What to Approve Before Final Render
An architectural animation project has three approval gates: the shot list, where structure and running time are settled at zero cost; the rough cut, where pacing and order are confirmed before any final rendering; and the final pass, which should be limited to colour, sound and small corrections. Approving only the finished piece means approving after every expensive decision has already been made.
Most buyers of architectural animation approve once, at the end, when the piece looks finished. It feels like the responsible moment to give an opinion and it is the worst one, because by then the structure is fixed, the frames are rendered and every meaningful decision has already been made and paid for.
The format has three natural approval points, and only the last one is obvious. The other two happen when the work looks unfinished, which is exactly why they get skipped.
This guide sets out what belongs at each gate, what it costs to move a decision past its gate, and how to run a review when the thing being reviewed does not look like the thing you are buying.
Why the gates exist at all
Because cost in animation is not distributed evenly across the timeline. It concentrates sharply at the end, when rendering begins.
Before rendering, changes are conversation. After rendering, changes are computation, and every second of finished footage carries hours of it.
So the gates are placed exactly where the cost of reversing a decision jumps, which is the only sensible place to put an approval.
Gate one: the shot list
What gets approved here is the structure: which shots exist, in what order, how long each holds, and how long the whole piece runs.
This is the cheapest gate by an enormous margin. Reordering a list, cutting three shots or halving the running time costs nothing at this stage and costs a great deal at any later one.
It is also the gate most often skipped, because a numbered table does not look like a deliverable and approving one feels premature.
What to actually check on a shot list
That each shot has a stated reason for existing. A shot without a reason is travel time, and travel time is what makes a piece feel long.
That the running time matches the channel where the piece will play, rather than matching how much building there is.
And that the opening establishes where this is before anything else happens, because a viewer who does not know what building they are looking at spends the first seconds working it out instead of watching.
The question to ask before approving the list
What is a viewer supposed to understand by the end. If nobody on either side can answer in one sentence, the list is a sequence of views rather than a piece, and approving it locks in that problem.
This is not a philosophical question. It determines which shots survive the cut and which are decoration.
Suppliers producing architectural animation services will happily build whatever is approved, so the responsibility for asking sits with the buyer.
Gate two: the rough cut
A grey, unlit, untextured version of the entire sequence with real timing. It looks crude and it is the most valuable thing you will see in the whole project.
What gets approved is pacing, order, transitions and length. Everything structural, confirmed in a form that costs almost nothing to change.
A supplier who does not offer one is either extremely confident or planning to charge for the consequences, and it is worth establishing which before signing anything.
How to review a rough cut without being distracted
The difficulty is real: it is hard to judge pacing when everything is grey and nothing looks finished. Reviewers instinctively comment on the appearance instead.
The discipline is to watch it three times and only ever comment on timing, order and length. Appearance is not on the agenda at this gate, and saying so out loud before playing it prevents most of the wasted discussion.
A useful test is to watch it once at double speed. Structural problems are more obvious when the content is stripped of detail and the rhythm is exaggerated.
The three questions the rough cut answers
Is it too long. It almost always is, and this is the only moment where shortening it is free.
Does anything repeat. Two shots teaching the same thing is the most common structural redundancy and it is invisible in a shot list.
Does it end or does it stop. A piece that stops because the building ran out reads as unfinished no matter how polished the frames become.
Gate three: the final pass
What belongs here is colour, grading, sound balance, small material corrections and text or branding placement.
What does not belong here is structure, camera position, running time or building geometry. Those were settled at gates one and two, and reopening them at this point means re rendering.
Stating that boundary in advance is not defensiveness on the supplier side. It is what keeps the last gate cheap for both parties.
What it costs to move a decision past its gate
A running time change at the shot list is free. At the rough cut it is an edit. After final rendering it is a re render of everything cut or extended.
A camera change at gate one is a line in a table. At gate three it is a new shot from nothing.
A building change at gate one is a modelling adjustment. After rendering it touches every frame the element appears in, which on a 3d walkthrough can be the entire piece.
The gate nobody schedules: the model review
Between the shot list and the rough cut there is a moment worth formalising, which is the first look at the building itself in grey, from the approved camera positions, before any movement is animated. It answers whether the geometry matches the drawings, whether the massing reads correctly and whether the site levels are right, and every one of those is cheap to correct at that point and expensive once cameras are moving through the model.
Practices familiar with still imagery already do this instinctively, because they review geometry before materials. In walkthrough animation the equivalent step gets skipped, because the deliverable is a moving piece and a set of static grey frames does not feel like progress toward it.
It costs a supplier almost nothing to export those frames and it removes an entire category of late discovery, which is the kind of trade worth insisting on even when nobody offers it.
