Architectural visualization lives and dies on iteration rounds: stills and walkthrough animations go out, the client's architects and developers mark up what's wrong, and the studio iterates — five to ten times per project, usually over email threads, WhatsApp screenshots, and cloud-drive folders nobody can reconstruct. This post is the review loop for that: timecoded comments on walkthroughs, drawn annotations on stills, and one version stack per round that both sides can trace.
Why the generic stack fails archviz
Email + Drive + WhatsApp produces three specific archviz failures: the walkthrough note "the camera move at the end feels slow" arrives with no timestamp, so the studio guesses between three shots; still feedback scatters across screenshots ("move this tree left") with no pin on the actual render; and by round six nobody can say which still set is current — the exact version chaos documented in the Drive vs review tool comparison. A dedicated archviz feedback product exists (archvizreview.com built for this niche), and ReviewStudio markets to archviz too — so if your firm lives entirely in stills and wants an architecture-native markup surface, evaluate the niche tools honestly. The loop below is what a video-review-first tool covers, and it is the strongest fit when walkthrough animations are half your deliverable.
The loop, per round
- Upload the round as a version. Stills grouped as a batch, the walkthrough as a video asset — each round one version on the stack, so "v3 stills vs v3 walkthrough" is a fact, not an inbox dig (video versioning guide).
- Timecoded comments on walkthroughs. The client pins "this pan is too fast — 0:22" and "show more of the courtyard here — 1:05"; the studio works the camera notes shot by shot instead of scrubbing for them. This is the same precision logic as timestamped comments vs annotations, applied to camera language.
- Drawn annotations on stills. The client circles the parapet detail and arrows the paving material; drawing on the frame beats describing a pixel region in words — the same mechanic as annotation-first review, where "show me where" replaces three emails of guessing.
- Reviewer lanes for the client side. Architect, developer, and marketing stakeholder comments stay attributable — conflicting notes ("more people in the scene" vs "keep it serene") get reconciled by the decision owner, the same rule as any multi-stakeholder round (see the approval workflow guide).
- Resolve and re-render. Threads close only when the fix ships in the next version — the stack becomes the project's iteration history, which is also what arbitrates "we approved this look" six weeks later.
What to state in the contract
Archviz disputes are usually round disputes, so borrow the proposal language that prevents them: a named number of iteration rounds, a feedback window per round, one decision owner consolidating the client's notes, and the definition of a round (one consolidated markup answered by one new version — the revision-rounds definition applies verbatim). Studios that moved from "unlimited tweaks" to this structure report the same thing other creative fields did: the client's feedback got more organized the moment the process was.
Honest limits
Two things a general review tool does not replace: live, real-time walkthrough sessions with a client on the call (some firms love them; run them in the real-time tool of your choice and treat the async review link as where the actual sign-off lives), and the render pipeline itself — review tools manage versions and notes, not scenes or farms. If your whole deliverable is stills-only and every stakeholder wants to draw in one shared canvas at once, try the archviz-specific tools first and keep the video-review loop for the animation half.
FAQ
How do clients comment on a walkthrough animation with timestamps?
Upload the walkthrough to a review link like any video: the client pauses, pins a comment at the exact moment ("0:22 — pan too fast"), and the studio's camera notes arrive shot-by-shot instead of as descriptions. No plug-ins for the client — a browser link works.
What's the best way to collect still-render feedback?
One review page per round with the still batch as a version, where the client draws directly on each frame. Screenshots-of-screenshots are where archviz notes go to die; drawn annotations keep the note attached to the exact parapet, tree, or material.
Do archviz firms need a special archviz review tool?
Not necessarily. Niche tools exist and suit stills-only shops that want architecture-native markup; once walkthrough animations are a core deliverable, a video-review-first loop (timecoded comments, version stacks, reviewer lanes) covers the whole deliverable in one place.
How many review rounds should an archviz contract include?
Three to five consolidated rounds is common, with the round defined in the proposal: one consolidated markup from one decision owner, answered by one new version. Extras bill as new scope — the definition is what prevents the ninth "small tweak" round.