Adding Godot Screen Shake and Hit Stop, a Devlog
The most upvoted comment in our last feedback thread was four words long: "hits feel like nothing." Eleven other players agreed under it. Our platformer had enemies, a dash attack and a boss, and none of it landed with any weight.

The most upvoted comment in our last feedback thread was four words long: "hits feel like nothing." Eleven other players agreed under it. Our platformer had enemies, a dash attack and a boss, and none of it landed with any weight. You pressed the button, a number went down, and the enemy blinked. That was it.
So this sprint we added screen shake and a few related bits of feedback in Godot 4.3, and then spent nearly as long taking some of it back out. This devlog covers what we built, the numbers we ended up with, and the one piece of popular advice that made our game worse before it made it better.
What we started with
Our camera was a plain Camera2D attached to the player, with position smoothing on and a speed of 6. Hits were handled in the enemy script: subtract health, play a white flash shader for 0.08 seconds, play a sound. No camera movement, no pause, no particles worth mentioning.
Before touching anything, we recorded a 40 second clip of the same fight with the boss and posted it in the feedback channel with a simple question: on a scale of 1 to 5, how much does each hit feel like it connects? The average across 23 replies was 2.1. We wanted something to compare against later, and a rough number from real players beat our own opinion.
Building the shake
We went with a trauma based shake, the approach a lot of Godot tutorials use, because it scales nicely when several things hit at once. Each event adds some trauma, a value between 0 and 1. Every frame, trauma decays, and the actual shake amount is trauma squared, so small hits barely move the camera while big ones really kick.
The camera script holds three settings: max_offset, which we set to 14 pixels; max_roll, set to 0.06 radians; and decay, set to 1.6 per second. Each frame, the script reads a FastNoiseLite value for x, y and rotation, multiplies by the squared trauma and the maximums, and applies it to the camera's offset and rotation. Using noise instead of randf_range gave a smoother, rolling shake. With random values every frame, the picture jittered at high frame rates in a way two testers described as "vibrating," which was not what we were after.
One thing that tripped us up: position smoothing on the camera fights the shake if you move the camera's position. Moving the offset instead avoided that completely. I lost most of an evening to that before reading the docs properly.
The trauma values we settled on after playtesting look like this:
| Event | Trauma added | Hit stop | Notes |
|---|---|---|---|
| Player sword hit on small enemy | 0.18 | none | Barely visible on its own, on purpose |
| Dash attack connects | 0.30 | 0.04 s | The one players noticed most |
| Enemy defeated | 0.35 | 0.06 s | Plus a burst of 12 particles |
| Player takes damage | 0.45 | 0.08 s | Also a red edge flash |
| Boss slam lands | 0.60 | none | Shake only, no freeze |
| Boss defeated | 0.85 | 0.20 s | The only time we let it go big |
The advice that made it worse
Almost every game feel talk and video we watched said the same thing: when in doubt, add more. More shake, more particles, more freeze frames. Overdo it, then dial it back. We took that literally in the first build, set the sword hit to 0.4 trauma and every enemy death to 0.6, and pushed it to our playtest group.
The response was the opposite of what we hoped. In a room with six enemies, the screen never stopped moving. Three players said they could not track the small spiked enemies during a fight. One said it gave her a headache after ten minutes, and she was not the only one; two more people mentioned it privately. Our rating question went up to 3.4, but two of the comments under it were about turning the effect off.
What we learned is that "overdo it, then dial back" works when you test a single hit in isolation. It falls apart in a real fight where events stack. Squared trauma helped, but the real fix was making small, frequent events tiny and saving the big shakes for rare ones. In our table, only two events cross 0.5, and you see each of them maybe once per level.
We also added a camera shake slider to the options menu, from 0 to 100 percent, defaulting to 80. That took about twenty minutes, and it is the change I would tell anyone to make first. Some players simply do not want it, and the game should not punish them for that.
Hit stop, the part that actually fixed the feel
Surprisingly, the shake was not what moved the needle most. A short freeze on impact did. When the dash attack connects, the game pauses for 0.04 seconds, which is just over two frames at 60 frames per second. You barely notice it as a pause, but the hit suddenly feels solid.
We did this by setting Engine.time_scale to 0.05 for a short timer and then back to 1.0. The timer from get_tree().create_timer() has to be created with ignore_time_scale set to true, otherwise it slows down along with everything else and the freeze lasts far longer than intended. That bug had the boss defeat freezing for almost four seconds in one build, which a tester found funny and we did not.
Sound mattered too. We moved the hit sound to fire on the first frame of the freeze, not after it, and shifted its pitch randomly by up to 8 percent so repeated hits do not sound identical.
Where it landed
We posted the same 40 second boss clip again with the final settings. The average rating came back at 4.0 from 27 replies, and nobody mentioned headaches. The most common comment was that the dash "finally feels like a dash," which is the exact thing we wanted.
If you are working on something similar, try this before anything else: record one fight, ask players to rate it, change one thing, and ask again. We sorted every comment the same way we did in our devlog on sorting 63 player comments, and the options slider ties into the menus we rebuilt for controller support in Godot. More progress notes are in the devlogs section.
FAQ
How do you add screen shake in Godot 4?
Move the Camera2D offset each frame using a noise value multiplied by a decaying trauma amount. Moving the offset avoids fighting position smoothing.
How much screen shake is too much?
If frequent events like normal hits shake the screen noticeably, it is too much. Keep common events small and save large shakes for rare moments.
What is hit stop in games?
A freeze of a few hundredths of a second when an attack lands. It makes impacts feel heavier without moving the camera at all.
Should screen shake have an option to turn it off?
Yes. A simple slider costs little to build and helps players who get motion discomfort or who just find it distracting.
More in Devlogs
Devlogs
Adding Sound and Music to Our Godot Game, a Devlog
We shipped the platformer demo to eleven playtesters with no audio at all, because audio was the thing we kept pushing to next...
Devlogs
Adding Controller Support to Our Godot Game, a Devlog
Three of the first ten people who played our demo picked up a gamepad, pressed A on the title screen, and nothing happened.
Devlogs
Adding a Godot Dialogue System to Our Game, a Devlog
Our platformer had six characters who needed to talk, and every line of their dialogue was hardcoded into a label node with a...