The GPU frame capture is warning:
your application called CAMetalLayer nextDrawable earlier than needed
I am calling present(drawable) after all my other GPU calls have been made, directly before committing the command buffer
guard let commandBuffer = commandQueue.makeCommandBuffer(),
let computeBuffer = commandQueue.makeCommandBuffer(),
let descriptor = view.currentRenderPassDescriptor else { return }
...
//First run the compute kernel
guard let computeEncoder = computeBuffer.makeComputeCommandEncoder() else { return }
computeEncoder.setComputePipelineState(computePipelineState)
dispatchThreads(particleCount: particleCount)
computeEncoder.endEncoding()
computeBuffer.commit()
//I need to wait here because I need values from a computed buffer first
//and also why I am not just using a single pipeline descriptor
computeBuffer.waitUntilCompleted()
//Next render computed particles with vertex shader to a texture
let renderEncoder = commandBuffer.makeRenderCommandEncoder(descriptor: descriptor0)!
renderEncoder.setRenderPipelineState(renderPipelineState)
...
renderEncoder.endEncoding()
//Draw texture (created above) using vertex shader:
let renderTexture = commandBuffer.makeRenderCommandEncoder(descriptor: descriptor)
renderTexture?.setRenderPipelineState(renderCanvasPipelineState)
...
renderTexture?.endEncoding()
//Finally present drawable and commit command buffer:
guard let drawable = view.currentDrawable else { return }
commandBuffer.present(drawable)
commandBuffer.commit()
I don't see how it is possible to request the currentDrawable any later. Am I doing something wrong or sub-optimal?
I started looking into the issue because most frames (basically doing the same thing), the wait for the current drawable is about 3-10ms. Occasionally the wait is 35-45ms.
I see a number of recommendations to use presentDrawable instead of present, but that does not seem to be an option with Swift.
Is there any way to request the current drawable when it is needed and make the warning go away?
You're calling view.currentRenderPassDescriptor at the top of the code you listed. The render pass descriptor has to have a reference to the drawable's texture as a color attachment. That implies obtaining the drawable and asking for its texture. So, that line requests the drawable.
Don't obtain the render pass descriptor until just before you create the render command encoder (let renderTexture = commandBuffer.makeRenderCommandEncoder(descriptor: descriptor)) using it.
Related
I’m using Metal with Swift to build a 3D viewer for iOS and I have some issues to make the depth working. From now, I can draw and render a single shape correctly in 3D (like a simple square plane (4 triangles (2 for each face)) or a tetrahedron (4 triangles)).
However, when I try to draw 2 shapes together, the depth between these two shapes doesn’t work. For example, a plane is placed at Z axes = 0 behind a tetra which is placed at Z > 0. If I look a this scene from the back (camera placed somewhere at Z < 0), it’s ok. But when I look at this scene from the front (camera placed somewhere at Z > 0), it doesn’t work. The plane is drawn before the tetra even if it is placed behind the tetra.
I think that the plane is always drawn on the screen before the tetra (no matter the position of the camera) because the call of drawPrimitives for the plane is done before the call for the tetra. However, I was thinking that all the depth and stencil settings will deal with that properly.
I don’t know if the depth isn’t working because depth texture, stencil state and so on are not correctly set or because each shape is drawn in a different call of drawPrimitives.
In other words, do I have to draw all shapes in the same call of drawPrimitives to make the depth working ? The idea of this multiple call to drawPrimitives is to deal with different kinds of primitive type for each shape (triangle or line or …).
This is how I set the depth stencil state and the depth texture and the render pipeline :
init() {
// some miscellaneous initialisation …
// …
// all MTL stuff :
commandQueue = device.makeCommandQueue()
// Stencil descriptor
let depthStencilDescriptor = MTLDepthStencilDescriptor()
depthStencilDescriptor.depthCompareFunction = .less
depthStencilDescriptor.isDepthWriteEnabled = true
depthStencilState = device.makeDepthStencilState(descriptor: depthStencilDescriptor)!
