Mantaflow Field Guide: What Every Parameter Actually Does

There’s a particular frustration with fluid simulation that anyone who’s spent twenty minutes in Blender’s Mantaflow system knows: you change one slider, wait twenty minutes for a bake, and end up with something that looks like either a toddler’s watercolor accident or a physics glitch from 1998. The problem isn’t that Mantaflow is bad - it’s actually quite good - it’s that every parameter speaks a slightly different dialect of the same physics language, and most tutorials either skip the nuance or dump you into a finished scene without explaining why the settings matter.

Alec Scoreborg’s Part 2 of his Mantaflow introduction takes a different approach. Rather than building a beach scene or waterfall, he treats the tutorial like a field guide - here’s what each parameter does, side-by-side, with honest trade-offs you can actually use for decision-making. No pretty final renders, just the engineering underneath.

The Flow Setup: Geometry vs. Inflow

Scoreborg starts where Part 1 left off: the domain is a cube, the liquid type is selected, cache ends at frame 200. The first decision point is what kind of source you’re working with. Flow behavior determines whether your fluid object is a one-time event or a continuous emitter.

Geometry means the fluid spawns once from the object’s shape - think a puddle, a spilled drink, a contained volume that starts where it is. Inflow keeps generating fluid every frame - think a faucet, a hose, a waterfall that never runs dry. The distinction matters because it changes what your scene can become. An inflow needs a drain or open border eventually, or your simulation will drown itself in particles.

The clever trick Scoreborg demonstrates here is animating the inflow - you can keyframe the flow object to stop emitting fluid at frame 50, which means you get the best of both worlds: a filling action that becomes static geometry once the tank is full.

Resolution, Time, and the Border Problem

Resolution divisions are the most obvious trade-off in fluid sim. Higher looks better, higher takes longer. What Scoreborg adds is the practical observation that the difference between 64 and 128 divisions isn’t just cosmetic - it changes how the liquid interacts with itself, whether the surface tension reads as convincing or “computer-y.” His advice: start low, get your motion right, then scale up.

Time scale is where Mantaflow defaults bite new users. The default speed of 1.0 is actually too fast for most natural-looking fluids. Scoreborg prefers 0.5 - which gives the liquid more weight, more viscosity, more sense that it’s actually water or oil or whatever you’re simulating rather than a caffeinated physics demo.

The border collision settings deserve more love than they typically get. You can toggle individual walls of the domain cube - meaning fluid can exit through the top (open container) or flow out a side (drain). The default behavior treats every wall as a solid bounce, which isn’t what most real-world containers do. Turning off top and side collision creates entirely different simulation possibilities.

Flip Ratio and Particle Physics

Flip ratio controls splash intensity, and this is where Mantaflow gets interesting. The default value of 0.97 produces aggressive, energetic splashing - great for explosions, less great for calm pools. Drop it to 0.1 and you get glass-smooth liquid with barely a ripple.

Scoreborg’s side-by-side comparisons here are genuinely useful: one value on the left, a different value on the right, same simulation otherwise. You can see exactly what the slider does before committing to a bake.

Particle radius has an inverse relationship with perceived volume. Higher radius shrinks the liquid, lower radius expands it. This is counter-intuitive - you’d expect larger particles to mean larger liquid - but it’s useful for faking fluid density without changing the physics. Need your liquid to look heavier? Shrink the particle radius slightly.

The max and min particle settings per cell control density at the simulation level. More particles means more detail but more computation. There’s no universal right answer here - it depends on your scene’s scale and your hardware’s patience.

Mesh Generation: Smoothing and the Blocky Water Aesthetic

Once the particles are calculated, Mantaflow generates a mesh from them. The up-res factor scales this mesh resolution - higher values mean smoother surfaces but longer processing. Positive smoothing melts the mesh toward nothing at extreme values, useful for stylized effects or fixing artifacts.

But the interesting discovery is negative smoothing. At extreme negative values (around -40), the mesh doesn’t just stay rough - it produces chunky, blocky, plateau-like water. It looks like a mistake until you realize it could be a deliberate aesthetic. Stylistic low-poly water, voxel-style fluid, something deliberately non-photoreal. Scoreborg points this out as potentially useful rather than broken, which is exactly the kind of observation that separates reference tutorials from tutorial tutorials.

Effectors and Secondary Particles

Effector objects (collision objects that interact with fluid) get a quick demonstration - a cube falling through the liquid, creating displacement and splash. The fractional obstacles setting controls boundary smoothing where fluid meets collision geometry. It’s subtle, but visible at corners where sharp edges should theoretically cut through the liquid.

The final section covers spray, foam, and bubbles - secondary particles that bake separately and appear as colored objects in the viewport. These aren’t just cosmetic; they’re a separate physics calculation that adds convincing surface detail without requiring impossibly high particle counts in the main simulation. The trick is that they generate based on the motion of the primary liquid - where the surface breaks, where velocity exceeds thresholds - so they read as physically connected rather than random decoration.

Recommended Gear for This Tutorial

Watch the Tutorial

This video won’t give you a portfolio piece. What it gives you is bookmark-worthiness - a reference you’ll return to when you’re ten parameters deep in a simulation and can’t remember whether flip ratio or particle radius controls splash size. The next video in the series promises actual application - beach, waterfall, or liquid chocolate - applying these settings in context. For now, this is the dictionary.

Level Up Your Blender Setup

Bottom line: Mantaflow is capable of impressive results, but only if you understand what each knob does. This tutorial saves you from the trial-and-error bake cycle that destroys evening projects. Watch it before your next fluid sim, take notes, and refer back when you’re wondering why your ocean looks like soup.