Define the decision the budget should support
Start with one uncertainty, such as whether a character handover can read clearly across two shots. Name the acceptance criteria and the point at which you will stop to review. A budget without a decision question can become permission to keep generating variations without learning from them.
Keep creative changes separate from repeated attempts at the same requirement. Changing the character design halfway through a test may be justified, but it changes what the original resource estimate can tell you. Record the new assumption before continuing.
Copy the budget and attempt record
Use the current estimate shown for the actual action you plan to take. If the estimate is unavailable or variable, mark that uncertainty rather than inserting an invented rate. Track human review effort separately from generation usage so neither disappears from the plan.
The planned subtotal is candidate count multiplied by the estimate per action when that estimate is applicable. Add revised estimates as separate rows instead of silently replacing the original assumption.
PROJECT / TEST ID / DATE:
QUESTION TO RESOLVE:
ACCEPTANCE CRITERIA:
CURRENT PRICING OR ESTIMATE CHECKED AT:
ACTION / MODEL OR WORKFLOW CHOICE:
PLANNED CANDIDATE COUNT:
CURRENT ESTIMATE PER ACTION:
PLANNED SUBTOTAL / UNCERTAINTY:
REVISION ALLOWANCE / REASON:
REVIEW OWNER / TIME ALLOWANCE:
STOP RULE / RESOURCE CAP:
ATTEMPT ID / INPUT VERSION:
ACTUAL USAGE:
ACCEPTED / REJECTED / NEEDS REVIEW:
REASON / OBSERVATION:
NEXT DECISION:
TOTAL OBSERVED USAGE:
WHAT THIS TEST DOES AND DOES NOT ESTABLISH:Worked illustration using deliberately invented units
The numbers below demonstrate the arithmetic only. They are invented planning units, not Wrong Opera credit prices, expected success rates or a production recommendation. Replace them with the current estimate for your actual workflow.
The example keeps its first attempt allowance separate from a possible revision. A review happens before the additional allowance is used.
Use observed usage to improve the next estimate
After the test, record what failed and what changed. A candidate rejected because the brief changed should not be grouped without explanation with one that failed an unchanged requirement. Those categories lead to different planning decisions.
Carry useful observations into a similar next test while retaining uncertainty for different actions, characters and formats. Check current pricing and estimate information again before committing. A worksheet is a record of assumptions and evidence, not a promise that a whole film will scale directly from one successful shot.
Questions & answers
Should I budget only accepted outputs?
No. Planned attempts, revisions and review work all affect the experiment. Record accepted results separately so you can understand the difference between generation activity and usable production material.
Can this calculate the cost of a finished minute?
Only if you supply relevant observed data and clearly state its limits. A single test does not establish a reliable finished-runtime rate across different scenes, models or creative requirements.
Explore the workflow with your own idea.