Separate accepted footage from generated footage
Suppose a thirty-second advertisement needs three five-second generated inserts. The accepted target is fifteen seconds. Planning three candidates for each insert instead produces nine initial candidates and forty-five generated seconds. This is a hypothetical production plan, not evidence that three candidates will guarantee an acceptable shot. Define the visible acceptance condition, unacceptable defects and a live-action fallback for each scene. Increasing candidate volume before the scene has a clear purpose makes it difficult to decide where the generation budget is best spent.
Fix the model and settings
The official pricing page illustrates different generation quantities for different models. Match the calculation to the intended model and relevant production settings, including source method, duration and resolution where applicable. Credits for different models do not represent one interchangeable seconds price. Specify the route used for each scene. If an inexpensive exploration stage is followed by a different final-generation stage, budget both explicitly. When changing models, update the conversion rather than continuing to estimate remaining capacity from the previous model assumptions.
Do not turn a trial balance into a recurring allowance
The Free offer describes its credits as a one-time allowance. Do not use an unspent trial balance as evidence of an equal refill next month. For continuing production, identify recurring allocations, purchases, expiration and rollover for the actual contract. Initial exploration may fit within a trial while release corrections and the following assignment do not. Keep free exploration, ordinary monthly access and an annual commitment separate in the comparison sheet. The useful trial result is a clearer acceptance process that can inform the recurring workload estimate.
Use rejection reasons to decide the next action
Classify rejected candidates by the actual problem: anatomy, product representation, motion continuity or framing. A stylistic mismatch and a false representation of a product require different responses. For the latter, live footage or a diagram may be a better choice than concealing the defect in editing. Set a stopping rule for repeated failures instead of iterating until the balance is exhausted. Previously spent credits are not a reason to continue. Each proposed generation should have a clear explanation of how it could move the scene toward the accepted deliverable.
Budget review time before the deadline
Short generation waiting time does not eliminate candidate inspection, caption integration or final sound and image adjustment. Work back from the deadline and separate framing decisions, candidate creation, acceptance and exported-file review. Where concurrency is a contract feature, consider how much material the reviewer can inspect at once. An actual pilot should record waiting and human work separately. Until that pilot exists, estimated times should remain planning assumptions rather than public performance claims. Sufficient credits alone do not establish that the proposed production schedule is workable.
Set a scene-level spending boundary
A shot showing a product accurately has a different acceptance standard from an atmospheric insert. Give each scene an initial candidate count and an additional-generation ceiling. One difficult scene should not silently consume the entire assignment budget. If it approaches the ceiling, compare a shorter shot, live footage or a diagram. The objective is a complete explanation delivered on time, not maximizing the number of generated seconds.
Describe what extra capacity would produce
Before buying more capacity, identify the unfinished scenes and additional candidates it would support. A single temporary assignment differs from an ongoing increase in monthly demand. Compare the charge payable now, the validity of extra capacity, the contract period and required delivery capabilities rather than monthly equivalents alone. Retain selected footage, source images, prompts, model information and acceptance reasons. This record supports a more informed estimate for the next assignment and helps decide when generated inserts should be replaced by live footage instead of another round of unstructured exploration.
Official sources
Pricing, availability and feature limits can change. Check the linked official pages before subscribing.
- Runway pricingInformation checked: 2026-10-26
- Free plan detailsInformation checked: 2026-10-26
- Commercial use of generated outputInformation checked: 2026-10-26