Struct Yuv420pReadback 

pub struct Yuv420pReadback { /* private fields */ }

A submitted GPU RGB-to-YUV420 conversion whose buffer is not mapped yet.

Implementations

impl Yuv420pReadback

pub const fn size(&self) -> RenderSize

Returns the luma-plane dimensions.

pub const fn staging_bytes(&self) -> u64

Returns bytes retained by the GPU-visible staging buffer.

pub fn wait_ready(&self) -> Result<(), RenderError>

Blocks a native caller until this readback is mapped.

Errors

Returns RenderError::TextureReadback when mapping or polling fails.

pub fn wait_for_submission(&self) -> Result<(), RenderError>

Blocks native GPU progress through this readback’s submission.

This can be paired with an earlier Self::map_async call when a scheduler wants mapping to begin before it eventually has to wait.

Errors

Returns RenderError::GpuTimeout when the submission does not complete within the renderer’s wait timeout, or RenderError::TextureReadback when device polling fails.

pub fn wait(self) -> Result<Yuv420pPlanes, RenderError>

Waits for this readback and returns tightly packed planes.

Errors

Returns RenderError::TextureReadback when mapping, polling, or row packing fails.

pub fn map_async( &self, on_ready: impl FnOnce(Result<(), RenderError>) + Send + 'static, )

Requests buffer mapping and reports readiness exactly once.

pub fn poll(&self)

Advances backends that require explicit polling.

pub fn finish_into( self, targets: &mut Yuv420pTargets<'_>, ) -> Result<(), RenderError>

Packs mapped Y and UV rows into plane buffers the caller already owns.

Yuv420pReadback::finish allocates three vectors per frame, which is what the browser path wants. A native caller that already owns the memory the encoder will read from can pack straight into it, which removes both that allocation and the copy that would follow it. The pixels written are the same either way.

Each target may be strided, because encoder plane memory usually is.

Errors

Returns RenderError::TextureReadback when the buffer is not mapped, its layout cannot be represented, or a target is too small, and RenderError::GpuWorkDropped when the GPU did not run the copy: the buffer holds no frame of this readback, so it is not recycled, and the renderer refuses further work.

pub fn finish(self) -> Result<Yuv420pPlanes, RenderError>

Packs mapped Y and UV rows into three tightly packed planes.

Errors

Returns RenderError::TextureReadback when the buffer is not mapped or its layout cannot be represented, and RenderError::GpuWorkDropped when the GPU did not run the copy, as Self::finish_into does.

Trait Implementations

impl Debug for Yuv420pReadback

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.