Detailed Description

Decodes P210 and P216 semi-planar 4:2:2 videos.

One layout, two bit depths:

Plane 0: Y (16-bit words), full resolution
Plane 1: UV interleaved (16-bit per component), half width, full height

P210 puts 10 bits in the HIGH bits of each word, so R16 unorm returns a value that is short of 1.0 at full scale and has to be renormalized – that is what the scale is for. P216 uses all 16 bits and wants no scaling at all, which is the only difference between the two and why they share this decoder.

P216 is what NDI delivers for a 16-bit source, and NDI delivers it as individual FIELDS: half-height frames at twice the rate. Those are uploaded into the two halves of one full-height texture (see planeRows) and resolved by score_tc, so nothing here weaves anything on the CPU.

Public Member Functions
	P210Decoder (Video::ImageFormat &d, const char *sampleScale=SCORE_GFX_MSB_ALIGNED_SCALE)

std::pair< QShader, QShader >	init (RenderList &r) override
	Initialize a GPUVideoDecoder.

void	exec (RenderList &, QRhiResourceUpdateBatch &res, AVFrame &frame) override
	Decode and upload a video frame to the GPU.

Public Member Functions inherited from score::gfx::GPUVideoDecoder
virtual void	release (RenderList &)
	This method will release all the created samplers and textures.

Public Attributes
Video::ImageFormat &	decoder

const char *	scale {SCORE_GFX_MSB_ALIGNED_SCALE}

Public Attributes inherited from score::gfx::GPUVideoDecoder
std::vector< Sampler >	samplers

bool	hasFrame {}

bool	failed {}

bool	formatChanged {}

Static Public Attributes
static const constexpr auto	frag

Additional Inherited Members
Static Public Member Functions inherited from score::gfx::GPUVideoDecoder
static QRhiTextureSubresourceUploadDescription	createTextureUpload (uint8_t *pixels, int w, int h, int bytesPerPixel, int stride)
	Utility method to create a QRhiTextureSubresourceUploadDescription.

static bool	isTopField (const AVFrame &frame) noexcept
	Whether this frame carries the EVEN lines of its picture.

static PlaneRows	planeRows (const Video::ImageFormat &fmt, const AVFrame &frame, int textureRows) noexcept

static QString	vertexShader (bool invertY=false) noexcept

Member Function Documentation

◆ exec()

void score::gfx::P210Decoder::exec	(	RenderList &	,
		QRhiResourceUpdateBatch &	res,
		AVFrame &	frame
	)

inlineoverridevirtual

Decode and upload a video frame to the GPU.

Implements score::gfx::GPUVideoDecoder.

◆ init()

std::pair< QShader, QShader > score::gfx::P210Decoder::init ( RenderList & r )

inlineoverridevirtual

Initialize a GPUVideoDecoder.

This method must :

Create samplers and textures for the video format.
Create shaders that will render the data put into these textures.

It returns a {vertex, fragment} shader pair.

Implements score::gfx::GPUVideoDecoder.

Member Data Documentation

◆ frag

const constexpr auto score::gfx::P210Decoder::frag

staticconstexpr

Initial value:

= R"_(#version 450
 
)_" SCORE_GFX_VIDEO_UNIFORMS R"_(
 
layout(binding=3) uniform sampler2D y_tex;
layout(binding=4) uniform sampler2D uv_tex;
 
layout(location = 0) in vec2 v_texcoord;
layout(location = 0) out vec4 fragColor;
 
%2
 
vec4 processTexture(vec4 tex) {
  vec4 processed = convert_to_rgb(tex);
  { %1 }
  return processed;
}
 
void main()
{
  const float s = %3;
  float y = s * texture(y_tex, score_tc(v_texcoord)).r;
  float u = s * texture(uv_tex, score_tc(v_texcoord)).r;
  float v = s * texture(uv_tex, score_tc(v_texcoord)).g;
 
  fragColor = processTexture(vec4(y, u, v, 1.));
})_"

◆ scale

const char* score::gfx::P210Decoder::scale {SCORE_GFX_MSB_ALIGNED_SCALE}

"1.0" for P216, which uses every bit; the MSB-aligned renormalization for P210, whose ten bits sit at the top of a 16-bit word.

The documentation for this struct was generated from the following file:

P210.hpp

Detailed Description

Public Member Functions

Public Attributes