// Library and pipeline descriptor & state
let library = try! device.makeLibrary(source: shaders, options: nil)
// Our vertex function name
let vertexFunction = library.makeFunction(name: "basic_vertex_function")
// Our fragment function name
let fragmentFunction = library.makeFunction(name: "basic_fragment_function")
// Create basic descriptor
let renderPipelineDescriptor = MTLRenderPipelineDescriptor()
// Attach the pixel format that si the same as the MetalView
renderPipelineDescriptor.colorAttachments[0].pixelFormat = .bgra8Unorm
renderPipelineDescriptor.depthAttachmentPixelFormat = .depth32Float_stencil8
renderPipelineDescriptor.stencilAttachmentPixelFormat = .depth32Float_stencil8
//renderPipelineDescriptor.stencilAttachmentPixelFormat = .stencil8
// Attach the shader functions
renderPipelineDescriptor.vertexFunction = vertexFunction
renderPipelineDescriptor.fragmentFunction = fragmentFunction
// Try to update the state of the renderPipeline
do {
renderPipelineState = try device.makeRenderPipelineState(descriptor: renderPipelineDescriptor)
} catch {
print(error.localizedDescription)
}
// Depth Texture
let desc = MTLTextureDescriptor.texture2DDescriptor(pixelFormat: .stencil8, width: 576, height: 723, mipmapped: false)
desc.storageMode = .private
desc.usage = .pixelFormatView
depthTexture = device.makeTexture(descriptor: desc)!
// Uniforms buffer
modelMatrix = Matrix4()
modelMatrix.multiplyLeft(worldMatrix)
uniformBuffer = device.makeBuffer( length: MemoryLayout<Float>.stride*16*2, options: [])
let bufferPointer = uniformBuffer.contents()
memcpy(bufferPointer, &modelMatrix.matrix.m, MemoryLayout<Float>.stride * 16)
memcpy(bufferPointer + MemoryLayout<Float>.stride * 16, &projectionMatrix.matrix.m, MemoryLayout<Float>.stride * 16)
}
And the draw function :
function draw(in view: MTKView) {
// create render pass descriptor
guard let drawable = view.currentDrawable,
let renderPassDescriptor = view.currentRenderPassDescriptor else {
return
}
renderPassDescriptor.depthAttachment.texture = depthTexture
renderPassDescriptor.depthAttachment.clearDepth = 1.0
//renderPassDescriptor.depthAttachment.loadAction = .load
renderPassDescriptor.depthAttachment.loadAction = .clear
renderPassDescriptor.depthAttachment.storeAction = .store
// Create a buffer from the commandQueue
let commandBuffer = commandQueue.makeCommandBuffer()
let commandEncoder = commandBuffer?.makeRenderCommandEncoder(descriptor: renderPassDescriptor)
commandEncoder?.setRenderPipelineState(renderPipelineState)
commandEncoder?.setFrontFacing(.counterClockwise)
commandEncoder?.setCullMode(.back)
commandEncoder?.setDepthStencilState(depthStencilState)
// Draw all obj in objects
// objects = array of Object; each object describing vertices and primitive type of a shape
// objects[0] = Plane, objects[1] = Tetra
for obj in objects {
createVertexBuffers(device: view.device!, vertices: obj.vertices)
commandEncoder?.setVertexBuffer(vertexBuffer, offset: 0, index: 0)
commandEncoder?.setVertexBuffer(uniformBuffer, offset: 0, index: 1)
commandEncoder?.drawPrimitives(type: obj.primitive, vertexStart: 0, vertexCount: obj.vertices.count)
}
commandEncoder?.endEncoding()
commandBuffer?.present(drawable)
commandBuffer?.commit()
}
Does anyone has an idea of what is wrong or missing ?
Any advice is welcome !
Edited 09/23/2022: Code updated
Few things of the top of my head:
First
let desc = MTLTextureDescriptor.texture2DDescriptor(pixelFormat: .depth32Float_stencil8, width: 576, height: 723, mipmapped: false)
Second
renderPipelineDescriptor.depthAttachmentPixelFormat = .depth32Float_stencil8
Notice the pixeFormat should be same in both places, and since you seem to be using stencil test as well so depth32Float_stencil8 will be perfect.
Third
Now another thing you seem to be missing is, clearing depth texture before every render pass, am I right?
So, you should set load action of depth attachment to .clear, like this:
renderPassDescriptor.depthAttachment.loadAction = .clear
Fourth (Subjective to your usecase)*
If none of the above works, you might need to discard framents with alpha = 0 in your fragment function by calling discard_fragment() when color you are returning has alpha 0
Also note for future:
Ideally you want depth texture to be fresh and empty when every new frame starts getting rendered (first draw call of a render pass) and then reuse it for subsequent draw calls in same render pass by setting load action .load and store action .store.
ex: Assuming you have 3 draw calls, say drawing polygons wiz triangle, rectangle, sphere in one frame, then your depth attachment setup should be like this:
Frame 1 Starts:
First Draw: triangle
loadAction: Clear
storeAction: Store
Second Draw: rectangle
loadAction: load
storeAction: Store
Third Draw: sphere
loadAction: load
storeAction: store/dontcare
Frame 2 Starts: Notice you clear depth buffer for 1st draw call of new frame
First Draw: triangle
loadAction: Clear
storeAction: Store
Second Draw: rectangle
loadAction: load
storeAction: Store
Third Draw: sphere
loadAction: load
storeAction: store/dontcare
Your depth texture pixel format is not correct, try to change its pixel format to: MTLPixelFormatDepth32Float or MTLPixelFormatDepth32Float_Stencil8.
