"In case anyone might find it useful...", Miscellaneous NS research, copied from my CAL forum thread Options

[a_civilian]

From this CAL forum thread, copied here because the CAL forums are viewable by members only. Edit: The CAL-NS forum no longer exists. I've also since added a number of entries.

Delays between attacks (inverse of rate of fire) at 100 fps, in seconds:

CODE

Light Machinegun: .060
Shotgun: .66
Heavy Machinegun: .060
Grenade Launcher: .61
Pistol (cap): .10
Mines: .50
Knife: .33

Skulk Bite: .40
Parasite: .25
Leap: .50

Spit: .40
Heal Spray: .80
Bile Bomb: .75
Web: .25

Lerk Bite: .32
Spores: 1.00
Umbra: .50

Swipe: .48
Metabolize: 1.45
Acid Rocket: .50

Gore: .45
Devour: 1.00
Stomp: .75

Should be accurate to within �1 to the last significant digit (�0.001 seconds for LMG and HMG, �0.01 seconds for all else). Also, note that there is no delay before the first attack.

Reload times, in seconds:

CODE

LMG: 3.00
Pistol: 3.00, 3.10 (magazine not empty, magazine empty)
Shotgun: 1.10+.47n
HMG: 6.30
GL: 2.20+1.10n

Should be accurate to within �0.01 seconds.

Both were measured using the wait command and the fact that at 100 fps (constant), each such command waits 0.01 seconds.

Weapon draw times, in seconds:

CODE

LMG: .40
Shotgun: .90
HMG: .60
GL: .30
Pistol: .35
Knife: .25
Welder: .55
Mines: .30
Grenade: .30

Skulk Bite: .10
Parasite: .10
Leap: ~0 (<.02)
Xenocide: .10

Spit: .10
Healspray: .10
Bilebomb: .10
Web: 1.00

Lerk Bite: .20
Spores: .60
Umbra: .10
Primal Scream: .60

Swipe: .10
Blink: .10
Metabolize: .10
Acid Rocket: .10

Gore: .50
Devour: .60
Stomp: .60
Charge: .60

Accurate to within �0.01 seconds.

Fall damage:

Fall damage is taken past 580 units/second, and is modeled by:

damage = 0.225v - 130

(vertical component of the velocity; downward is taken to be positive)

Leap speed:

Leap adds 500 units of speed in the direction you're aiming.

If you're on the ground, it also jumps, adding vertical speed.

Marine bunnyhopping:

A marine can bunnyhop on landing if the vertical component of his velocity is less than his base speed.

Hence the height difference h needed to bunnyhop at base speed v, jump speed* s, and sv_gravity g is given by:
h = (s^2 - v^2)/(2g)

And conversely, the base speed v needed to bunnyhop on a height difference h with jump speed* s and sv_gravity g is given by:
v = sqrt(s^2 - 2gh)

*see entry on jump speed below

Dependence of movement speed on framerate:

For the purposes of movement, the unit time varies with framerate. This causes variation in movement speed that does not show up on cl_showspeed, and is the reason for drastically lengthened jump durations at certain very high framerates.

Here is a plot of the factor by which the unit time is lengthened, against framerate. Notable troughs are labeled (major troughs in red, minor troughs in orange); these are desirable framerates as they maximize speed (lower time multiplier is better).

(Major troughs, in text form, occur at framerates 20, 25, 27, 37, 40, 50, 62, 71, 83, 100, 111, 125, 142, 166; minor troughs at 30-33, 43, 45, 47, 52, 55, 58, 66, 76, 90)

Dependence of attack rate on framerate:

The dependence of attack rate on framerate appears defined by the statement that there must be a constant integer number of frames between attacks at any framerate. The measured numbers of frames between attacks for each framerate value are shown below for selected weapons:

