Skip to content

unienv_interface.wrapper.video_record

Classes:

Name Description
EpisodeRenderStackWrapper
EpisodeVideoWrapper
EpisodeWandbVideoWrapper

Functions:

Name Description
image_specs_from_space

The image leaves of space declared through metadata["kind"] == "image".

image_to_hwc_rgb

Convert an image described by layout / format metadata to (H, W, 3) RGB.

Attributes:

Name Type Description
ImageSpecs

ImageSpecs module-attribute

ImageSpecs = Dict[Optional[str], Tuple[Optional[str], Optional[str]]]

EpisodeRenderStackWrapper

EpisodeRenderStackWrapper(
    env: Env[
        BArrayType,
        ContextType,
        ObsType,
        ActType,
        RenderFrame,
        BDeviceType,
        BDtypeType,
        BRNGType,
    ],
    *,
    done_index: Optional[int] = None,
    rendered_index: Optional[int] = None,
    nested_separator: str = ".",
    frame_source: Literal["render", "observation"] = "render",
    frame_space: Optional[Space] = None,
)

Bases: Wrapper[BArrayType, ContextType, ObsType, ActType, RenderFrame, BDeviceType, BDtypeType, BRNGType, BArrayType, ContextType, ObsType, ActType, RenderFrame, BDeviceType, BDtypeType, BRNGType]

Initialize the wrapper. For a batched environment, done_index is used to check episode termination on a specific batch index. If rendered_index is provided, render outputs that are batched tensors are indexed at this batch index before caching/logging.

Frames are taken from env.render() (frame_source="render") or from the observations returned by reset / step (frame_source="observation"). frame_space describes the frames; it defaults to env.observation_space for observation frames. When its leaves declare Space.metadata, only the leaves with kind == "image" are recorded, converted to (H, W, 3) RGB according to their layout ("CHW" is moved channel-last) and format ("bgr", "rgba", "gray", ...). Without any kind metadata, every render frame is recorded as an RGB image (and, for observation frames, every (H, W, 3) leaf).

Methods:

Name Description
step
reset
render
close
sample_space

Sample from space using and updating self.rng.

sample_action

Sample one action from action_space.

sample_observation

Sample one observation from observation_space.

sample_context

Sample one context value if context_space is defined.

update_observation_post_reset

Merge masked reset observations back into a full batched observation.

update_context_post_reset

Merge masked reset contexts back into a full batched context.

has_wrapper_attr

Checks if the given attribute is within the wrapper or its environment.

get_wrapper_attr

Gets an attribute from the wrapper and lower environments if name doesn't exist in this object.

set_wrapper_attr

Sets an attribute on this wrapper or lower environment if name is already defined.

Attributes:

Name Type Description
done_index
rendered_index
nested_separator
frame_source
frame_space
episodic_frames Union[List[Any], Dict[str, List[Any]]]
video_episode_num
video_step_num
video_episode_start_step
metadata Dict[str, Any]

Returns the :attr:Env :attr:metadata.

render_mode Optional[WrapperRenderFrame]
render_fps Optional[int]
backend ComputeBackend[Any, WrapperBDeviceT, Any, WrapperBRngT]
device Optional[WrapperBDeviceT]
batch_size Optional[int]
action_space Space[WrapperActT, WrapperBDeviceT, Any, WrapperBRngT]
observation_space Space[WrapperObsT, WrapperBDeviceT, Any, WrapperBRngT]
context_space Optional[Space[WrapperContextT, WrapperBDeviceT, Any, WrapperBRngT]]
rng WrapperBRngT

Returns the :attr:Env :attr:rng attribute.

unwrapped Env

Returns the base environment of the wrapper.

prev_wrapper_layer Env[BArrayType, ContextType, ObsType, ActType, RenderFrame, BDeviceType, BDtypeType, BRNGType]
env

done_index instance-attribute

done_index = done_index

rendered_index instance-attribute

rendered_index = rendered_index

nested_separator instance-attribute

nested_separator = nested_separator

frame_source instance-attribute

frame_source = frame_source

frame_space instance-attribute

frame_space = frame_space

episodic_frames instance-attribute

episodic_frames: Union[List[Any], Dict[str, List[Any]]] = []

video_episode_num instance-attribute

video_episode_num = 0

video_step_num instance-attribute

video_step_num = 0

video_episode_start_step instance-attribute

video_episode_start_step = 0

metadata property writable

metadata: Dict[str, Any]