How can i export collage video using different resolution videos? I'm trying to achieve like showing first image below, I'm using AVCustomEdit demo and have done so far, I created AVMutableVideoComposition pass all video trackIDs to customVideoCompositorClass and getting all videos CVPixelBuffer and than converting in MTLTexture than render all textures but problem is my video output size is square(destinationTexture) and videos size is portrait or landscape thats why every video is squeezed also how can i rotate scale position and mask shape every video? also how can i apply cifilters? should i convert every CVPixelBuffer to ciImage and ciImage back to CVPixelBuffer?
override func renderPixelBuffer(backgroundTexture: MTLTexture,
firstPixelBuffer: CVPixelBuffer,
secondPixelBuffer: CVPixelBuffer,
thirdPixelBuffer: CVPixelBuffer,
fourthPixelBuffer: CVPixelBuffer,
destinationPixelBuffer: CVPixelBuffer) {
// Create a MTLTexture from the CVPixelBuffer.
guard let firstTexture = buildTextureForPixelBuffer(firstPixelBuffer) else { return }
guard let secondTexture = buildTextureForPixelBuffer(secondPixelBuffer) else { return }
guard let thirdTexture = buildTextureForPixelBuffer(thirdPixelBuffer) else { return }
guard let fourthTexture = buildTextureForPixelBuffer(fourthPixelBuffer) else { return }
guard let destinationTexture = buildTextureForPixelBuffer(destinationPixelBuffer) else { return }
/*
We must maintain a reference to the pixel buffer until the Metal rendering is complete. This is because the
'buildTextureForPixelBuffer' function above uses CVMetalTextureCacheCreateTextureFromImage to create a
Metal texture (CVMetalTexture) from the IOSurface that backs the CVPixelBuffer, but
CVMetalTextureCacheCreateTextureFromImage doesn't increment the use count of the IOSurface; only the
CVPixelBuffer, and the CVMTLTexture own this IOSurface. Therefore we must maintain a reference to either
the pixel buffer or Metal texture until the Metal rendering is done. The MTLCommandBuffer completion
handler below is then used to release these references.
*/
pixelBuffers = RenderPixelBuffers(firstBuffer: firstPixelBuffer,
secondBuffer: secondPixelBuffer,
thirdBuffer: thirdPixelBuffer,
fourthBuffer: fourthPixelBuffer,
destinationBuffer: destinationPixelBuffer)
// Create a new command buffer for each renderpass to the current drawable.
let commandBuffer = commandQueue.makeCommandBuffer()!
commandBuffer.label = "MyCommand"
/*
Obtain a drawable texture for this render pass and set up the renderpass
descriptor for the command encoder to render into.
*/
let renderPassDescriptor = setupRenderPassDescriptorForTexture(destinationTexture)
// Create a render command encoder so we can render into something.
let renderEncoder = commandBuffer.makeRenderCommandEncoder(descriptor: renderPassDescriptor)!
renderEncoder.label = "MyRenderEncoder"
guard let renderPipelineState = renderPipelineState else { return }
modelConstants.modelViewMatrix = matrix_identity_float4x4
// Render background texture.
renderTexture(renderEncoder, texture: backgroundTexture, pipelineState: renderPipelineState)
var translationMatrix = matrix_float4x4(translation: simd_float3(-0.5, 0.5, 0))
// var rotationMatrix = matrix_float4x4(rotationZ: radians(fromDegrees: -90))
var scaleMatrix = matrix_float4x4(scaling: 0.25)
var modelMatrix = translationMatrix * scaleMatrix
modelConstants.modelViewMatrix = modelMatrix
// Render first texture.
renderTexture(renderEncoder, texture: firstTexture, pipelineState: renderPipelineState)
// translationMatrix = matrix_float4x4(translation: simd_float3(0.5, -0.5, 0))
// rotationMatrix = matrix_float4x4(rotationZ: radians(fromDegrees: -45))
// scaleMatrix = matrix_float4x4(scaling: 0.5)
// modelMatrix = translationMatrix * scaleMatrix * rotationMatrix
// modelConstants.modelViewMatrix = modelMatrix
// // Render second texture.
// renderTexture(renderEncoder, texture: secondTexture, pipelineState: renderPipelineState)
//
// // Render third texture.
