Cruzz's The More You Know

Can we please not fill this wonderful thread with unnecessary troll/rageboi/flame/nonconstructive/negative posts?

I WANT TO LEARN ABOUT THE GAME NOT READ YOUR DAMN WHINY COMMENTS

+1

Can we please not fill this wonderful thread with unnecessary troll/rageboi/flame/nonconstructive/negative posts?

I WANT TO LEARN ABOUT THE GAME NOT READ YOUR DAMN WHINY COMMENTS

If you want to learn something, then have some of this:\

I often see people screaming about squads running out of cover and giving a chase, even some of the more renown streamers make that mistake so this is how you fix it:

If you're fighting more then 1 squads and want to switch targets, right click squad you want to attack and then press S(normal hotkeys) or STOP order, unit will change target, but will not re position or chase the squad then it retreats and will switch to another target after first one goes out of range.

Thanks for your help, again.

I'm in the process of writing something between a bug report and a guide about how to use artillery properly (and why the UI is counter-intuitive).

Therefore, if you have the time, could you review what I have written here about scatter?

http://www.coh2.org/topic/47925/l2aim-the-definitive-guide-to-artillery-in-coh2

I would gladly appreciate if you could point out any inaccuracies/incorrect assumptions etc.

Reticule type (which one to pick from the ui_reticule folder) and sizes are defined at the end of the ability files. They have no link whatsoever to the actual scatter of the weapons being used, the values are entered purely by hand and are mostly completely wrong because they've barely been updated in the years since the game was released.

No scatter affecting effect in this game is ever updated over to the reticules either, even though it's very easy to do scripting wise.

Now, regarding the reticules, could we go through one example to see if I got it right; let's take the Katyusha.

If I got this correctly, the targeting circle values for Katyusha are under fatality_katyusha_rockets, and the radius of the reticule is 20.

According to your section about scatter in the OP, the radius of the reticule should correspond to a distance of ~120, if we ignore distance_scatter_max (the distance_scatter_ratio of the Katyusha is 0.1667).
Show Spoiler

If I understand horizontal scatter well enough, the targeting reticle of the Katyusha should correspond to what happens at range ~229 (well beyond the maximum range of the Katyusha)
Show Spoiler

If I got this correctly, the targeting circle values for Katyusha are under fatality_katyusha_rockets, and the radius of the reticule is 20.

Fatality is for the end of game fatality abilities that hit your opponents base.

abilities\soviet\modal_ability\artillery\artillery_barrage_ability\kaytusha_rocket_truck_barrage_mp
&&
abilities\soviet\modal_ability\artillery\artillery_barrage_ability\kaytusha_rocket_truck_barrage_vet3_mp

The regular barrage has an UI Radius from min 8.33 to 22.5 max (range 50-150)
Weapon stats:
Range min 50 max 150 (200 on weapon because it's also reused for the 200 range vet3 barrage)
Angle 10
Distance max 15
Ratio 0.1667

Angular scatter estimate as in first post: 5 - 15 (50*0.1 150*0.1)
Actual(?) angular scatter: 4.37 - 13 (distance/tan(90-angle/2))
Distance scatter: 8.33 - 15 (50*0.1667 150*0.1667 > 15)

So the Katyusha reticule overestimates angular (horizontal) scatter at all ranges. Error on distance scatter isn't as bad.

While testing this stuff it definitely feels like the explanation for offset in the mod tools is completely wrong. I always expected the hit area to be a half circle at 1 based on that, but instead it seems the offset just moves the entire possible hit area forward (or backward with a negative value). With a value of 1 all targeting is now centered around what used to be the edge of the old scatter area.

Suxton and its 0.3 offset after around 40 rounds or so:
Show Spoiler



Light red circle is ingame UI reticule circle. Light green ellipsoid part is calculated scatter centered around the targeting circle. Blue ellipsoid overlayed on top is an offset moving the entire targeting area forward. Ignoring height differences, no rounds will ever hit in the light green or light red areas, everything will hit in the area overlayed by blue (so dark green+blue+purple areas).

