heard (back in Closed Beta I believe) that PGI used some generic CryEngine stuff, some models that come for free with the SDK itself. Perhaps the cause of mechs having "realistic" size in game is that PGI wanted to use some of the SDK assets without rescaling them? I'm not into 3D modelling myself, so don't know how true can it be. As far as I understood you, levels of detail are coded so in the engine that you can't just have a 25 meters shadow, but make it 5 times less detailed than 5 meters shadow and expect it to put same strain on the hardware... is it so?

Still I don't understand how can we expect better hitbox fidelity if it depends not on the model size, but on polygon count. Whatever are the sizes of the polygons, hitbox will still have fewer polygons than visual model, and so there will be a possibility of mismatch, and this mismatch will be the same relative to the model size.

Time compression is very strange idea. I don't see how it can do anything but confusing people. We can alter sizes and keep the time scale and change speeds proportionally to sizes (i. e. be it 10 meters Commando running at 30 m/s or 1 meter Commando running at 3 m/s, they both cover 3 x their height per second), but slowing time sounds like cheating. What is it, our speedometer shows us that Commando runs at 30 m/s (108 kph), while it only really runs at 22.5 m/s and covers 2.25 x his height per second? It can be easily disclosed by in-game testing.

[Koniving]

Morang, on 07 September 2013 - 01:34 PM, said:

I once heard (back in Closed Beta I believe) that PGI used some generic CryEngine stuff, some models that come for free with the SDK itself. Perhaps the cause of mechs having "realistic" size in game is that PGI wanted to use some of the SDK assets without rescaling them? I'm not into 3D modelling myself, so don't know how true can it be. As far as I understood you, levels of detail are coded so in the engine that you can't just have a 25 meters shadow, but make it 5 times less detailed than 5 meters shadow and expect it to put same strain on the hardware... is it so?

Still I don't understand how can we expect better hitbox fidelity if it depends not on the model size, but on polygon count. Whatever are the sizes of the polygons, hitbox will still have fewer polygons than visual model, and so there will be a possibility of mismatch, and this mismatch will be the same relative to the model size.

Time compression is very strange idea. I don't see how it can do anything but confusing people. We can alter sizes and keep the time scale and change speeds proportionally to sizes (i. e. be it 10 meters Commando running at 30 m/s or 1 meter Commando running at 3 m/s, they both cover 3 x their height per second), but slowing time sounds like cheating. What is it, our speedometer shows us that Commando runs at 30 m/s (108 kph), while it only really runs at 22.5 m/s and covers 2.25 x his height per second? It can be easily disclosed by in-game testing.

While it depends on polygons, consider the background. Imagine how many meters tall those pillars on Tourmaline are. They're often in excess of 200 meters. Cut the polygons down by 8 on those, and they become far simpler, far oversized, and when you shoot 5 meters away from the pillar wall -- you've hit the server-side pillar. Notice there's nothing actually there?

[Here5y]

I think I will wait a little bit longer before i buy my first light mech.

[Koniving]

Morang, on 07 September 2013 - 01:34 PM, said:

I once heard (back in Closed Beta I believe) that PGI used some generic CryEngine stuff, some models that come for free with the SDK itself. Perhaps the cause of mechs having "realistic" size in game is that PGI wanted to use some of the SDK assets without rescaling them? I'm not into 3D modelling myself, so don't know how true can it be. As far as I understood you, levels of detail are coded so in the engine that you can't just have a 25 meters shadow, but make it 5 times less detailed than 5 meters shadow and expect it to put same strain on the hardware... is it so?

Still I don't understand how can we expect better hitbox fidelity if it depends not on the model size, but on polygon count. Whatever are the sizes of the polygons, hitbox will still have fewer polygons than visual model, and so there will be a possibility of mismatch, and this mismatch will be the same relative to the model size.

Time compression is very strange idea. I don't see how it can do anything but confusing people. We can alter sizes and keep the time scale and change speeds proportionally to sizes (i. e. be it 10 meters Commando running at 30 m/s or 1 meter Commando running at 3 m/s, they both cover 3 x their height per second), but slowing time sounds like cheating. What is it, our speedometer shows us that Commando runs at 30 m/s (108 kph), while it only really runs at 22.5 m/s and covers 2.25 x his height per second? It can be easily disclosed by in-game testing.

Not sure about the the m/s thing. But time compression would affect everything. Including how fast your crosshair moves. In truth with the CryEngine the network can be set to have an Atlas moving "50 kph" move as fast as a Jenner does at 150 (and you don't wanna imagine how fast that Jenner flies). Then again, they can reduce the speed of the simulation so that it takes 7 real seconds for an AC/20 round to hit you in the face, and even with all that reaction time to think, every aspect of your mech is also time-compressed. Looking, reticule tracking, torso twisting, acceleration, deceleration, firing, beam time, everything would be slower. Because of this, there's insanely accurate hit detection, with all the time in the world to judge whether or not the hit will occur even if your ping was over 2,000.