Returns the :attr:Env :attr:metadata.

render_mode property

render_mode: Optional[WrapperRenderFrame]

render_fps property

render_fps: Optional[int]

backend property

backend: ComputeBackend[Any, WrapperBDeviceT, Any, WrapperBRngT]

device property

device: Optional[WrapperBDeviceT]

batch_size property

batch_size: Optional[int]

action_space property writable

observation_space property writable

observation_space: Space[WrapperObsT, WrapperBDeviceT, Any, WrapperBRngT]

context_space property writable

context_space: Optional[Space[WrapperContextT, WrapperBDeviceT, Any, WrapperBRngT]]

rng property writable

Returns the :attr:Env :attr:rng attribute.

unwrapped property

unwrapped: Env

Returns the base environment of the wrapper.

This will be the bare :class:gymnasium.Env environment, underneath all layers of wrappers.

prev_wrapper_layer property

prev_wrapper_layer: Env[
    BArrayType,
    ContextType,
    ObsType,
    ActType,
    RenderFrame,
    BDeviceType,
    BDtypeType,
    BRNGType,
]

env instance-attribute

env = env

step

step(
    action: ActType,
) -> Tuple[
    ObsType,
    Union[SupportsFloat, BArrayType],
    Union[bool, BArrayType],
    Union[bool, BArrayType],
    Dict[str, Any],
]

reset

reset(
    *args, mask: Optional[BArrayType] = None, seed: Optional[int] = None, **kwargs
) -> Tuple[ContextType, ObsType, Dict[str, Any]]

render

render() -> Optional[Union[RenderFrame, Sequence[RenderFrame]]]

close

close()

sample_space

sample_space(space: Space) -> Any

Sample from space using and updating self.rng.

sample_action

sample_action() -> ActType

Sample one action from action_space.

sample_observation

sample_observation() -> ObsType

Sample one observation from observation_space.

sample_context

sample_context() -> Optional[ContextType]

Sample one context value if context_space is defined.

update_observation_post_reset

update_observation_post_reset(
    old_obs: ObsType, newobs_masked: ObsType, mask: BArrayType
) -> ObsType

Merge masked reset observations back into a full batched observation.

update_context_post_reset

update_context_post_reset(
    old_context: ContextType, new_context: ContextType, mask: BArrayType
) -> ContextType

Merge masked reset contexts back into a full batched context.

has_wrapper_attr

has_wrapper_attr(name: str) -> bool

Checks if the given attribute is within the wrapper or its environment.

get_wrapper_attr

get_wrapper_attr(name: str) -> Any

Gets an attribute from the wrapper and lower environments if name doesn't exist in this object.

Parameters:

Name Type Description Default
name str

The variable name to get

required

Returns:

Type Description
Any

The variable with name in wrapper or lower environments

set_wrapper_attr

set_wrapper_attr(name: str, value: Any)

Sets an attribute on this wrapper or lower environment if name is already defined.

Parameters:

Name Type Description Default
name str

The variable name

required
value Any

The new variable value

required

EpisodeVideoWrapper

EpisodeVideoWrapper(
    env: Env[
        BArrayType,
        ContextType,
        ObsType,
        ActType,
        RenderFrame,
        BDeviceType,
        BDtypeType,
        BRNGType,
    ],
    *,
    store_dir: PathLike | str,
    format: Literal["mp4", "gif", "webm"] = "webm",
    done_index: Optional[int] = None,
    rendered_index: Optional[int] = None,
    nested_separator: str = ".",
    frame_source: Literal["render", "observation"] = "render",
    frame_space: Optional[Space] = None,
    stream: bool = False,
)

Bases: EpisodeRenderStackWrapper[BArrayType, ContextType, ObsType, ActType, RenderFrame, BDeviceType, BDtypeType, BRNGType]

stream=True encodes each frame as it is captured instead of holding the episode's frames in memory until it ends (mp4 / webm only): long episodes of large frames, or many recorded envs, otherwise take frames x height x width x 3 bytes each. The files and their names are the same either way.