So the take home message is that for artillery barrages its generally better to position the center of the UI slightly "in front" of where you want most of the rockets/shells to land?

Angular scatter estimate as in first post: 5 - 15 (50*0.1 150*0.1)
Actual(?) angular scatter: 4.37 - 13 (distance/tan(90-angle/2))

I am not sure what you mean by the first statement (0.1 stands for what?)

From your second statement, am I right to deduce that you consider the scatter area to be a rectangle?

However, wouldn't you say that the scatter area looks a bit more like this?
Show Spoiler

Your findings on scatter offset, though look both surprising and interesting. I had the same opinion, which was also influenced by the modding tools (and your OP!).

From own, personal very thorough experience with PIATs (scatter_offset: 0.2). It seems that PIAT projectiles barely ever undershoot. Moreover the maximum possible undershoot distance seems by far smaller than the overshoot distances I have experienced.

Perhaps this is influenced a lot by elevation, weapon angle etc?

Would it be possible to test Katyusha vs Panzerwerfer scatter? Both of them have the same scatter_offset (0.1), comparable scatter values. However, the Katyusha seems to fire its rockets at a steeper angle.

Also, do weapons with 0 scatter_offset have the same scatter pattern? It might be worth to try SU-76 (scatter_offset = 0, very steep firing angle).

I am not sure what you mean by the first statement (0.1 stands for what?)

It's just a quick guesstimate like I had in the first post, you can ignore it.

From your second statement, am I right to deduce that you consider the scatter area to be a rectangle?

The scatter area is an ellipsoid. The tan calculation corresponds to checking the distance between Target and the end of the angle scatter line in my picture

However, wouldn't you say that the scatter area looks a bit more like this?
Show Spoiler

No, if you look at weapons with very similar distance and angle like mortars they'll form a circle with enough rounds fired. If you put more extreme values in distance scatter for your weapons you will not get rounds flying off to the sides, all the very long distance rounds will be going down the center.

Perhaps this is influenced a lot by elevation, weapon angle etc?

Terrain elevation will definitely affect results.

Suxton and its 0.3 offset after around 40 rounds or so:

just to be clear: you were targeting the point itself?

just to be clear: you were targeting the point itself?

I was targeting where the cursor is in the image, on the white border of the sector capture area more or less closest to where my suxton was. Barrage should've been centered according to the targeting circle shown, but obviously wasn't, which made me realize how wrong I was about offset.

I was targeting where the cursor is in the image, on the white border of the sector capture area more or less closest to where my suxton was. Barrage should've been centered according to the targeting circle shown, but obviously wasn't, which made me realize how wrong I was about offset.

ah, opening the image in a new tab lets me see the circle. i was just looking at the shrunk version and the targeting circle doesn't show up in that.

This seems to be getting more and more bizarre

I'll probably play around with modding this weekend. I am getting curious about the effect of scatter_offset on units.

- How would sexton perform with a scatter offset of 0 or an offset of 1?
- How would mortars perform with a scatter offset of 1? Would that cause them to overshoot, or to undershoot? (mortar angle is way more than 45 degrees; aiming further up would cause the shell to land closer).

PS: Is there any way we can inquire Relic to ask them how scatter rolls are calculated. Obviously, since the effect doesn't follow the description of the modding tools, one of the two has to be a bug.

...
While testing this stuff it definitely feels like the explanation for offset in the mod tools is completely wrong. I always expected the hit area to be a half circle at 1 based on that, but instead it seems the offset just moves the entire possible hit area forward (or backward with a negative value). With a value of 1 all targeting is now centered around what used to be the edge of the old scatter area....
Light green ellipsoid part is calculated scatter centered around the targeting circle. Blue ellipsoid overlayed on top is an offset moving the entire targeting area forward.