But alas, at first you'll "feel" that difference no matter how subtle.

[Fred013]

Curse the Cryengine nerds for derailing the thread from cute buggy adorableness!
Back on topic, I think they could have made it cuter if they kept the walking toaster look. After all, why do people like the Urbanmech? (walking trashcan)

Edited by Fred013, 07 September 2013 - 08:30 PM.

[Dread Raider]

Do people actually like the Urbanmech?

Edited by Ganjhitsu, 07 September 2013 - 08:35 PM.

[Fred013]

Ganjhitsu, on 07 September 2013 - 08:34 PM, said:

Do people actually like the Urbanmech?

Just.... just.... just.....

FACEPALM

[Koniving]

Ganjhitsu, on 07 September 2013 - 08:34 PM, said:

Do people actually like the Urbanmech?

[Jin Ma]

Ganjhitsu, on 07 September 2013 - 08:34 PM, said:

Do people actually like the Urbanmech?

there are people who don't like the urbanmech ?

[Morang]

Koniving, on 07 September 2013 - 07:41 PM, said:

While it depends on polygons, consider the background. Imagine how many meters tall those pillars on Tourmaline are. They're often in excess of 200 meters. Cut the polygons down by 8 on those, and they become far simpler, far oversized, and when you shoot 5 meters away from the pillar wall -- you've hit the server-side pillar. Notice there's nothing actually there?

Well... imagine those pillars being 20 meters high. Cut down polygons by 8, and they become far simpler, and when you shoot 0.5 meters away from visible pillar wall, you still hit collision model. But your mech is now 1 meter high, not 10 meters (like it was with 200 m pillar). So when you hit invisible wall 5 meters away from 200 meters pillar while sitting in a 10 meters mech, it's bad, because it's half the height of your mech. But if you hit invisible wall 0.5 meters away from 20 meters pillar while sitting in a 1 meter mech, it's just the same bad thing. No difference. Still you think that you have exposed half your mech above the ridge, but boom, you lasers hit nothing in front of you. To change anything you have to increase the polygon count of a pillar collision model (you don't need to increase polygon count of the visible model though, as you should just bring the pillar hitbox closer to visible model).

[Morang]

Jin Ma, on 07 September 2013 - 09:30 PM, said:

there are people who don't like the urbanmech ?

I think that if PGI release Urbanmech (engine cap = 60 x 1.4 = 85, max speed with tweak = 50.5 kph), everybody who buys it should have all their other mechs locked and must then proceed to master 3 Urbanmech variants before they can play other mechs again. It will shut the Urbanmech cult forever.

[stjobe]

The Flea, folks. Focus.

[Fred013]

When are we getting the Flea? September? October? March 2014?

Edited by Fred013, 08 September 2013 - 12:46 AM.

[Mechwarrior0311]

Yes, give those Light pilots their Flea and then please release another 100 ton mech of any choosing so the Atlas isn't so lonely on the battlefield!!

[Fred013]

Mechwarrior0311, on 08 September 2013 - 01:42 AM, said:

Yes, give those Light pilots their Flea and then please release another 100 ton mech of any choosing so the Atlas isn't so lonely on the battlefield!!

King Crab maybe for launch, or surprise banshee.

[Baine3FreakofNature3]

This is not a flea. The flea has no arms and is a pod mech. That's a locust..... PGI kills lore once again.

Why? The flea, at 20 tons, is a minimal-complication design. Essentially weapons on legs, it's made with the intent for speed in manufacturing and ease of construction(less acutators/arms, less parts). This is no flea PGI and I hope you can change it's name at least, to the locust. If you're using an existing franchise, at least respect the established lore.

Edited by Baine3FreakofNature3, 08 September 2013 - 02:52 AM.

[Morang]

But it have vertically traverseable "arms" in the TRO art. And it have shoulder and upper arm actuators in the Record Sheet.

[Mechwarrior0311]

Fred013, on 08 September 2013 - 01:54 AM, said:

King Crab maybe for launch, or surprise banshee.

King Crab! King Crab!!

[Hexenhammer]

Mechwarrior0311, on 08 September 2013 - 03:22 AM, said:

King Crab! King Crab!!

Urbanmech!

[Belorion]

Morang, on 07 September 2013 - 10:40 PM, said:

I think that if PGI release Urbanmech (engine cap = 60 x 1.4 = 85, max speed with tweak = 50.5 kph), everybody who buys it should have all their other mechs locked and must then proceed to master 3 Urbanmech variants before they can play other mechs again. It will shut the Urbanmech cult forever.

That's kind of how I play anyway... My kdr other stats would be way better if I played my best mechs more often.