Phasmophobia – Collection of Tips, Game Mechanics and Oddities

Here’s a bunch of mechanics that may be less well known. This isn’t intended as a beginner’s guide.


This guide is not aimed at beginners. Neither is it a summary of what every ghost can do. (There are lots of those, am I making one? Not at the moment.) It’s a collection of random stuff that random players at random stages might want to know. If you’re just a new player trying to figure out how to survive when the ghost appears, this guide… isn’t for you? Probably?

There are some tips that beginners can use, though. If you don’t have time just read the summary for each tip/mechanic.

Stuff That Works Through Floors (and Doesn’t)

Summary: Can’t yell, can’t power the Raiju, can die on the Maple Lodge balcony. And other things.

Here’s a (non-exhaustive) list of things you can’t do/won’t happen across different floors:

  • Ghosts can’t hear you across floors. If you’re trying to test for a Yokai or attract it in general make sure it is on the same floor before trying, or hold an active electronic. (Just be careful for a Raiju)
  • The Raiju doesn’t get powered by active electronics that are located on a different floor.
  • Electronics don’t flicker if the ghost is on a different floor.
  • Ghost events probably (?) don’t happen across floors. Unless it walks to your floor. But since it can happen to roam to your floor before triggering a ghost event, then whether it does do “real” cross-floor ghost events or not isn’t very useful.

Fun fact! Sometimes during hunts the ghost bugs out and thinks it hasn’t changed floors. So the rules still apply, just that now you’re shouting at the ghost behind the Brownstone stairs and think it’s a Yokai because it Doesn’t Care.

Things that can happen:

  • Your Edgefield Street House Master Bedroom Demon can chomp the crucifix that is still hanging inside the van. 99% a Demon if this happens. (A Garage Demon can chomp it too, and it’s much easier.)
  • Ghosts can see you across floors. Which means that the balcony hiding spot on Maple Lodge (Campsite) is actually more dangerous than actually hiding indoors with the door closed, even while crouching. Also, Campsite landscaping is just bad for LoS blocking in general so you’ll have a full-speed Spirit out of nowhere and go “it’s a Jinn” or something.

You May Not Have to Change Your Hiding Spot Every Hunt. Here’s Why!

Summary: You only need to change hiding spots if you were detected in the spot (for closets, lockers, and small tents), and in or near the hiding spot (if it is furniture-based).

How it works

The actual mechanic was first introduced in

  • The ghost now has a 50% chance to remember where it last saw you during the last hunt and will search that area in the next hunt.

I’d assume electronics and voices also count. (If it doesn’t let’s just assume it does)

What this means specifically, is that if the last time the ghost detected you in Hunt 1 was in the Living Room at the table, it has a 50% chance at the start of Hunt 2 to set a waypoint at the Living Room table and go there first looking for you.

It’s something like the Escape the ghost during a hunt objective. Simply hiding won’t work, you need to be seen by the ghost (and everyone must survive that hunt) for the objective to be completed. Similarly, you need to be detected by the ghost for it to roll the 50% chance whether to head straight there first thing in the next hunt. If no one was detected at all, the ghost will not remember where anyone is.

The caveat for furniture-based/organic hiding spots: if your RNG is really bad, the ghost might go there anyways since… RNG. Also, if someone was detected near (not necessarily in) a organic hiding spot, because the ghost will search the immediate vicinity for a short while before moving on, it might chance upon you in your spot.

If you are in a closet or a locker, ghosts can’t physically enter them (only brush by), so good! You will be safe for the next 74148 hunts as long as you weren’t detected in there, and you hold the door shut.

A short-but-not-so-short summary:

  • If you weren’t detected in a locker or closet, you can hide indefinitely in that same locker or closet.
  • If no one was detected in or near an organic hiding spot, changing hiding spots is generally not needed.
  • If someone was detected in or near an organic hiding spot, it is usually a better idea to change to a hiding spot further away.
  • If you have a Deogen, get the hell out of your hidey hole.

I Need to See You to Kill You

Summary: They need to have LoS and be near enough to kill you, or you get safe spots.

Ghosts are very simple. They see you, they touch you, they kill you.