CODE

LMG and HMG:
framerate    frames/attack
20        1
21-38        2
39-58        3
59-76        4
77-90        5
91-111        6
112-125        7
126-142        8
143-166        9
167-?        10

pistol (cap):
framerate    frames/attack
20-29        2
30-40        3
41-47        4
48-58        5
59-66        6
67-76        7
77-83        8
84-100        9
101        10
102-111        11
112-125        12
126-142        14
143-166        16
167-?        19

shotgun:
framerate    frames/attack
20-21        14
22        15
23-24        16
25        17
26-27        18
... (I interpolate insignificant attack rate variation)
91-95        65
96-100        66
101        69
102-111        73
112-125        82
126-142        93
143-150        108
151-166        109
167-?        127

bitegun:
framerate    frames/attack
20            8
21-22        9
23-25        10
26-27        11
28-29        12
30-32        13
... (as previously)
63-66        27
67-71        29
72-76        31
77-83        34
84-90        37
91-99        40
100        41
101-102        44
103-111        45
112-125        50
126-142        57
143-166        67

Obviously, the attack rate is maximized by setting the framerate to the highest value that preserves any given number of frames per attack (that is, the highest value in one of the given ranges).

The attack rate (in attacks/second) can be determined by dividing the framerate by the number of frames per attack.

(the CAL forums formatted this last one nicely :( )


Addition, 3/1/07: Blink
----------------
Each attack of blink adds 225 units of speed. At 100 fps, blink fires every 0.06 seconds (6 frames). (This is in 3.2.)

Thus it takes 0.19 seconds* to reach maximum non-celerity horizontal speed (aiming between 0 and 36 degrees from the horizon) and 0.25 seconds to reach maximum celerity horizontal speed (aiming between 0 and 32 degrees from the horizon).

* remember that the first attack fires instantly - that is, in one frame.

Blinking while standing on the ground also automatically jumps for an additional 268 units of speed upward.
----------------

Addition 2, 3/1/07: Jetpack
----------------
Jetpack vertical acceleration is modeled at 100 fps by a = 7.16s - 324 (in units/second^2), where s is the base run speed. (Note: This does not include the effect of gravity; to account for that, subtract 800.)

The dependence of this on framerate, as with gravity, is as the inverse of the time multiplier with respect to speed or the inverse square of the time multiplier with respect to distance.

If a directional command is held while using the jetpack, 12 units of speed are added to the horizontal velocity per frame, regardless of the framerate.

The jetpack's vertical and horizontal thrusts are independent.
----------------

Addition, 3/9/07: Friction
----------------
At speeds above 100 units/second, the acceleration due to friction is -sv_friction*v (in units/second^2), where v is the horizontal velocity.

At nonzero speeds below 100 units/second, the acceleration due to friction is a constant 100*sv_friction units/second^2 opposite the direction of motion.
----------------

Addition 2, 3/9/07: Surface acceleration
----------------
The ground acceleration of every class is (base run speed)/(0.1 seconds). (Note this does not include the effects of friction.)
----------------

Addition, 5/21/07: Jump speed
----------------
Jump speed for every class is 268 units/second. However, HL seems to apply changes in velocity before changes in position, so one frame's worth of gravity is subtracted from the speed before you even leave the ground.

Therefore the correct formula for jump speed v is

v = 268 - sv_gravity/framerate
----------------

This post has been edited by a_civilian: Sep 26 2009, 06:24 AM

[Comprox]

This was next on my list of things to figure out. You get a hero sticker sir!

[a_civilian]

Moved to general discussion as I figured there was no reason to keep this thread available only to PTs.

[JazzX]

Moved to general discussion as I figured there was no reason to keep this thread available only to PTs.

Yeah and this way everyone gets a taste of how shockingly exciting PT Forum threads get.

[a_civilian]

Are you implying that's not exciting!?

[Trayder]

That's pretty interesting, does the command cl_quickselecttime impact on any of these values? Most likely would be drawtime I'd imagine.

[MrMojo]

wow talk about effort

[todd1Ok]

So, way i read this, higher framerates would cause you to fire slower?

also, what about framerates in the high 500's? i dont mean to brag, but i regularly get 400+ fps on combat maps and 300+ on classic. how would this impact my play?

[a_civilian]

So, way i read this, higher framerates would cause you to fire slower?

It actually looks something like this.

also, what about framerates in the high 500's? i dont mean to brag, but i regularly get 400+ fps on combat maps and 300+ on classic. how would this impact my play?

I can't answer that, unfortunately, as my computer can't go anywhere near that high and I can't discern any clear pattern from which to extrapolate.

[Nikon]

sooo, basically fps_max 60 would be the most benefical setting for maximizing your shots per second?