Methods:

Name Description
step
reset
render
close
sample_space

Sample from space using and updating self.rng.

sample_action

Sample one action from action_space.

sample_observation

Sample one observation from observation_space.

sample_context

Sample one context value if context_space is defined.

update_observation_post_reset

Merge masked reset observations back into a full batched observation.

update_context_post_reset

Merge masked reset contexts back into a full batched context.

has_wrapper_attr

Checks if the given attribute is within the wrapper or its environment.

get_wrapper_attr

Gets an attribute from the wrapper and lower environments if name doesn't exist in this object.

set_wrapper_attr

Sets an attribute on this wrapper or lower environment if name is already defined.

Attributes:

Name Type Description
store_dir
store_format
stream
metadata Dict[str, Any]

Returns the :attr:Env :attr:metadata.

render_mode Optional[WrapperRenderFrame]
render_fps Optional[int]
backend ComputeBackend[Any, WrapperBDeviceT, Any, WrapperBRngT]
device Optional[WrapperBDeviceT]
batch_size Optional[int]
action_space Space[WrapperActT, WrapperBDeviceT, Any, WrapperBRngT]
observation_space Space[WrapperObsT, WrapperBDeviceT, Any, WrapperBRngT]
context_space Optional[Space[WrapperContextT, WrapperBDeviceT, Any, WrapperBRngT]]
rng WrapperBRngT

Returns the :attr:Env :attr:rng attribute.

unwrapped Env

Returns the base environment of the wrapper.

prev_wrapper_layer Env[BArrayType, ContextType, ObsType, ActType, RenderFrame, BDeviceType, BDtypeType, BRNGType]
env
done_index
rendered_index
nested_separator
frame_source
frame_space
episodic_frames Union[List[Any], Dict[str, List[Any]]]
video_episode_num
video_step_num
video_episode_start_step

store_dir instance-attribute

store_dir = store_dir

store_format instance-attribute

store_format = format

stream instance-attribute

stream = bool(stream)

metadata property writable

metadata: Dict[str, Any]

Returns the :attr:Env :attr:metadata.

render_mode property

render_mode: Optional[WrapperRenderFrame]

render_fps property

render_fps: Optional[int]

backend property

backend: ComputeBackend[Any, WrapperBDeviceT, Any, WrapperBRngT]

device property

device: Optional[WrapperBDeviceT]

batch_size property

batch_size: Optional[int]

action_space property writable

observation_space property writable

observation_space: Space[WrapperObsT, WrapperBDeviceT, Any, WrapperBRngT]

context_space property writable

context_space: Optional[Space[WrapperContextT, WrapperBDeviceT, Any, WrapperBRngT]]

rng property writable

Returns the :attr:Env :attr:rng attribute.

unwrapped property

unwrapped: Env

Returns the base environment of the wrapper.

This will be the bare :class:gymnasium.Env environment, underneath all layers of wrappers.

prev_wrapper_layer property

prev_wrapper_layer: Env[
    BArrayType,
    ContextType,
    ObsType,
    ActType,
    RenderFrame,
    BDeviceType,
    BDtypeType,
    BRNGType,
]

env instance-attribute

env = env

done_index instance-attribute

done_index = done_index

rendered_index instance-attribute

rendered_index = rendered_index

nested_separator instance-attribute

nested_separator = nested_separator

frame_source instance-attribute

frame_source = frame_source

frame_space instance-attribute

frame_space = frame_space

episodic_frames instance-attribute

episodic_frames: Union[List[Any], Dict[str, List[Any]]] = []

video_episode_num instance-attribute

video_episode_num = 0

video_step_num instance-attribute

video_step_num = 0

video_episode_start_step instance-attribute

video_episode_start_step = 0

step

step(
    action: ActType,
) -> Tuple[
    ObsType,
    Union[SupportsFloat, BArrayType],
    Union[bool, BArrayType],
    Union[bool, BArrayType],
    Dict[str, Any],
]

reset

reset(
    *args, mask: Optional[BArrayType] = None, seed: Optional[int] = None, **kwargs
) -> Tuple[ContextType, ObsType, Dict[str, Any]]

render

render() -> Optional[Union[RenderFrame, Sequence[RenderFrame]]]

close

close()

sample_space

sample_space(space: Space) -> Any

Sample from space using and updating self.rng.

sample_action

sample_action() -> ActType

Sample one action from action_space.

sample_observation

sample_observation() -> ObsType

Sample one observation from observation_space.

sample_context

sample_context() -> Optional[ContextType]