Right. What actually goes on is: you have a total of 6 “tracking points” on your player – 3 around the lower half and 3 around the upper half. If the ghost can see all 3 on either half, this is considered line-of-sight (LoS). If the ghost has LoS with you and it is close enough to you (effectively “colliding” or entering “kill zone”), you die. If it sees 2 on your feet and 2 on your head, you can’t die, even if it’s right beside you. If it’s next to you and it sees 3 points on your head, time to collect your $42.

Now, how do we figure out what allows the ghost to “kill”? Let’s take two safe spots:

Scenario 1: Brownstone stairs underside

If you crouch under the stairs (going as deep as possible), you will not die even if the ghost sees you there. If the ghost goes under the stairs (as opposed to getting stuck on the stairs), it sees you and speeds up; this means that it can see the player.

Hence we establish that the ghost has a “hitbox” or “kill zone” where you will die if you get too close.

Scenario 2: Asylum nurse room safe spot

There’s a safe spot between the bed frame and the other metal frame. By attracting the ghost to the long side of the metal frame, if you crouch, you can survive:

But if you stand up, you get the following:

We can hence infer that the ghost also requires LoS to be able to kill a player.

Thus the ghost must satisfy two conditions to be able to kill a player:

  • Be near enough to the player
  • Can see the player (LoS)
  • ??? who knows maybe there’s a third requirement! But so far no evidence for this

The 20 or so safe spots across all the maps are a result of the ghost not meeting one (or both) of the requirements.


Since we need LoS to kill players, and the Deogen always has LoS, that means that you can get killed through the wall by a Deogen if it’s thin enough and/or both of you are pressing against it. Which also means that half of existing safe spots don’t work with a Deogen. Which also means just loop it it’s fun to take your pet for a walk.

Smudging, and Banshees

I’ll leave you to decide how this one works.

You Don’t Need Both Closet Doors Closed

Summary: Hide in the very back corner and hold the door that is closer to you. You (probably) won’t die. As long as you do the other things you’re supposed to do.

The why

Closets! If you stand in one of the corners of the closet at the very back, and hold the door that is closer to you shut, even if the other door is fully open, the ghost cannot establish the three-point LoS (discussed in previous section) needed to 1) see you, and 2) kill you. (To repeat, ghosts cannot walk directly into closets.)

Why hold the door? If you are holding any type door, ghosts cannot override it unless it detected you (if you have your flashlight on, for example) or if you’re hiding behind a normal door and the ghost gets very close. Which means that as long as you hold your door, the ghost won’t get you unless you do a dum-dum and let the ghost know that you’re there. There’s also the added benefit of letting your teammate enter to hide, unless they’re the dum-dum.

Props to Shuee (the level 24k streamer) for doing this often enough for me to pick up on this.

The Ghost Has Other Detection Priorities

Summary: Can’t really summarise this or you might misinterpret it. Just read the More Info section.

Are you a Banshee?

Sometimes you get ghosts that don’t really act quite right. Like, clearly you’re closer to it, but it chooses to walk past you to another player? And it’s not a Banshee (nor a Deogen)?

They’re trying to test for a Banshee with the setup shown. The ghost room is in the garage (magenta). If it was not a Banshee, they were expecting the ghost to come out of the room, see Player A (it’s labelled differently on the original), and chase them. Player A smudges and hides, and Player B sees if the ghost chases them.

If it was a Banshee, the ghost would either ignore Player A in the first place, or ignore Player B. In reality, although the ghost gave zero thought to Player A and directly chased Player B, it turned out to be a Raiju.

What was their conclusion? After several of the same occurrences happening, they decided that when the ghost starts a hunt and the grace period ends, If there are multiple players detected, it will choose the nearest player by straight-line distance (ignoring obstacles in the way) to chase, instead of pathfinding distance (pathfinding distance is the distance it takes for the ghost to walk to a player from its current position).

More Info

I’m going to add further to this: if two players are detected at exactly the same time (such as the start of a hunt), it will “lock on” to the nearer player even if they then become the “further player” later on, until the ghost loses detection of them. So if Player A is at one end of a High School hallway, the ghost starts chasing A from the other end (after coming down the staircase), then Player B suddenly pops up at the same end as the ghost (from behind), the ghost is going to ignore them, even if they’re way closer! Unless Player A goes behind a wall, then the ghost will now reverse and chase Player B.