// renderTexture(renderEncoder, texture: thirdTexture, pipelineState: renderPipelineState)
//
// // Render fourth texture.
// renderTexture(renderEncoder, texture: fourthTexture, pipelineState: renderPipelineState)
// We're done encoding commands.
renderEncoder.endEncoding()
// Use the command buffer completion block to release the reference to the pixel buffers.
commandBuffer.addCompletedHandler({ _ in
self.pixelBuffers = nil // Release the reference to the pixel buffers.
})
// Finalize rendering here & push the command buffer to the GPU.
commandBuffer.commit()
}
I would recommend to use a library called MetalPetal. It is an image processing framework based on Metal.You have to convert the CVPixelBuffer in to MetalImage that is MTIImage. and then you can do anything in the image like there are premade filters and you can apply to it or you can use even CIFilter or your custom filters and you can transform , rotate , crop every frame so that collage frames are accurate . then you have to convert the MTIimage to cvpixelbuffer again . Here you can also CIImage but it will be slow i guess. And you are getting box images maybe for the render size . Please see the render size .
I'm trying to understand GPU image processing with Metal and have successfully manipulated and rendered images onto a screen.
I currently render my manipulated image by using a custom vertex and fragment shader that takes in the size of the texture and the screen, and then resizes it to the correct aspect ratio. It is redrawn using MTKViewDelegate. This code is shown below
guard let currentRenderPassDescriptor = view.currentRenderPassDescriptor,
let currentDrawable = view.currentDrawable,
let renderPipelineState = renderPipelineState,
let encoder = commandBuffer.makeRenderCommandEncoder(descriptor: currentRenderPassDescriptor)
else {
semaphore.signal()
return
}
encoder.pushDebugGroup("RenderFrame")
encoder.setRenderPipelineState(renderPipelineState)
encoder.setVertexBytes(&textureOccupiedDimensions, length: MemoryLayout<vector_int2>.size, index: 0)
encoder.setVertexBytes(&viewportSize, length: MemoryLayout<vector_int2>.size, index: 1)
encoder.setFragmentTexture(texture, index: 0)
encoder.drawPrimitives(type: .triangleStrip, vertexStart: 0, vertexCount: 4, instanceCount: 1)
encoder.popDebugGroup()
encoder.endEncoding()
commandBuffer.addScheduledHandler { [weak self] (buffer) in
guard let unwrappedSelf = self else { return }
unwrappedSelf.didRenderTexture(texture, withCommandBuffer: buffer, device: device)
unwrappedSelf.semaphore.signal()
}
commandBuffer.present(currentDrawable) // schedules display at next possible refresh point
commandBuffer.commit()
I now want to draw some primitive shapes (lines, circles, etc) on my rendered image that don't require knowledge of the texture layout of the image underneath, but am not sure what's the best way to approach this challenge. These shapes will have to update each time the image is refreshed as well. I understand how to use compute encoders to copy the whole texture while altering certain points, but I don't know if this is the most efficient way to accomplish this. Any advice would be much appreciated!
I am new to Metal. How many MTLRenderpassdescriptor we can create in a single Application any limit ?
I want to have three MTLRenderpassdescriptor is it possible. One I need to draw quads according to finger position. It is like Offscreen Rendering. There will be texture output.
I want to pass the output texture to another renderpass descriptor to process the color of it which is also similar to Offscreen Rendering.Here new output texture will be available
I want to present new output Texture to the screen. Is it Possible ? When I create three renderpass descriptors only two is available.
First offscreen rendering It works fine
let renderPass = MTLRenderPassDescriptor()
renderPass.colorAttachments[0].texture = ssTexture
renderPass.colorAttachments[0].loadAction = .clear
renderPass.colorAttachments[0].clearColor = MTLClearColorMake( 0.0, 0.0, 0.0, 0.0)
renderPass.colorAttachments[0].storeAction = .store
let commandBuffer = commandQueue.makeCommandBuffer()
var commandEncoder = commandBuffer?.makeRenderCommandEncoder(descriptor: renderPass)
for scene in scenesTool{
scene.render(commandEncoder: commandEncoder!)
}
commandEncoder?.endEncoding()
This renderpass descriptor is not available in Debug as well.
let renderPassWC = MTLRenderPassDescriptor()
renderPassWC.colorAttachments[0].texture = ssCurrentTexture
renderPassWC.colorAttachments[0].loadAction = .clear
renderPassWC.colorAttachments[0].clearColor = MTLClearColorMake( 0.0, 0.0, 0.0, 0.0)
renderPassWC.colorAttachments[0].storeAction = .store
let commandBufferWC = commandQueue.makeCommandBuffer()
let commandEncoderWC = commandBufferWC?.makeRenderCommandEncoder(descriptor: renderPassWC)
for scene in scenesWCTool{
print(" msbksadbdksabfkasbd")
scene.render(commandEncoder: commandEncoderWC!)
