VFX review is a version-control problem wearing an artistic costume: a campaign might carry forty shots through five iterations each, and the workflow that survives that volume is not "more meetings" - it is per-shot versions, frame-accurate annotations pinned to the exact artifact, and one explicit approval per shot. This guide covers the shot review loop from turnover to final, the annotation discipline artists can actually act on, the different failure modes of motion graphics rounds, and the honest line where a review link stops being enough.
Why VFX review is harder
The math is the problem: shot count times iteration count. Twenty CG shots at four versions is eighty reviewable artifacts, each needing notes that reference the exact frame and the exact version - "the tracking slips at 1:12 on v3" fixes; "the shot still looks weird" costs a phone call and a day. Add the feedback-chain reality (client notes supervisor notes artist) and every imprecise note compounds. Precision is not pedantry here; it is the only force that keeps eighty artifacts converging instead of orbiting.
The shot review loop
1. Turnover. Editorial locks the plates: shot IDs, handles, reference frames, and the cut context each shot lives in. A clean turnover is the cheapest schedule insurance in VFX - every downstream ambiguity traces back to it.
2. Artist version. Each work iteration uploads as a version of its shot - v001, v002 - never overwriting. The version stack is the shot's history, and notes attach to the version they describe.
3. Internal pass. The supervisor or lead reviews on the link before the client ever sees it - client rounds should see convergence, not process. Frame-accurate annotations at this stage are cheap; the same notes discovered by the client are expensive.
4. Client pass. Client stakeholders review converged versions with the cut context visible (a shot in isolation reads differently than in sequence). Notes pin to exact frames; the supervisor translates and batches them for artists.
5. Explicit approval per shot. "Approved" is a state on the version, not a nod in a call - and final approval happens in cut context, because a shot that works alone can break a sequence. The video versioning guide covers the version/approval mechanics in depth.
Annotations artists can act on
The annotation standard that works: circle the artifact - a drawing on the frame beats a paragraph; name the frame - the exact timecode, so nobody scrubs to find it; name the version - notes belong to v003, not "the shot"; reference the plate when the problem is upstream ("the plate's shadow direction fights the comp at 0:04"). This is the full annotation discipline from timecoded comments vs frame-accurate annotations - in VFX it stops being a nice-to-have, because imprecision at eighty-artifact scale is schedule death. Shot-level precision vocabulary is also why the animation-studio persona runs a nearly identical loop - see video review software for animation studios.
Motion graphics rounds are different
MoGraph review fails differently than CG: the risks are text accuracy (a wrong price, a misspelled name, an outdated claim - the client's legal exposure, not the artist's), brand compliance (exact colors, fonts, logo clear space), and timing-to-voiceover (animations landing on the beat of the VO). Build the round around those: a text-accuracy pass where a stakeholder reads every word against the source doc, a brand pass against the guidelines, and a sync pass with audio. The annotation habit is the same - circle the wrong digit, pin the frame - but the checklist is the value, because moGraph errors are invisible to "does it look good" review.
When you outgrow a review link
The honest line: heavy film-scale VFX pipelines - shot databases, per-department workflows, USD/scene graph state, render-farm integration - run on specialist tools (the cineSync/ftrack/ShotGrid class) that review links do not replace, and pretending otherwise wastes everyone's time. But agency and studio-scale work - brand campaigns, mograph packages, title sequences, episodic promos - fits a review tool's loop completely: per-shot versions, frame-accurate annotation, threaded notes, approvals on record. The test is whether your pipeline needs shot database state or just shot review state; most commercial work is the second, and the review-side precision vocabulary - frame-accurate review - is the same either way.
How this runs in Frooty
Every mechanism in this guide is review-side and ships in the product: versions per shot with frame-accurate comments attached, drawing annotations on paused frames, threaded resolution so each note closes in place, explicit approval states per version, and reviewer links that cost nothing extra - clients review without seats. The capability list is on the review and approvals page; the version/approval mechanics are in the versioning guide.
FAQ
How do clients review VFX shots?
Through review links organized per shot: each shot's versions stack in one place, comments pin to exact frames, and the client reviews converged versions in cut context. Internal passes happen before client rounds so clients see convergence, not work-in-progress. Approval lands per shot, on the version, on the record.
What is a turnover in VFX?
The handoff from editorial to VFX: locked plates with shot IDs, frame handles, reference frames, and the cut context each shot sits in. Everything downstream - tracking, comping, review references - inherits from the turnover, which is why a clean one is the cheapest schedule insurance in the pipeline.
How many versions does a VFX shot typically go through?
Somewhere between three and six is normal for a commercial shot, but the honest answer is "until approved" - what matters is that every version is kept, notes attach to the version they describe, and approval is explicit per version. Version count is a fact of the work, not a quality signal either way.
Do I need special software to review VFX shots?
For film-scale pipelines with shot databases and per-department state, yes - the cineSync/ftrack class exists for that. For agency and studio-scale work (campaigns, mograph, titles), a review tool with per-shot versions, frame-accurate annotation, and approval states covers the loop. The test: do you need shot database state, or shot review state?