[Golden]

Actually, its either 75 or 77, think its 77. One of my clan memebers uses fps_max of 77 while playing. At the start of his jump script he has his fps_max go to I think 200 and back to 77 at the end. This allows him to fire at max speed and his jumps fire faster allowing for less of a speed loss when bunnyhopping.

[a_civilian]

sooo, basically fps_max 60 would be the most benefical setting for maximizing your shots per second?

That's 58. And 20 gives a higher attack rate - see the single point above 20 fps.

Actually, its either 75 or 77, think its 77. One of my clan memebers uses fps_max of 77 while playing. At the start of his jump script he has his fps_max go to I think 200 and back to 77 at the end. This allows him to fire at max speed and his jumps fire faster allowing for less of a speed loss when bunnyhopping.

Are you sure it's not 76? 77, in my measurements, gave a significantly lower attack rate (it is one of the troughs).

---

Edit: similar plot of the pistol's attack rate is here.

This post has been edited by a_civilian: Aug 23 2006, 08:04 PM

[Droggog]

That's quite interesting. Thanks for the analysis.

So basically, a good compromise for having decent fps, decent bhopping (time multiplier) and lmg rof would be 111 fps?

Hmmm, apparently i have to play in "developer 1" or i seem to be capped at 100 fps even with v-sync off. And HL seems to reset to dev 0 randomly when i play?! May need to bind a key to re-reset it to 1 easily while playing.

Another question; when i set fps_max to 111, i'm capped at 110 fps (or so cl_showfps tells me). Setting fps_max to 112 seems to do the trick (capped at 111 then), is that the exact setting i have to use? Or do i have to set it at 111 regardless of what cl_showfps and r_speeds 1 tells me?

H4x!

[Shockwave]

Droggog - It's worth noting (afaik) that developer 1 does not give the FPS it claims. As a Developer tool, it shows your potential FPS you would be getting, were the engine to bother rendering that many frames. Of course, I'm sure one of the Developer team could elaborate more, I'm not an expert by any stretch of the imagination.

Interesting research though, and certainly worth (if a bit dry) reading.

- Shockwave

[Golden]

That was my theory too Shockwave, at least until some friends of mine demonstrated that FPS above 100 does do some interesting things... such as jetpacking along the ground with 400 FPS allows you to get up to 1300 ground speed.

[a_civilian]

A framerate above 100 fps definitely affects ingame timings; whether the video device actually renders more than 100 frames in a second is what is uncertain (and usually the answer I hear is indeed "no").

This post has been edited by a_civilian: Sep 7 2006, 12:40 AM

[Hellabeans]

Yes, jetpacking with 300 fps for some reason gives you more horizontal speed and much, much, less veritcal lift, that's why it is easier to obtain high speeds when jetpacking with above 100 fps

Try it yourself, try jetpacking with 100 fps then with 200fps+, and you'll notice a different behavior with the jatpack

[Droggog]

Yeah, curious, i played a bit with this, and i clearly noticed some changes at 111 fps compared to 100 fps (about attack rates, no jetpack tests so far);

With the knife, or skulk bite, i clearly felt a negative effect (it's a bit slower, it seems). About the lmg rof, honestly, i felt no change. It probably does, but compared to 100 fps it's so minimal it feels the same to me. Ideally, we would need a graph for each weapon so we could load a .cfg for each class (marine, skulk, lerk, fade, etc...) with optimal settings, for testing purpose.

About bhopping, the only thing i noticed is that i may need to add one more "wait" in my script, nothing else

Then again, i'm judging from what i see client-side. How the server interpret this i've no idea.

This post has been edited by Droggog: Sep 7 2006, 08:02 PM

[Jabba_The_Hunt]

also, what about framerates in the high 500's? i dont mean to brag, but i regularly get 400+ fps on combat maps and 300+ on classic. how would this impact my play?

That is not at all unusual considering the age of the half life engine, one thing you should note is that you can not see more frames than your monitors refresh rate, some people believe having a higher frame rate still makes gameplay seem "smoother" even when your monitor isn't displaying the frames and I would suggest its the above type of effects which lead to this.

[Alcapwn]

I thought this stuff was fixed in the 3.0final/3.1 patch? (I remember the dev team talking about fixing it)


Or is it just another HL1 engine quirk?