AI-generated B-roll is arriving on timelines faster than most studios anticipated. Where location shoots, stock libraries, and second-unit crews once filled the gaps between interview segments and hero shots, tools like Runway and Pika are generating passable cutaway footage from text prompts in minutes. For production budgets, that shift is real and measurable. For creative quality, it's more complicated.
What AI B-roll actually costs to produce
The raw cost of AI-generated footage is low. A monthly subscription to a mid-tier generative video tool runs between $50 and $150 AUD. Compare that to a single half-day stock location shoot in Melbourne, which can run $2,500 to $6,000 once you factor in crew, hire, and post. On that arithmetic alone, the case for AI B-roll looks obvious.
It isn't. The hidden cost sits in iteration time. A prompt that sounds precise ("aerial view of a glass office tower at dusk, photorealistic") rarely produces usable footage on the first attempt. Studios report spending two to four hours generating, reviewing, and refining a 10-second clip that a stock library would have delivered in three minutes. That iteration labour has to land somewhere in the budget. If it's absorbed into a producer's day rate, the saving disappears quickly. If it isn't tracked at all, the project looks cheaper than it was.
West Melbourne Studios treats AI B-roll as a tool with a specific use case, not a universal budget lever. The format earns its place in projects where the required footage is genuinely impossible to capture practically: extreme weather, abstract conceptual sequences, historical settings, or scenes requiring crowd simulation at scale. For those use cases, the iteration cost is still lower than the alternative.
Where the quality gap remains visible
Current AI video generators handle texture and light inconsistently. Skin, water, and fabric in motion are the three most reliable tells. A human hand opening a laptop will still look slightly wrong in most outputs today, with fingers that move at an uncanny rhythm or a wrist that bends past its natural range. Viewers don't consciously identify the defect, but they register it as a feeling of unease. That feeling damages the brand impression the video was built to create.
This matters most in corporate and professional services content, where credibility is the product. A documentary about a remote landscape can absorb a slightly artificial sky. A CEO explainer video cannot absorb an AI-generated shot of a handshake that reads as wrong. Understanding how AI-generated voiceovers are changing video production is useful context here: the same quality ceiling applies, where the format works at a distance but breaks under close scrutiny.
The practical rule West Melbourne Studios applies is simple. If a B-roll shot would appear within 2 metres of a human face in the edit, shoot it practically. If it's establishing context, suggesting scale, or illustrating an abstract idea, AI generation is worth testing.
How this changes the pre-production conversation
The decision to use AI-generated B-roll should happen before the shoot, not in post when the budget has already been spent. That means the brief needs to flag it. When clients review a treatment that includes AI-generated sequences, they're entitled to know. Some will be indifferent. Others will object on creative or reputational grounds, and those objections are legitimate.
The production brief is the right place to make this explicit. If you're already thinking through how motion graphics explain complex ideas in video, the same pre-production logic applies to AI footage: decide the use case early, set quality benchmarks before generation starts, and review outputs against the brief rather than against each other.
Studios that skip this conversation create two problems. First, the client discovers the AI content in the review process and questions what else was substituted without disclosure. Second, the production team has no quality standard to work toward, so iteration becomes open-ended and eats the time saving the tool was supposed to deliver.
The talent and workflow implications
AI B-roll doesn't eliminate second-unit shoots. It changes who runs them. The skill set required to prompt, evaluate, and integrate generated footage is different from the skill set required to operate a camera. Editors and post-production coordinators are picking up this work by default, often without adjusted day rates or additional time in the schedule.
Studios need to decide where AI B-roll generation sits in the workflow before it arrives on a project. Is it a producer task? A post task? Does it require a dedicated prompt editor, a role that's starting to appear in production job descriptions? These aren't abstract questions. They determine whether the cost saving lands in the budget or evaporates into untracked labour hours.
West Melbourne Studios runs AI generation as a post task, with the editor given a prompt brief approved in pre-production. That keeps the creative direction consistent with the shoot and prevents the footage from pulling the edit in a direction the director didn't intend.
What to put in the budget line
Budgeting AI B-roll honestly means separating three costs that often get collapsed into one. Tool access (the subscription) is a studio overhead. Iteration time (the prompting and review process) is a line item, billed at the relevant day rate for whoever runs it. Integration time (colour matching, grain addition, motion blur adjustments to make AI footage sit alongside practical footage without jarring) is a post-production line item, typically 30 to 60 minutes per 10-second clip depending on the grade.
When those three costs are tracked separately, the real saving from AI B-roll becomes visible. In the right use cases, it still exists. In the wrong ones, it doesn't, and the budget is better spent on a focused half-day shoot that delivers footage the edit can use with confidence.
The tools will improve. The quality ceiling on AI video generation is rising fast, and what reads as uncanny today will likely read as acceptable within 18 months. But the workflow, disclosure, and budgeting disciplines required to use it well are the same regardless of how convincing the output becomes. Studios that build those disciplines now will use the better tools more effectively when they arrive.