Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

Recommended Posts

I know it's been documented before, but here's another 3-shot AV kill with an SSG (plays back with Vile Challenge (vilchal.wad), -complevel 2. (I got a faster time somehow right after this demo, so I'll eventually release the faster demo with this as a bonus)

Somewhat notable since there are no other monsters, so the randomization factor comes solely from randomly flickering sectors.

Also, there's not one but TWO other AVs that die from three SSG shots+AV fire blast damage, so the first part of this demo is very lucky (shame I screwed up with the BFG at the end).

vilc-110.zip

Share this post


Link to post
dew said:

A revenant in one rocket.

I deal 358 damage to him, so I assume it's another case of what RjY described. wad, demo (the max one).

I agree. Also, thanks for the credit in the video description :-) In this case the sleeping monster "catalyst" is the second revenant behind the wall (thing 265). Four P_DamageMobjs occur during gametic 3796:

1. thing 264 is hit directly by the rocket, taking 120 damage
2. thing 264 then takes 119 splash damage
3. thing 265 takes 63 splash damage, is woken up, is moved back to start of list of actors in its blockmap square.
4. thing 264 gets processed again and takes another 119 splash damage. 120+2*119=358.

The double splashes were observed to be the same magnitude both times in all double splashes thus far analysed (marineController noted the previous page's demo had 2*110 and 2*100). It seems reasonable to claim this will be the case generally when the target is already awake when struck. The target will have a large momentum/velocity set after the first splash damage but its position will not alter until P_XYMovement in the next gametic. Thus the distance from explosion to target remains the same for the second splash, which entirely determines the magnitude of the damage (there is no random component).

Share this post


Link to post

I updated the first post with our "achievement list". A revenant in 4 SG shots is the same as 2 non-aligned pinkies in a row in 4 SG shots, so indeed it is not rare. Such posts may be removed. Nevertheless, feel free to post demos containing:
- unusually resistant monsters
- kills which you're unsure whether they're "rare"

Share this post


Link to post

Oh sorry. I didn't realize there was criteria for it. They just seemed rare to me.

Share this post


Link to post

Not rare in itself, but that it happened like this was at least rare for me:

At 0:28, two zombies with a shotgun shot (not lined up)
At 0:38, two shotgunners with a shotgun shot (not lined up), and with the very next shot, another two (also not lined up), almost from behind the wall.

In this run, I was aiming for anything below 50 seconds, which I found very hard at the time (December last year). I just got lucky, also it helped that I grinded (ground?) for this for many hours.

Plays back with Rebirth1.wad.

Link

Share this post


Link to post

I think it still counts as a rare kill if an enemy takes as many shots as possible.
A mancubus eats 4 rockets and still craves more.
Occurs in hellbnd.wad, prboom+ 2.5.1.3 cl2, at around 4:33 (-skipsec 273)

hb04rare.zip

Share this post


Link to post

Also in Fitzkrieg's demo, the SSG shot that finishes off the manc just barely does (it had 24 health at the last shot and the SSG did 25), which I think contributes to it being a little more rare.

Share this post


Link to post
doomdaniel95 said:

Also in Fitzkrieg's demo, the SSG shot that finishes off the manc just barely does (it had 24 health at the last shot and the SSG did 25), which I think contributes to it being a little more rare.


Well, that's just a similar effect to how you can line up monsters. Some of the histcans hit the manc, the rest go through.

On the subject of that, I am curious about how the engine decides which pellets hit the closest target and which hit further targets. I've sort of never considered that question until now.

Share this post


Link to post

Shots pass through the parts of a monster that are in a different blockmap square from the monster's centre. So if a monster is standing on the line between two blockmap squares half your shells will pass through it and hit the thing behind.

This is the somewhat famous blockmap bug. ZDoom calls it (along with a related issue) compat_hitscan, Odamex calls it co_blockmapfix. (Is it on Doomwiki.org anywhere? I can't find it) Edit: turns out it is flawed collision detection on the Doom wiki. Edit2: dr_sean reminded me of cph's extremely detailed explanation of the bug.

Share this post


Link to post

Revenant takes full ssg shot without flicking. I think chances for that are rather astronomically low. Plays back with stardate.

heh.zip

Share this post


Link to post
j4rio said:

Revenant takes full ssg shot without flicking. I think chances for that are rather astronomically low.

