Grenade trajectories are not static; they are shaped by game patches, collision hulls, release height, and server tick. This article breaks down why your default lineups drift after updates and how to rebuild them efficiently.
Introduction
Have you ever had this experience: a smoke that perfectly blocked Mirage mid last week now lands half a body off after this week's update. Your teammates say you "whiffed," but you know you didn't. The problem isn't your hand—it's the physics layer. Grenades are never static; their trajectory is a ledger written by the patch, collision hulls, release height, and server tick. This ledger gets quietly revised every few patches, and your default lineups were memorized from an old ledger.
1. Bounce Decay: Not "Bouncing Lower," but a New Energy Budget
First, let's clarify a commonly confused concept: grenade bouncing is not simply "reflection angle." In CS2, the collision response of HE, flash, and smoke includes three tunable quantities—restitution, tangential friction, and angular damping. Restitution determines how much energy is returned vertically, tangential friction eats horizontal velocity, and angular damping affects how far the grenade "rolls" after landing.
Among these three, restitution is the most sensitive. A change of just 0.02 can make a smoke that bounces off the wooden barrel on Inferno Banana to Trips land a full crate width off. Worse, restitution is not a global constant—different material surfaces (wood, metal, concrete, sand) have their own response curves. When a patch adjusts the physics material of a certain surface type, your "barrel to Trips" lineup is dead, but your concrete-floor lineup might be completely unchanged.
So the first step to relearning is not "throw it again," but to confirm: which surface does this lineup depend on? Lineups relying on barrels, metal sheets, or sand have completely different patch risks.
2. Collision Hull Precision: You Aim at the Texture, the Engine Calculates the Convex Hull
The second variable is more hidden: grenade collision detection does not use the visible model, but a set of simplified convex hulls. When a map author updates the collision box of a prop, visually there may be almost no change, but if the convex hull's vertices move, the grenade's bounce normal moves with them.
A typical scenario is "wall-hugging throws." You aim at the visual edge of the wall at Ancient's donut gap, thinking the grenade will graze the wall, but it actually hits a convex hull that extends two centimeters beyond the texture, and the trajectory is deflected early. Two centimeters at the release point amplifies to half a meter at the landing point.
Professional teams record the three-stage coordinates of "release point—collision point—landing point" for each default utility, not just the landing point. Once the collision point is changed by a patch, the landing point will inevitably drift, but if you only memorize the landing point, you won't know which stage the drift starts from.
3. Release Height: Standing, Crouching, and the "Invisible Offset" While Moving
The third variable comes from yourself. The grenade's initial position is bound to the character's hand bone, and the hand bone's position is affected by stance, crouching, and even movement speed. If a patch adjusts the hand bone's binding offset, or changes the relative relationship between the camera and hand when crouching, your release point changes even with the same crosshair placement.
This is why some default smokes work fine when "standing throw," but are wildly off when "jump throw" or "crouch throw." Jump throw is especially dangerous: the vertical velocity at the moment of jumping adds to the grenade's initial velocity, and the vertical velocity value is affected by patch physics parameters. You think jump throw just flies a bit further, but it actually changes the shape of the entire parabola.
4. The Right Way to Relearn: Rebuild Your Coordinate System in 20 Minutes
Don't blindly rethrow a hundred times. Follow this order:
- First, fix your stance and crosshair anchor, change only one variable, and confirm whether the release point or collision point has drifted.
- Use "short-distance low toss" as a baseline test to eliminate cumulative air resistance errors from long distances.
- For each map's default utility, prepare at least two redundant lineups—one relying on bounce, one on direct throw. If one side changes, the other can still work.
- Write the new landing point in your notes, but annotate "patch number + test date," because it may change again next update.
Grenades are not memorized; they are calculated. The patch tax must be paid every time. The only difference is whether you pay it in advance or get taken out in a match by an opponent's smoke you can't block.