Action-conditioned gameplay trajectories for world models and game-playing agents. Align rendered video with player inputs, camera motion, state, events, rewards, goals, and outcomes.
01
A usable gameplay trajectory keeps rendered video synchronized with keyboard, mouse, or controller inputs and the resulting player, object, and environment state.
Datoric structures episodes as state, action, and future-state sequences, with camera motion, events, goals, rewards, and success or failure outcomes aligned on the same trajectory.
02
Environment and task coverage are selected for the target workflow, from navigation and third-person action to platforming, driving, puzzles, resource collection, inventory, exploration, and multi-step goals.
03
Rendered frames show what the agent sees. State and event logs explain the transition: camera pose, player position and velocity, inventory, status, object interactions, goal progress, rewards, and outcomes.
Optional depth, segmentation, object, minimap, NPC, collision, physics, and caption fields are subset-specific and identified during delivery review.
04
QA checks action-state synchronization, frame rate, input and state frequency, goal completion, reward signals, and metadata consistency. Corrupted, malformed, and duplicate episodes are filtered, with rights-clearance review included.
05
Inspect complete trajectories and confirm which environments, inputs, state fields, episode lengths, and frame rates are represented in the selected delivery.
06
Choose target environments, input methods, state and event fields, goals, reward structure, and evaluation criteria. Datoric scopes the selected delivery and any new collection around those requirements.
Program standards
Frequently asked
It is gameplay video aligned with the actions and state transitions behind it. Each trajectory can pair frames with keyboard, mouse, or controller inputs, camera motion, game state, events, goals, rewards, and outcomes.
Passive video shows what happened. Action-conditioned trajectories also preserve the inputs, state, and resulting future state, so a world model or agent can learn how actions change an environment.
Available coverage includes first-person navigation, third-person action, platforming, driving, puzzle solving, resource collection, object interaction, inventory, exploration, and multi-step goals.
Yes. The buyer review package can include real gameplay video, input logs, state logs, QA summaries, and licensing documentation for the selected environments.
Tell us the environments, controls, state fields, and outcomes your model needs, and we will scope an action-conditioned gameplay program around them.
Datoric