Just to be very clear: even if the ghost “locks on” to one player, other players will still die if they touch a ghost. Unless its a Banshee while their target is inside the house. And ghosts do not “target” someone permanently for the rest of the hunt. This mechanic is mostly about the specific case where multiple players are detected at once.


How about a Deogen? Since they always have line-of-sight with everyone, they will simply pick the closest player by straight-line distance. So if the player below the Deogen on a different floor is closer than the player on the same floor, good luck to the player on the same floor. (Unless it happens to trigger the “refreshing” of targets before it runs them over).

Mirror, Mirror, on The Wall, How Much Sanity Do I Fall?

Summary: By playing around with video times we can measure to a high accuracy the sanity drain rate of the Haunted Mirror. (It’s closer to 10% not 7.5%)


When the Cursed Possessions first released some people were trying to figure out some of their mechanics, like “How much sanity does the Haunted Mirror drain?” or “Am I practically immortal when the Ouija Board breaks in my hand?” (For a very short time the answer was yes, by the way.) So here I was testing the Haunted Mirror by taking a timer and timing it from 100% sanity, starting when I right clicked, and stopping when it broke.

The result I got from that was approximately 10%/s. The only issue is, one of the developers said on one of Insym’s streams that it was 7.5%/s. And most people who know this have tagged along with it. Eight months later, I have finally figured out a more accurate way of measuring the drain rate.


I went in the house, used the mirror a second, then ate pills to regain full sanity. Then I used the mirror until it broke. I had that recorded.

So there’s this thing called VLC, and I installed some plugin that allows you to get the timestamp (in milliseconds) of each frame. I then took the time beginning from the first frame where the mirror started to raise, all the way until the first frame where it broke. VLC isn’t the best for this stuff but it works.


10.077 seconds to break. Which is around 9.92%/s… about 10%! Unfortunately I have an issue called My Computer Isn’t Powerful Enough To Look At The Mirror With 60 FPS, so it’ll be slightly off (but not seconds off). 15 fps is fine.


There’s no sanity penalty for simply performing the right click; it’s purely based on the amount of time you spend with the mirror active. If you right click and walk out and walk back in and right click and walk out and walk back in and – it drains slowly, millisecond by millisecond.

Can The Twins Be Faster Than a Revenant? An Analysis of Ghost Speeds in Phasmophobia

Summary: This is probably the closest we can get to accurately measuring ghost speeds without asking the developers directly. If you don’t want to read a research paper just skip to the bullet points. Most of the speeds are listed there.


Some have tried different ways to calculate ghost speeds during hunts. The main way is usually “eye power” (Psycho does this I think?), asking the developers themselves, or the developers deciding “here’s the specifics in the patch notes :)”. Here’s an alternative method.

The setup

Remember the motion sensors from the Demon case? These can be used (pretty accurately!) to find out the ghost’s speed. Here’s an example setup for measuring the maximum ghost speed under LoS (I used a different hallway because in the actual game there are obstacles for this one):

  • Magenta = ghost room
  • Yellow = player location during test
  • Red = motion sensors (their reach isn’t actually that far)
  • Dark blue = ghost path
  • Light blue = distance between the ghost room door and the first motion sensor (~30 to 35 metres)

What’s happening here: we need the ghost to attain maximum speed, so we have the first 30-ish metres for this to happen. The next distance across the motion sensors is where we measure the speed.

To minimse wandering/prevent room switching, these tests were done on only Amateur or Intermediate. The whole “runway” is about 50 to 70 metres, depending on where the ghost room is.

Motion sensors

To place the motion sensors, I used a macro (via Logitech G HUB since my mouse works with it), setting a fixed time for moving – I’m have no idea what I’m saying here’s a screenshot.

I placed motion sensors one each before and after the macro, and one during the 3 second wait time.

To ensure that the player walks roughly parallel to the wall, I aligned the camera such that the black lines at the bottom of the wall are as un-jagged as possible. The longer the motion sensors are placed apart, the better the precision.

Preferably the motion sensors are placed on the same side as the ghost room (or where the ghost will enter from), so that it’s a straight line and we don’t underestimate the speed. If we have 3 motion sensors, we can check if the ghost is still accelerating between the first and second sensors (since the time taken will be different).