Sample one context value if context_space is defined.

update_observation_post_reset

update_observation_post_reset(
    old_obs: ObsType, newobs_masked: ObsType, mask: BArrayType
) -> ObsType

Merge masked reset observations back into a full batched observation.

update_context_post_reset

update_context_post_reset(
    old_context: ContextType, new_context: ContextType, mask: BArrayType
) -> ContextType

Merge masked reset contexts back into a full batched context.

has_wrapper_attr

has_wrapper_attr(name: str) -> bool

Checks if the given attribute is within the wrapper or its environment.

get_wrapper_attr

get_wrapper_attr(name: str) -> Any

Gets an attribute from the wrapper and lower environments if name doesn't exist in this object.

Parameters:

Name Type Description Default
name str

The variable name to get

required

Returns:

Type Description
Any

The variable with name in wrapper or lower environments

set_wrapper_attr

set_wrapper_attr(name: str, value: Any)

Sets an attribute on this wrapper or lower environment if name is already defined.

Parameters:

Name Type Description Default
name str

The variable name

required
value Any

The new variable value

required

EpisodeWandbVideoWrapper

EpisodeWandbVideoWrapper(
    env: Env[
        BArrayType,
        ContextType,
        ObsType,
        ActType,
        RenderFrame,
        BDeviceType,
        BDtypeType,
        BRNGType,
    ],
    *,
    wandb_log_key: str,
    format: Literal["mp4", "gif", "webm"] = "webm",
    control_wandb_step: bool = False,
    log_wandb_episode_id: bool = True,
    done_index: Optional[int] = None,
    rendered_index: Optional[int] = None,
    nested_separator: str = ".",
    frame_source: Literal["render", "observation"] = "render",
    frame_space: Optional[Space] = None,
)

Bases: EpisodeRenderStackWrapper[BArrayType, ContextType, ObsType, ActType, RenderFrame, BDeviceType, BDtypeType, BRNGType]

Methods:

Name Description
step
reset
render
close
sample_space

Sample from space using and updating self.rng.

sample_action

Sample one action from action_space.

sample_observation

Sample one observation from observation_space.

sample_context

Sample one context value if context_space is defined.

update_observation_post_reset

Merge masked reset observations back into a full batched observation.

update_context_post_reset

Merge masked reset contexts back into a full batched context.

has_wrapper_attr

Checks if the given attribute is within the wrapper or its environment.

get_wrapper_attr

Gets an attribute from the wrapper and lower environments if name doesn't exist in this object.

set_wrapper_attr

Sets an attribute on this wrapper or lower environment if name is already defined.

Attributes:

Name Type Description
wandb
wandb_log_key
log_wandb_episode_id
control_wandb_step
store_format
metadata Dict[str, Any]

Returns the :attr:Env :attr:metadata.

render_mode Optional[WrapperRenderFrame]
render_fps Optional[int]
backend ComputeBackend[Any, WrapperBDeviceT, Any, WrapperBRngT]
device Optional[WrapperBDeviceT]
batch_size Optional[int]
action_space Space[WrapperActT, WrapperBDeviceT, Any, WrapperBRngT]
observation_space Space[WrapperObsT, WrapperBDeviceT, Any, WrapperBRngT]
context_space Optional[Space[WrapperContextT, WrapperBDeviceT, Any, WrapperBRngT]]
rng WrapperBRngT

Returns the :attr:Env :attr:rng attribute.

unwrapped Env

Returns the base environment of the wrapper.

prev_wrapper_layer Env[BArrayType, ContextType, ObsType, ActType, RenderFrame, BDeviceType, BDtypeType, BRNGType]
env
done_index
rendered_index
nested_separator
frame_source
frame_space
episodic_frames Union[List[Any], Dict[str, List[Any]]]
video_episode_num
video_step_num
video_episode_start_step

wandb instance-attribute

wandb = wandb

wandb_log_key instance-attribute

wandb_log_key = wandb_log_key

log_wandb_episode_id instance-attribute

log_wandb_episode_id = log_wandb_episode_id

control_wandb_step instance-attribute

control_wandb_step = control_wandb_step

store_format instance-attribute

store_format = format

metadata property writable

metadata: Dict[str, Any]

Returns the :attr:Env :attr:metadata.

