Imagine two animators join the same production on the same day.
Over 6 months they each finish about a hundred shots. Same supervisor, same rigs, same pipeline.
By the wrap party, one of them is animating at a visibly higher level while the other didn't progress at all.
The difference was not talent. It wasn't the volume of work either.
The one who progressed opened every shot with something they wanted to experiment with, instead of just finishing as fast as possible.
Repetition doesn't make you better, it's progressive overload: the intention of aiming for higher quality during common practice.
It costs no extra hours. Your tracker already holds the data: every version you posted, every annotation drawn on a frame, every round trip before a task went green. Attaching an objective for deliberate practice to a shot takes 90 seconds and turns work you were doing anyway into training.
What deliberate practice actually means
Psychologist Anders Ericsson studied how people reach expert performance, in violinists, chess players and surgeons, and found that improvement comes from a particular structure of practice rather than from accumulated hours. Four things have to be present: a specific goal instead of a general intention to improve, work pitched just past what you can already do reliably, feedback that arrives soon enough to connect to what you actually did, and reflection on it before the next attempt.
The popular version compressed all of that into a famous number of hours (10,000 hours), a reading Ericsson spent much of the rest of his life objecting to. Hours were never the mechanism.
Naive practice is doing the work until it's approved. Deliberate practice is doing the work to find something out.
Naive practice is not laziness. It's what a schedule rewards and how producers are programmed to track progress.
- Deadlines make the shot the unit of work rather than the development of your skills. With eleven shots to land before the Thursday delivery, "finished" is the only state anyone tracks, and the fastest route to get things done is the approach you already trust.
- Approval is binary. All shots end up green on the board, and nobody asks what the four rounds taught you, even though those rounds are the most expensive thing on the schedule.
- And feedback reads as fixes: "the foot pops on the settle at 31, hold it two more frames". But it's also a lesson about how long you let a weight land before you move off it, but nobody says the second half out loud. Supervisors are solving the shot and rarely teaching the animator. Better-framed notes help, but even the best annotations are aimed at the shot.
You cannot change the deadline or make approval stop being binary. You're responsible, however, about how to make the most of your time while you're working on a shot.
The method: one question per shot
Before you open the scene file, write one sentence naming what you want this shot to teach you.
Not a goal for the shot ("get approved"). A goal for you, running alongside.
Keep it narrow enough. "How do I get better at weight?" cannot be answered by a single shot of a character. "Does starting the hip drop two frames before the foot contacts read as heavier?" is actionable.
Draw the assignment from where your weak spot. You can look up your own retake history to figure it out. If three separate supervisors wrote some version of "the head lead feels mechanical," this is it.
At approval, write the answer down in three lines. It can be as short as "I didn't find out" (when the shot changed, or the note came back about something else entirely). Recording it honestly stops you from believing you tested something you didn't.
Then let the answer set up the next one. If two frames of early hip drop fixed the floatiness, the next question is where that stops working: does the same trick hold on a run, where the contact is much shorter? Questions that chain compound into systems you can reuse across your craft, throughout your entire career.
What makes a good question
A good question is specific and observable. You should be able to look at a playblast and say yes or no.
- Weak: "How do I make my jumps better?"
- Strong: "Does holding the anticipation two extra frames make the jump read better?"
or,
- Weak: "How do I animate more appealing faces?"
- Strong: "If I offset the brow from the eyelid by two frames, does the blink stop reading as a twitch?"
A good question targets a weak spot. The shots where you already know the answer feel better to work on but teach you nothing. If you are strong on facial and shaky on body mechanics, the question goes on the body mechanics even when the shot is a close-up.
A good question fits the shot's schedule. If answering it needs an extra day on a shot budgeted for two, it's a risk for the delivery.
Example questions by discipline
The pattern is the same in every department: something you suspect about your own work, testable in the shot in front of you.
Animation:
- Can I block this shot with fewer keys and still get the pose read on the first pass?
- Where does the weight shift actually start in this action, and does starting it earlier fix the "floaty" note?
- Can I get this lip sync approved without a retake on mouth shapes?
Layout:
- Does this lens choice support the emotional beat, or am I defaulting to my usual focal length?
- Can I stage this so the eye lands on the subject within the first twelve frames?
- How much does camera height change how the director reads the character's status?
Compositing:
- Can I match black levels to the plate without relying on a grade node at the end?
- Which single change fixes "doesn't sit in the plate": light wrap, grain, or edge softness?
- Can I build this comp so a late CG update drops in without rework?
Rigging:
- Which controls did animators actually use on the last show, and which were ignored?
- Can I cut the number of retakes caused by deformation at extreme poses?
- Does this rig evaluate fast enough that animators stop asking for a proxy?
Mining your shot history as raw material
"Kitsu is the project's collective memory."
Axel Tillement, CG Supervisor at Cube Creative
You don't have to invent questions or think too hard. A production tracker records everything that went wrong and got fixed:
- Version history. Step back through a finished task's previews from v1 to final. The delta between them is a list of things you did not see the first time. In Kitsu, we built the review engine to keep every version on the task and put two side by side in full screen to make the changes obvious.
- Comments and annotations. Look for recurring comments and annotations across a batch of your shots.
- Retake counts. Production reports break retakes down by task type and shots, so you can see which work consistently takes you more rounds.
- Time spent against estimate. Timesheets and quotas show where you are consistently over. Sustained overrun on one task type usually denotes a skill gap.
Logging what you learned
A personal log turns a hundred separate answers into a readable journal. For example, six fields per entry:
- Shot - the task ID, so you can find the versions again
- Question - the sentence you wrote before starting
- What I tried - the actual change, in specifics: "hip drop moved from frame 12 to frame 10"
- What happened - the supervisor's response and the round count
- The answer - yes, no, or didn't find out
- Next steps - what next questions the answer opens up
Keep it somewhere you will actually reach. A comment on the task has the least friction, since the entry sits beside the versions and notes it refers to, and task comments in Kitsu are already part of your approval routine. A personal doc works if you would rather keep experiments out of the production record.
Don't spend too long on each entry to make it a consistent habit. A log that takes ten minutes is easy to abandon when the schedule tightens.
Read it back regularly, looking for repeats: the same question asked three times means you have not answered it, and a run of "didn't find out" means your questions are too broad. It's a post-mortem at the scale of one artist.
Tagging entries by skill lets you group all your questions together and see whether the answers can be sharper.
Bringing it to the team
A team gets more out of this habit at almost no cost to supervisors.
Supervisors can add a question to the first-pass comment instead of only listing fixes.
Dailies can have a short "what did this shot teach you?" section: thirty seconds per artist, after the review notes, to surface answers other people can use.
"Once we can get everyone involved to give visual feedback, we get team ownership of the shots and we build a communication tool, a language that the whole team shares."
Chris Unterberg, Founder & Animation Supervisor at Tetsuo Animation
Leads can read across artists. When four people log questions, the shared weak points stand out, and a half-day on what four people are struggling with beats a generic workshop. Juniors can use the format to improve their onboarding.
You just need to keep in mind deliberate practice is about growth, not surveillance. The moment retake counts become a performance score, artists pick safe questions or stop logging honestly, and the data stops being worth reading.
Conclusion: the difference was the intention
Think back to the story about the two animators. Every shot you touch already generates feedback. Somebody looks at it, forms a judgement, and tells you what to change, and your tracker keeps all of it: the versions, the drawings on the frame, the round counts, the hours. You just need to decide what you want to find out to start your own deliberate practice routine.
Pick a shot. Write down a question, and write the answer underneath it when the task status goes green. Start simple.