Approving the camera positions specifically
Camera position is the decision that behaves worst when it is left implicit. A viewpoint that reads well as a still can be wrong for movement, because a moving camera has to arrive from somewhere and continue to somewhere, and positions chosen in isolation frequently produce transitions that make no spatial sense.
The practical check is to look at the sequence of positions on a site plan rather than one at a time. If the camera teleports across the site between consecutive shots, the viewer will feel the discontinuity even if each individual frame is well composed.
Suppliers offering 3d walkthrough services will happily animate whatever positions are approved, and correcting a spatial contradiction after rendering means producing new shots rather than adjusting existing ones.
What happens when a gate is skipped
The pattern is consistent enough to predict. Skipping the shot list produces a piece that is too long and has no ending, because nothing ever decided where it should stop. Skipping the rough cut produces a pacing argument at the most expensive possible moment, usually with a launch date approaching.
Skipping the model review produces geometry corrections discovered in finished frames, which is the most demoralising version of the problem because the frames look finished and are wrong underneath.
None of these are supplier failures. In every case the supplier delivered what was approved, and what was approved was the only thing put in front of anybody.
Why unfinished work is hard to approve
There is a genuine psychological obstacle here and it is worth naming. People find it difficult to sign off on something that looks crude, because approval feels like an endorsement of appearance rather than of structure.
The reframing that helps is to treat each gate as approving a decision rather than a deliverable. At the shot list you are approving what the piece will contain. At the rough cut you are approving how long each thing lasts. Neither is a judgement about how it looks.
Saying that out loud before the review begins is a small act of facilitation that reliably improves the quality of the feedback, because it tells reviewers what kind of opinion is useful.
Running the gates against a launch date
Working backwards is the only sensible way to schedule this, because rendering time is real and non compressible. From the launch date, allow time for final rendering, for the final pass review, for the rough cut review and its corrections, and for the shot list to be agreed.
That arithmetic almost always reveals that the shot list needs to be settled weeks earlier than anybody planned, which is uncomfortable and true, and it is far better discovered at the start than at the point the schedule fails.
An architectural animation company that pushes for early structural decisions is protecting your date rather than being difficult about process.
Who should sign off, and who should not
One person with authority to approve, holding the direction and consolidating everybody else. Animation attracts more reviewers than any other visual format because watching requires no expertise.
Each additional reviewer adds an impression without adding information, and contradictory direction is far more expensive here than with still images, because resolving it means re rendering rather than re editing.
Everybody who wants to comment can comment. Only one person decides what is passed to the supplier.
How to write feedback at each gate
At the shot list: name the shot number and state the change. Cut shot seven. Move shot four before shot two. Reduce total to forty seconds.
At the rough cut: use timecodes. The shot at eleven seconds holds too long, take it to four. The transition at twenty two seconds is confusing.
At the final pass: name the frame or the range and the correction. Ambiguity at any gate becomes a guess, and in this format a guess costs a render cycle.
The design change problem
Development designs move late and animation punishes that more than any other format. A facade adjustment after rendering starts is not a revision, it is new work.
The protection is sequencing rather than contract language: lock the elements that appear on screen before final rendering begins, and animate as late as the schedule allows while still finishing before the campaign.
Where that is impossible, agreeing the cost of a late change in advance is better than discovering it a week before a launch.
Approving the variants
If the plan includes several cuts, a launch version, a listing version, a silent loop and a short social version, each needs its own confirmation and they should be approved together.
Approving only the long version and assuming the others follow is how a project ends up with a silent gallery loop whose pacing depended on music that is not playing.
The variants are cheap when produced together and awkward when requested after a project has been closed by an architectural animation company.
The silent version deserves its own review
Because it is a different piece. Most architectural animation leans on music for rhythm and emotion, and removing the track exposes any shot that was holding on atmosphere rather than on content.
Reviewing it silently, in the room where it will actually play if possible, catches problems that never appear in a headphones review at a desk.
This takes ten minutes and it is skipped almost universally.
What to check on the delivered files
Resolution and aspect ratio against each channel. Whether the branded and unbranded versions both exist. Whether the audio is licensed for the intended use, including advertising if that is the plan.
And whether the first frame works as a still, because most platforms will use it as a thumbnail and a thumbnail chosen by an algorithm is frequently the worst frame in the piece.
None of this is interesting and all of it is cheaper to catch now than after publication.
Keeping the approval trail
Record which version was approved, by whom, on what date, and keep the shot list and the rough cut that produced it.
Development projects run for years and pieces get reused, recut and extended across phases. Eighteen months later nobody remembers whether the version in the campaign was the approved one.
It also settles the recurring question of whether something was requested or invented, which is otherwise two memories disagreeing under deadline pressure.
