How to Budget an Interactive Film Game: Look Beyond Finished Minutes

An interactive film game cannot be budgeted solely by the length of a single playthrough. The real cost driver is the “total volume of unique content”: everything that must be written, rehearsed, filmed, edited, encoded, localized, and tested, multiplied by branching complexity and change risk. The budget should show cash, person-days, and usage of critical resources together.

The first is typical playthrough minutes, for the player experience; the second is total unique media minutes, for post-production and storage; the third is shooting pages and setup count, for on-set production; the fourth is nodes and combinations of conditions, for programming and testing. The same work might take 90 minutes to complete yet contain 240 minutes of unique video and hundreds of test paths.

Publicity can state the number of endings, but internal budgets must break this down into unique scenes, shot variants, and reusable core material. A change to epilogue text and an entirely new scene cannot both be counted simply as “one ending.”

Development includes research, structure, scripts, prototypes, and tools; pre-production includes casting, locations, art direction, costumes and makeup, storyboards, rehearsals, and legal clearances; filming includes personnel, equipment, locations, transportation, catering, insurance, and overtime; post-production includes editing, sound, color grading, visual effects, subtitles, and encoding; software includes the player, UI, state management, saves, analytics instrumentation, and platform integration; release includes testing, localization, store assets, ratings, marketing, and support.

For each item, record the quantity, unit price or person-days, owner, dependencies, and basis for the estimate. A lump-sum figure without an explainable basis makes later trade-offs difficult.

Label what each branch adds: dialogue only, performance variants, separate shots, separate scenes, actors, locations, state logic, or test combinations. Do not count shared core material twice; calculate the entry and exit segments separately. This lets the team immediately see the cross-department impact when a writer proposes a new branch.

A thirty-second exterior branch may require half a day to move locations, permits, and nighttime equipment, making it far more expensive than two minutes filmed in a studio. Even with equal running times, setup costs can differ completely, so the budget unit should be “scene setup + shot variants + finished footage,” rather than duration alone.

An unlocked script, an unvalidated technical prototype, tight actor availability, and uncertainty about the provenance of AI assets all increase risk. Create a risk register for known uncertainties: probability, cost impact, trigger signals, mitigation measures, and owner. A contingency reserve should cover these specific risks, rather than being an arbitrary percentage added on top.

Increase estimate precision progressively across stages. At the concept stage, provide ranges and assumptions; after prototyping, update estimates using measured filming efficiency and first-frame performance; establish the filming baseline only after the script is locked. Preserve the reasons for differences in every version.

A founding team may reduce cash spending by contributing its own time, but time is still a cost. Record internal person-days at their opportunity cost to compare outsourcing with doing the work in-house. Too many parallel tasks for key team members can also create scheduling bottlenecks, even when no payment is required on paper.

Also record equipment capacity, location availability windows, actor-days, storage, and bandwidth. Once these hard constraints are exceeded, a small budget increase may not be enough to resolve them.

The minimum version delivers the core promises and a complete experience; the target version achieves the intended quality; the expanded version adds high-value content that can be removed independently. These three versions are not a uniform 20% cut: they specify which branches, scenes, and platforms can be dropped while keeping the narrative intact.

The minimum version of 《零点回拨》 could retain one main location and three endings; the target version adds the journalist subplot and a second location; the expanded version adds a separate perspective for a second playthrough. This enables structured decisions when funding changes, instead of hastily cutting ten seconds from every scene.

Calculate the all-in cost per unique finished minute, minutes delivered per shooting day, and person-hours for integrating and testing each node. These metrics should not be compared blindly across projects, but they can reveal anomalies within this project: if a subplot consumes 15% of the budget, is reached through very few paths, and serves no thematic purpose, its justification needs to be revisited.

Use conservative, base, and optimistic scenarios for revenue forecasts, and deduct platform revenue shares, taxes and fees, refunds, marketing, and ongoing support. Budget decisions cannot rely on becoming a guaranteed hit.

After locking the baseline, submit a change request for new content: creative benefits, affected assets, person-days, cash, schedule, and the priority items it will replace. No small change is free; even one new line of dialogue may require an actor to return to set, subtitle updates, and regression testing across all paths.

Update forecasts weekly using three columns: “budget spent,” “committed,” and “needed to complete.” Looking only at payments already made understates signed contracts and unfinished work. When deviations exceed the agreed threshold, the team must immediately choose to reduce scope, revise the approach, or add resources, rather than waiting until post-production to discover that the funds can finish only half the work.

Next step: measure the current version using the four measures of scale, then break down the cross-department incremental costs of the ten most expensive branches. Use this to build minimum, target, and expanded versions, rather than quoting a total first.