Avoiding death

Now comes recording the whole thing! Again, my computer isn’t great, so I had to minimise the game to a windowed 320×200 pixels to achieve 60 FPS while recording the entire hunt (or so I hope it did). We then take the footage into VLC, and use the same method we used for estimating the mirror’s sanity drain, just that now we measure the time taken between the first frame the first motion sensor lights up, and the first frame where the second motion sensor lights up.


How accurate is it? On Tanglewood, I measured the Deogen’s slow crawl speed (before they revealed it) at around 0.411 m/s. The actual speed is 0.4 m/s, so for a very short distance to test, that is a good start. I measured the Thaye’s slowest speed at 1.1 m/s (slightly off from 1 m/s, but it was done in Tanglewood, so again, not the best testing site.)

I did a test in Asylum with the Revenant with LoS, and with the longer distance offered, it came to about 2.97 m/s (I think the most commonly accepted value is 3 m/s). It’s fairly accurate. Here are the numbers (including players for reference):

EntityConditionBase speedSpeed with continuous LoS
Standard ghost1.72.8
DeogenPath distance to player < 2.5m0.4N/A
DeogenPath distance to player < 2.5m3N/A
HantuDepending on room temperature1.4 – 2.7N/A
Jinn> 3m from player & fuse box on & player in LoS2.5N/A
MoroiDepending on average sanity1.5 – 2.32.5 – 3.7
RaijuWithin range of electronics2.5N/A
RevenantWhether player is in LoS13.0 (instant speed change)
ThayeDepending on age1.0 – 2.8N/A
The TwinsMain “twin”? 1.6? 2.6
The TwinsDecoy “twin”1.9? 3.0 ~ 3.1

All numbers in m/s and to 1 d.p.

Base speed vs top speed

Let’s take the Moroi:

  1. Maximum speed at 0% sanity: 3.71 m/s
  2. Minimum speed at 0% sanity: 2.25 m/s
  3. Divide: 3.71 ÷ 2.25 ≈ 1.65

So we can assume (at least for the Moroi) that it uses a multiplier (setting it manually to 3.71 or u + 1.46 wouldn’t really make sense). With more decimal places, the exact max speed would be 3.7125 m/s.

Now I’ll take a bold step and say that this applies to all normal LoS speed-up ghosts. Which would match up with most of the data… except The Twins. Can a max speed Twins decoy be faster than a Revanant? …No idea. If we take our measured maximum speed (2.62 to be more precise) then divided it by 1.65, it would be about 1.58. Which is different from what the developers say: 10% slower than the normal ghost speed, or 1.53!

  1. We don’t actually know if it’s a 1.65 multiplier for all LoS ghosts.
  2. Precision isn’t high enough.

That’s why the value above has a question mark beside it. I’ll try to test this more precisely in the next few weeks on Asylum, or else I’ll have to do this in Sunny Meadows! I heard they’re removing infinite line-of-sight though, which might suck. Not for most players just for me testing this out. Don’t quote me on this.


I thought that acceleration would be impossible to measure, but now that I think about it, it is totally possible. The next sentence contains a physics equation. (Take two sets of average speeds when it’s speeding up between two sets of motion sensors then use v = u + at) Since I haven’t tested it yet, the only thing I can currently give you (based on the data from my speed testing) is that ghosts do not accelerate slower than 0.07 m/s². (I think it’s 0.1 m/s²?)

Some Other Smaller Tips and Mechanics

If you don’t know, now you do. If you did know, here’s affirmation. I did not do a survey on how well known these are (neither did I for any of the other tips):

  • The D.O.T.S Projector silhouette you see is the same as the ghost model you see in events and hunts.
  • Baby ghost? Baby feet. Here’s a comparison between the regular one and the one for Tooth-Faced Devil:
  • (You might have seen this image on the wiki… because I was the one who uploaded this image. To the wiki. I took the screenshot.)
  • The Obake’s fingerprints have a 1/6 chance to occur per successful “imprint”.
    • Sample size of ~200 yielded just over 30 obake fingies.
    • There are 6 handprint designs, 1 of them being Obake. Same for fingerlight and keyboard smash. Which works out to 1/6. Seems logical to me.
  • There isn’t any increased drain during a ghost event/hunt simply for being in LoS with the ghost except for the Phantom (which is either 0.4%/s or 0.5%/s). Is it bugged? No idea. But it doesn’t seem to exist for the moment.

You May Also Like

Be the first to comment

Leave a Reply

Your email address will not be published.