graph editor habits that actually improved my animation

31 views 2 replies

I want to talk about the Graph Editor — specifically how underutilized it is by animators who came up through blocking-first workflows. For a long time I treated it as a cleanup tool, something you visited after the poses were solid. That turned out to be a pretty limiting mindset.

The shift for me came when I started working in Blender alongside colleagues who had Maya backgrounds. They were shaping curves as they were blocking, not after. Watching them manipulate the velocity of a movement directly — flattening a curve to create a hold, pushing a tangent to juice an ease-in — made me realize I had been doing an extra round of polish that was really just correcting poor curve decisions I had baked in earlier.

A few habits that changed how I work:

First — default to flat tangents on every new key. Auto-tangents in Blender and Maya are convenient but they introduce overshoot you didn't ask for. Starting flat and building the curve intentionally forces you to make a decision about every transition.

Second, check your curves before you check your viewport. If the curve looks wrong, a bump where there should be a smooth arc or a sharp spike on a rotation channel, the viewport is going to confirm it. Fixing the data is faster than tweaking poses until the visual problem disappears.

Third, separate your channels when things get complex. Stacked curves on a single graph panel hide problems. Isolating the Y translation of a hip bone will immediately show you whether your weight shift is actually reading the way you think it is.

I've heard Cascadeur pushes some of this further with its physics-assist tools, though I haven't had time to dig into it seriously. Would be curious whether it changes how people think about curve work or just abstracts it away entirely.

What are your Graph Editor habits — things you do consistently that you'd consider non-negotiable at this point?

the habit that changed how i work in the graph editor more than anything else was stopping relying on the dopesheet to set timing and then going to the graph editor to fix the feel. now i do timing and spacing in the graph editor from the start, even when the keys are still rough.

it forces you to think about velocity rather than just position. if you place a key and immediately look at the curve slope, you start to internalize what that slope means for the actual motion, whether it's going to ease in, snap, float. the dopesheet abstracts that away and you end up doing more corrective work later.

also: weighted tangents. if you're still on auto tangents for most of your work, switch. the time you spend managing them manually pays back when you stop fighting the curves to do what you already know you want.

the point about working in normalized time in the graph editor is something i picked up from a mentor and it changed how i think about spacing. blocking in stepped, then switching to linear just to check timing before going spline — that order matters more than i used to give it credit for. the mistake i kept making was going spline too early and then losing track of what the timing actually was under all the easing.

one habit i'd add: i keep a second NLA track with the original stepped block as a reference strip, muted, so i can toggle it back in if the spline pass drifts away from the intent.