If we forget for a sec the other factors like elevation difference and the tendency or not to overshoot the formulas you provided do not justify an ellipsoid shape. The only why one can get an ellipsoid shaped is if the probabilities of distance and angle scatter are linked, and a high distance scatter shot will have a low angle scatter and vice versa.

If the angular scatter and distance are statistically unrelated (and i have no idea if they are) one should be getting one of the following shapes depending on weather angular scatter uses distance to target to be calculated, or distance actually traveled:


If the length scatter and the angular scatter are actually related and the shape is actually ellipsoid what distance is used to calculated angular scatter would effect the shape. If the angular scatter use distance actually traveled the shape would be symmetrical over the X axis.
If angular scatter use distance to target then it overshot would tend to scatter angularly less while undershots tend to scatter angularly more and the shape would me more egg shaped..

If we forget for a sec the other factors like elevation difference and the tendency or not to overshoot the formulas you provided do not justify an ellipsoid shape. The only why one can get an ellipsoid shaped is if the probabilities of distance and angle scatter are linked, and a high distance scatter shot will have a low angle scatter and vice versa.

First, let's all agree that none of us know how Relic has implemented the scatter feature into their game.

According to the way scatter features are phrased in the attribute editor, I would be inclined to agree with Vipper.

However, according to Cruzz:

No, if you look at weapons with very similar distance and angle like mortars they'll form a circle with enough rounds fired. If you put more extreme values in distance scatter for your weapons you will not get rounds flying off to the sides, all the very long distance rounds will be going down the center.

I have not done the mortar test myself. This is probably the first thing I should try out.

My intuition is that Relic/whoever implemented the game engine, would have gone for an easy-to-implement option. The Vipper-model is very easy to implement. However, implementing an elipsoid scatter area is not as complicated as it sounds (it also happens to better match players' expectations).
Show Spoiler

Edited to add: There are also two things about the attribute editor that tip me off to believe Vippers theory more, and it has to do with horizontal scatter:
- Why is angle_scatter used, instead of some kind of scatter ratio?
Show Spoiler
- Why is there no max_horizontal_scatter attribute in the editor?
Show Spoiler

Then, you have two options

Option A:
2. Pick an angle from 0 to 360 that will determine the direction of the deviation from the center
3. Pick a value between 0 and the maximum distance until you exit the elipsoid. This would determine the distance

I have no idea how relic has implemented scatter.
In your option A the actual area would be a closer to circle around the intended target depending on the scatter value and the limit ...

As I said before if RNG gives 2 separate values 1 to calculate distance and 1 to calculate angular scatter the shape will probably not be ellipsoid.

An ellipsoid shape indicated that shot that scatter max at distance will scatter zero at angle and shot that scatter max at angle will scatter zero at distance and thus indicates a non pure RNG relationship to the two values....

It could be achieved for instance if one used the RNG produced value A and the other used 1/A value...

do not justify an ellipsoid shape.

You're right and I'm wrong. Finally got around to testing with some zero firing delay weapons to actually get meaningful amounts of rounds fired off and it's pretty obvious.

Higher firing angle barrages result in circular-ish results,
Show Spoiler

but the farther and lower angle you get the more clear the cone becomes.
Show Spoiler

I guess I was just trying to think of a single good reason for relic to always provide circular scatter previews ingame, was always kinda silly to expect them to use a more computationally complex ellipse instead of just the cone that comes naturally from how the firing is set up.

Excellent work Cruzz!

Now we know that the earth is not flat, but doughnut-shaped.

However, there is one crucial detail missing; what is the effect of the scatter offset on:
a) The scatter area
b) The distribution of shots in the scatter area

Previously, I think we were both led to believe that this only affects (b). However, your sexton test uncovered that it actually affects (a) (and perhaps not b at all).

Verifying (b) is tedious. However, we could still benefit from what happens with respect to (a)

Would it be possible to create 3 clones of the same unit (e.g. Sexton) and give each clone a different offset value?
- 0 offset
- 0.5 offset
- 1 offset

My intuition is that the effect of the offset value is to "nudge" the scatter area scatter_offset * vertical_scatter units further. If you do the math, this has the same effect as making (50 + 50 * scatter_offset)% of shells to overshoot.