Indeed: revenant pain chance is 100/256, so chance of no pain with all shots on target is (1-100/256)^20, just under one in twenty thousand. Congratulations :)

Your pain chance rolls (0-255, need to be 0-99 to trigger) were: 117, 126, 217, 255, 101, 126, 104, 222, 166, 132, 148, 238, 168, 125, 225, 153, 221, 125, 172, 230

Share this post


Link to post
RjY said:

Your pain chance rolls (0-255, need to be 0-99 to trigger) were: 117, 126, 217, 255, 101, 126, 104, 222, 166, 132, 148, 238, 168, 125, 225, 153, 221, 125, 172, 230

What are those numbers? I looked at rndtable[] and it doesn't even include 117, 126, and several others.

Seems to me that if it's possible to hit a revenant with a full SSG blast and not trigger pain, it must be at worst a 1/256 chance, since you're entering P_Random() at some point and it's otherwise fully deterministic.

Share this post


Link to post
Linguica said:

What are those numbers? I looked at rndtable[] and it doesn't even include 117, 126, and several others.

The demo is -complevel 9 (Boom 2.02). Boom's random number generator uses one of these. Its range of return values is a superset of rndtable.

prboom-plus/src/m_random.c is the code, in particular

boom = rng.seed[pr_class]; // in this case pr_class is pr_painchance (7)
rng.seed[pr_class] = boom * 1664525ul + 221297ul + pr_class*2;
boom >>= 20;
return boom & 255;

Share this post


Link to post

Demon dies in seven berserk punches, demo plays back with truetest.wad. Punches did 20,40,20,20,20,20,140 damage to demon.

Use the parameter -skipsec 75 in PrBoom+ if you want to skip to the demon kill, -trace_thingshealth 146 to see damage done to demon.

t1t7demn.zip

Share this post


Link to post
Kotzugi said:

How rare is a hell knight kill with two rockets? I'm not familiar with the maths on rare kills. Plays back with HR.WAD and happens at around 1:30.


Not too rare, my rarest with HKs and rocket launcher was 2 hell knights in three rockets in one of my Scythe episode 2 NM-Speed attempts, so one hell knight in 2 rockets isn't too unusual (though it is uncommon).

Share this post


Link to post
Kotzugi said:

How rare is a hell knight kill with two rockets?

Such a probability is roughly 10-15%. I've calculated it based on this doomwiki page and particularly this graph:



BEWARE - MATH:

Spoiler

There's a 12% probability that a rocket hit will deal 288 damage, and 11.5% probability that it will deal 268 damage. That's roughly 22% probability that the damage will be above 250 HP, which is half of Hell Knight's health 500.

22% = 0.22.
Two rockets = the probabilities multiply each other.
0.22 * 0.22 = 0.0484 = 4.84%.

Then there's also 13% probability that a rocket hit will deal 248 damage, slightly below HK's half health. If one of the rockets deals 248 damage and the other one deals even more, it will kill the Hell Knight too. It can also be 228 + 288. Taking this into account, it increases the overall probability - actually more than twice, because now there's more combinations possible and the probabilities are roughly the same for all of them. So I estimate 10-15%.

If I was at school, I'd have to calculate it precisely, but I've already ran out of enthusiasm to do so. :)

Well, this is not the correct way how to do math, but the estimate should be right. The doomwiki page I've linked states that 2.86 rockets are needed to kill a Hell Knight on average. That means that there's about 10-15% probability that only 2 rockets will be needed, otherwise it would be 3.

Share this post


Link to post

odysseyofnoises said:
The chances of killing a revenant with 1 ssg shot is 1 in 5 million. has this ever happened to anyone anywhere?

source: http://www.gamers.org/pub/idgames/lmps/tyson/weapons.html


Never with vanilla, but I had it happen with an early BOOM beta (which I'd have in a demo, but that beta ate demos) and four times since in BOOM descendants. I take this to mean I've killed 25 million hellspawn, which seems about right. ;)

I've had a few WTF kills in ObHack maps with 1000+ monsters. In one case it went like this: Three monsters up on a platform, no prior damage far as I could tell, no gunfire in the area. Hell knight offended the revenant, who then went and beat up on the innocent mancubus... and the hell knight died.

Share this post


Link to post

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×