render_mode property

render_mode: Optional[WrapperRenderFrame]

render_fps property

render_fps: Optional[int]

backend property

backend: ComputeBackend[Any, WrapperBDeviceT, Any, WrapperBRngT]

device property

device: Optional[WrapperBDeviceT]

batch_size property

batch_size: Optional[int]

action_space property writable

observation_space property writable

observation_space: Space[WrapperObsT, WrapperBDeviceT, Any, WrapperBRngT]

context_space property writable

context_space: Optional[Space[WrapperContextT, WrapperBDeviceT, Any, WrapperBRngT]]

rng property writable

Returns the :attr:Env :attr:rng attribute.

unwrapped property

unwrapped: Env

Returns the base environment of the wrapper.

This will be the bare :class:gymnasium.Env environment, underneath all layers of wrappers.

prev_wrapper_layer property

prev_wrapper_layer: Env[
    BArrayType,
    ContextType,
    ObsType,
    ActType,
    RenderFrame,
    BDeviceType,
    BDtypeType,
    BRNGType,
]

env instance-attribute

env = env

done_index instance-attribute

done_index = done_index

rendered_index instance-attribute

rendered_index = rendered_index

nested_separator instance-attribute

nested_separator = nested_separator

frame_source instance-attribute

frame_source = frame_source

frame_space instance-attribute

frame_space = frame_space

episodic_frames instance-attribute

episodic_frames: Union[List[Any], Dict[str, List[Any]]] = []

video_episode_num instance-attribute

video_episode_num = 0

video_step_num instance-attribute

video_step_num = 0

video_episode_start_step instance-attribute

video_episode_start_step = 0

step

step(
    action: ActType,
) -> Tuple[
    ObsType,
    Union[SupportsFloat, BArrayType],
    Union[bool, BArrayType],
    Union[bool, BArrayType],
    Dict[str, Any],
]

reset

reset(
    *args, mask: Optional[BArrayType] = None, seed: Optional[int] = None, **kwargs
) -> Tuple[ContextType, ObsType, Dict[str, Any]]

render

render() -> Optional[Union[RenderFrame, Sequence[RenderFrame]]]

close

close()

sample_space

sample_space(space: Space) -> Any

Sample from space using and updating self.rng.

sample_action

sample_action() -> ActType

Sample one action from action_space.

sample_observation

sample_observation() -> ObsType

Sample one observation from observation_space.

sample_context

sample_context() -> Optional[ContextType]

Sample one context value if context_space is defined.

update_observation_post_reset

update_observation_post_reset(
    old_obs: ObsType, newobs_masked: ObsType, mask: BArrayType
) -> ObsType

Merge masked reset observations back into a full batched observation.

update_context_post_reset

update_context_post_reset(
    old_context: ContextType, new_context: ContextType, mask: BArrayType
) -> ContextType

Merge masked reset contexts back into a full batched context.

has_wrapper_attr

has_wrapper_attr(name: str) -> bool

Checks if the given attribute is within the wrapper or its environment.

get_wrapper_attr

get_wrapper_attr(name: str) -> Any

Gets an attribute from the wrapper and lower environments if name doesn't exist in this object.

Parameters:

Name Type Description Default
name str

The variable name to get

required

Returns:

Type Description
Any

The variable with name in wrapper or lower environments

set_wrapper_attr

set_wrapper_attr(name: str, value: Any)

Sets an attribute on this wrapper or lower environment if name is already defined.

Parameters:

Name Type Description Default
name str

The variable name

required
value Any

The new variable value

required

image_specs_from_space

image_specs_from_space(
    space: Optional[Space], nested_separator: str = "."
) -> Optional[ImageSpecs]

The image leaves of space declared through metadata["kind"] == "image".

Returns {flattened key: (layout, format)} (key None when space itself is the image), or None when no leaf of space declares a kind at all, i.e. the space carries no usable description and callers should fall back to treating every frame as an RGB image.

image_to_hwc_rgb

image_to_hwc_rgb(
    image: ndarray, layout: Optional[str] = None, format: Optional[str] = None
) -> ndarray

Convert an image described by layout / format metadata to (H, W, 3) RGB.

layout "CHW" is moved channel-last; format "bgr" / "bgra" are reordered, "rgba" drops alpha and "gray" is repeated over three channels. None (or "HWC" / "rgb") leaves the image unchanged.