When to stop reviewing
When the remaining comments are preferences rather than corrections, and when feedback starts reversing earlier feedback.
Every project reaches a point where further rounds average the piece rather than improve it. In animation that point arrives sooner than people expect, because the audience watches once, briefly, on a small screen.
The difference between round three and round four of a 3d architectural animation is invisible to everybody except the people who commissioned it.
The cost of an approval that never arrives
Delay is its own expense in this format and it is rarely counted. A supplier holding a rendering queue for an approval that is two weeks late does not simply pause, they reallocate the machines and the people, and rejoining the queue afterwards is not instant.
That is why a launch date can be missed without anybody having done anything wrong: the work was ready, the decision was not, and the slot was gone by the time it arrived.
Agreeing turnaround for approvals, in both directions, is as useful as agreeing turnaround for delivery, and it is almost never in the brief.
What a good approval looks like in writing
Short, specific and dated. Approved the shot list of eleven shots at a total of forty seconds, with the changes to shots four and seven noted below. That sentence protects both sides and takes thirty seconds to write.
What does not protect anybody is a verbal approval in a meeting where six people were present and three of them had reservations they did not voice clearly.
An architectural animation company keeping a written trail is not being bureaucratic, it is preventing the argument that begins with somebody remembering the conversation differently.
The same applies to a rejection. Recording what was not approved, and why, is what stops the same option being proposed again two weeks later by somebody who was not in the room the first time.
Approving work produced by somebody else
Occasionally a sequence arrives from a previous supplier, a partner or an agency, and somebody has to decide whether it is usable. The same gates apply in reverse: work out what its structure was meant to be, whether the pacing holds, and whether it can be recut rather than replaced.
The decisive question is whether the source files exist. A delivered video file can be trimmed and little else. A project with its model, scene and edit intact can be recut, extended, shortened for a new channel and updated when the building changes.
That difference is worth establishing before any judgement about quality, because a mediocre piece with its source files is frequently more valuable than a good one delivered as a flat export.
The approval that protects the schedule
The one nobody writes down is the confirmation that the building is locked. Not the whole design, only the elements that appear on screen: the facade, the visible interiors, the immediate landscape and anything the camera passes.
Getting that confirmed in writing before final rendering starts is the single most effective protection available, because it converts a later design change from an argument into a known and priced event.
Development teams resist it because locking anything feels premature, and the alternative is discovering the cost of a late change with a launch date already announced.
Reviewing on the right screen
A piece reviewed on a large calibrated monitor in a quiet room and then published to a phone in daylight is being judged in conditions its audience will never share. Contrast that looked subtle becomes invisible, and pacing that felt measured feels slow.
Where the piece will play on a sales gallery screen, review it on that screen if it exists, in the light it will actually have, with whatever sound the room provides.
This is a ten minute discipline that catches a class of problem no amount of desk review will surface, and almost nobody does it.
Signing off the thumbnail
Most platforms select a frame automatically and the automatic choice is frequently the worst frame in the sequence: a transition, a blur, an empty corridor.
Choosing the thumbnail deliberately is part of the final approval, because for a large share of the audience the thumbnail is the only frame they will ever see, which makes it functionally the most important image in the entire production.
It costs nothing, it takes a minute, and providers of 3d walkthrough services will supply a still from any frame on request.
The honest summary
Approve the structure when it is a table, approve the pacing when it is grey, and limit the final approval to colour, sound and small corrections.
Name one person to decide, write feedback with shot numbers and timecodes, and lock the building before rendering begins.
Do that and an architectural walkthrough closes in three gates. Approve only the finished piece and every disagreement arrives at the most expensive possible moment.
To run a project with a shot list and a rough cut as real approval gates, request a quote.
Frequently asked questions
What is the most important approval point in an animation project?
The rough cut, a grey unlit version of the full sequence with real timing. Pacing, order and length are still free to change there. After final rendering the same decisions cost hours of computation per second of footage.
Why should a shot list be formally approved?
Because it fixes structure and running time at zero cost. Reordering a list, cutting shots or halving the length is free at that stage and expensive at any later one. It is skipped most often because a numbered table does not feel like a deliverable.
What belongs in the final approval?
Colour, grading, sound balance, small material corrections and branding placement. Structure, camera positions, running time and building geometry do not, because reopening those after rendering means recomputing every frame they touch.
How many people should approve an animation?
One, holding the direction and consolidating everybody else. Animation attracts more reviewers than any other format because watching needs no expertise, and contradictory direction here is resolved by re rendering rather than re editing.
Does the silent version need a separate review?
Yes. Most architectural animation leans on music for rhythm, so removing the track exposes shots that were holding on atmosphere rather than content. Reviewing it silently, ideally in the room where it will play, takes ten minutes and is almost universally skipped.