Thus, scatter_offset will act more like an offset than a bias.

This would also perfectly explain what I experience when aiming PIATs.

You're right and I'm wrong...

To me it is not about who is wrong or right but understanding how things work and your contribution is far greater than mine, even in this subject you did most testing and come up with a result while i speculated more in theory, so the credit should go to you..

The shape in your second pic also seems to indicated that the distance used in calculating the scatter angle is the range to target and not the actual distance traveled (contrary to what would happen in real life) and thus my second schematic should be closer to the in game area..

.. and it seems that Cruzz was right about scatter_offset.

I did some trials with a 1.0 offset sexton and a 0.0 offset sexton.

Predictably, the 0.0 sexton had an even scatter area and the 1.0 offset sexton had its scatter area displaced entirely to the further side of the target.

I forgot to take pictures though

Updated the bulletin section a bit.

Bulletins

Stack.

Well, except for these:
soviet\medium-tanks\t34_85_armor
west_german\artillery\ig18_ability_recharge
west_german\light-vehicles\ostwind_armor
german\assault_gun\stug_e_armor
medium_tank\officer_tank (armor bulletin for sherman/okw panther, WILL stack on Panther but will NOT stack on Sherman)
NEW: brits_units\light_vehicles\m3\m3_armor

Team Weapons and Bulletins

Models used to recrew team weapons will still gain bulletin effects that have entity targets, which is most of them. So using grenadiers to recrew a maxim, your new maxim squad will have grenadier bulletin effects on them (these will not improve the machine gun itself because the team weapon is its own entity, only the groupies).

The default recruit/officer/veteran training bulletins for HMGs and ATGuns will function on all captured team weapons of the same type. So if you have a 4% penetration bulletins equipped for atguns as ostheer, an allied (or raketenwerfer, doesn't matter) atgun you steal will also have +4% penetration.

Some Ostheer bulletins affect multiple different units:

Note: This is an Ostheer only thing. Other factions have their bulletins targeted properly.

Note2: this is generally a one way effect. Assgrenadiers are grenadiers, but grenadiers are not assgrens. Specialist unit bulletins will generally not work with the base unit, its only base unit bulletins working on the specialist version.

Grenadiers & Assault Grenadiers & Assault Officer (entity type: grenadiers)
&& Ostruppen (entity type: conscripts)
Infantry Recruit Training (Grenadier & conscript accuracy)
Infantry Officer training (Grenadier & conscript reload)
Infantry veteran training (Grenadier & conscript cooldown+reload)

Ostruppen aren't grenadiers, but they are typed as conscripts so they get the soviet side of the bulletin applied to them. Relic pls.

Grenadiers & Assault Grenadiers & Assault Officer (entity type: grenadiers)
Grenadier cooldown

Tiger Ace and Tiger: (entity type: tiger)
Tiger penetration
Tiger armor
Tiger sight & turret rotation

PZIV command tank and PZIV: (entity type: panzer)
PZIV armor
PZIV Rotation speed
PZIV Reload speed
PZIV penetration

Stug G and Stug E and Elefant (entity type: stug)
The 3 recruit/veteran/officer training bulletins for su-76 and stug
Stug G Mobility
Stug E rate of fire (the one exception to the specialist bulletin thing)
Stug G penetration

Panzer Grenadiers & Stormtroopers (entity type: panzer_grenadier)
Panzer Grenadier Accuracy
Panzer Grenadier Received Accuracy

Well ok, USF has something similar though very minor

The 4 riflemen in a lieutenant squad , and 2 guys in a Captain squad are infact AEF_riflemen so they will get benefit from riflemen bulletins
2 guys in captain and major squads are REs for that sweet +1% accuracy bulletin too.

Prostruppen confirmed twin brothers of conscript squads.