}
commandEncoderWC?.endEncoding()
Third Renderpass Descriptor is working Fine.
let descriptor = view.currentRenderPassDescriptor
commandEncoder = commandBuffer?.makeRenderCommandEncoder(descriptor: descriptor!)
for x in canvasTemporaryScenes{
x.updateCanvas(texture: ssTexture!)
x.render(commandEncoder: commandEncoder!)
}
commandEncoder?.endEncoding()
guard let drawable = view.currentDrawable else { return }
commandBuffer?.present(drawable)
commandBuffer?.commit()
commandBuffer?.waitUntilCompleted()
MTLRenderPassDescriptor objects are cheap to create, unlike some other kinds of objects in Metal (e.g. pipeline state objects). You are expected to create descriptors any time you need them. You don't need to keep them for a long period of time. Just recreate them each time you are going to create a render command encoder. For example, creating multiple render pass descriptors per frame is perfectly normal. There's no specific limit on the number you can create.
I have no idea what you mean by "When I create three renderpass descriptors only two is available" and "This renderpass descriptor is not available in Debug". What do you mean by "available" in these sentences? Explain exactly what results you want and expect and what's actually happening that's different than that.
Im trying to apply the live camera filters through metal using the default MPSKernal filters given by apple and custom compute Shaders.
Im applying the custom filters on a collection view in a grid with the combination of default and the custom kernel functions.
It looks like in the app Clips.
But what I observed is that using the custom filters there is a lot of memory leaks compared to the default kernel functions given by apple.
I don't know what mistakes I made though if any.
Here is my custom compute shader.
kernel void customFunction1(
texture2d<float, access::read> inTexture [[texture(0)]],
texture2d<float, access::write> outTexture [[texture(1)]],
uint2 gid [[thread_position_in_grid]]){
const float4 colorAtPixel = inTexture.read(gid);
const float4 outputColor = float4(colorAtPixel.r, colorAtPixel.g, colorAtPixel.b, 1);
outTexture.write(outputColor, gid);
}
Regarding to my creating my pipeline and the dispatching through thread groups the code goes here
let blur = MPSImageGaussianBlur(device: device, sigma: 0)
let threadsPerThreadgroup = MTLSizeMake(4, 4, 1)
let threadgroupsPerGrid = MTLSizeMake(destinationTexture.width / threadsPerThreadgroup.width, destinationTexture.height / threadsPerThreadgroup.height, 1)
let commandEncoder = commandBuffer.makeComputeCommandEncoder()
commandEncoder.setComputePipelineState(pipelineState!)
commandEncoder.setTexture(sourceTexture, at: 0)
commandEncoder.setTexture(destinationTexture, at: 1)
commandEncoder.dispatchThreadgroups(threadgroupsPerGrid, threadsPerThreadgroup: threadsPerThreadgroup)
commandEncoder.endEncoding()
autoreleasepool {
let inPlaceTexture = UnsafeMutablePointer<MTLTexture>.allocate(capacity: 1)
inPlaceTexture.initialize(to: destinationTexture)
blur.encode(commandBuffer: commandBuffer, inPlaceTexture: inPlaceTexture, fallbackCopyAllocator: nil)
}
The Pipeline state with the custom Shader is created like this.
let defaultLibrary = device.newDefaultLibrary()
let kernelFunction = defaultLibrary!.makeFunction(name: name)
do {
pipelineState = try device.makeComputePipelineState(function: kernelFunction!)
} catch {
fatalError("Unable to create pipeline state")
}
And in instrumentation it shows there is a leak in some Malloc 16 bytes and in [Mtkview draw] method.
The screenshot is shown below.
I want help in finding where and how the issue is occurring from.
Thanks.
There's no reason to explicitly allocate an UnsafeMutablePointer to store the in-place texture parameter. Incidentally, that's the source of your leak: you allocate the pointer and then never deallocate it.
Use a local variable to pass the texture instead:
var inPlaceTexture = destinationTexture
blur.encode(commandBuffer: commandBuffer, inPlaceTexture: &inPlaceTexture, fallbackCopyAllocator: nil)
By the way, you're (eventually) going to have a bad time if you call the in-place encode method without providing a fallback allocator or checking the return value. In-place encoding will fail in certain situations, so you should provide a closure that allocates an appropriate texture in the event of failure.