imgui_impl_wgpu.cpp

// dear imgui: Renderer for WebGPU
// This needs to be used along with a Platform Binding (e.g. GLFW)
// (Please note that WebGPU is currently experimental, will not run on non-beta browsers, and may break.)

// Implemented features:
//  [X] Renderer: User texture binding. Use 'WGPUTextureView' as ImTextureID. Read the FAQ about ImTextureID!
//  [X] Renderer: Large meshes support (64k+ vertices) with 16-bit indices.
//  [X] Renderer: Expose selected render state for draw callbacks to use. Access in '(ImGui_ImplXXXX_RenderState*)GetPlatformIO().Renderer_RenderState'.

// You can use unmodified imgui_impl_* files in your project. See examples/ folder for examples of using this.
// Prefer including the entire imgui/ repository into your project (either as a copy or as a submodule), and only build the backends you need.
// Learn about Dear ImGui:
// - FAQ                  https://dearimgui.com/faq
// - Getting Started      https://dearimgui.com/getting-started
// - Documentation        https://dearimgui.com/docs (same as your local docs/ folder).
// - Introduction, links and more at the top of imgui.cpp

// CHANGELOG
// (minor and older changes stripped away, please see git history for details)
//  2024-10-07: Expose selected render state in ImGui_ImplWGPU_RenderState, which you can access in 'void* platform_io.Renderer_RenderState' during draw callbacks.
//  2024-10-07: Changed default texture sampler to Clamp instead of Repeat/Wrap.
//  2024-09-16: Added support for optional IMGUI_IMPL_WEBGPU_BACKEND_DAWN / IMGUI_IMPL_WEBGPU_BACKEND_WGPU define to handle ever-changing native implementations. (#7977)
//  2024-01-22: Added configurable PipelineMultisampleState struct. (#7240)
//  2024-01-22: (Breaking) ImGui_ImplWGPU_Init() now takes a ImGui_ImplWGPU_InitInfo structure instead of variety of parameters, allowing for easier further changes.
//  2024-01-22: Fixed pipeline layout leak. (#7245)
//  2024-01-17: Explicitly fill all of WGPUDepthStencilState since standard removed defaults.
//  2023-07-13: Use WGPUShaderModuleWGSLDescriptor's code instead of source. use WGPUMipmapFilterMode_Linear instead of WGPUFilterMode_Linear. (#6602)
//  2023-04-11: Align buffer sizes. Use WGSL shaders instead of precompiled SPIR-V.
//  2023-04-11: Reorganized backend to pull data from a single structure to facilitate usage with multiple-contexts (all g_XXXX access changed to bd->XXXX).
//  2023-01-25: Revert automatic pipeline layout generation (see https://github.com/gpuweb/gpuweb/issues/2470)
//  2022-11-24: Fixed validation error with default depth buffer settings.
//  2022-11-10: Fixed rendering when a depth buffer is enabled. Added 'WGPUTextureFormat depth_format' parameter to ImGui_ImplWGPU_Init().
//  2022-10-11: Using 'nullptr' instead of 'NULL' as per our switch to C++11.
//  2021-11-29: Passing explicit buffer sizes to wgpuRenderPassEncoderSetVertexBuffer()/wgpuRenderPassEncoderSetIndexBuffer().
//  2021-08-24: Fixed for latest specs.
//  2021-05-24: Add support for draw_data->FramebufferScale.
//  2021-05-19: Replaced direct access to ImDrawCmd::TextureId with a call to ImDrawCmd::GetTexID(). (will become a requirement)
//  2021-05-16: Update to latest WebGPU specs (compatible with Emscripten 2.0.20 and Chrome Canary 92).
//  2021-02-18: Change blending equation to preserve alpha in output buffer.
//  2021-01-28: Initial version.

// When targeting native platforms (i.e. NOT emscripten), one of IMGUI_IMPL_WEBGPU_BACKEND_DAWN
// or IMGUI_IMPL_WEBGPU_BACKEND_WGPU must be provided. See imgui_impl_wgpu.h for more details.
#ifndef __EMSCRIPTEN__
    #if defined(IMGUI_IMPL_WEBGPU_BACKEND_DAWN) == defined(IMGUI_IMPL_WEBGPU_BACKEND_WGPU)
    #error exactly one of IMGUI_IMPL_WEBGPU_BACKEND_DAWN or IMGUI_IMPL_WEBGPU_BACKEND_WGPU must be defined!
    #endif
#else
    #if defined(IMGUI_IMPL_WEBGPU_BACKEND_DAWN) || defined(IMGUI_IMPL_WEBGPU_BACKEND_WGPU)
    #error neither IMGUI_IMPL_WEBGPU_BACKEND_DAWN nor IMGUI_IMPL_WEBGPU_BACKEND_WGPU may be defined if targeting emscripten!
    #endif
#endif

#include "imgui.h"
#ifndef IMGUI_DISABLE
#include "imgui_impl_wgpu.h"
#include <limits.h>
#include <webgpu/webgpu.h>

// Dear ImGui prototypes from imgui_internal.h
extern ImGuiID ImHashData(const void* data_p, size_t data_size, ImU32 seed = 0);
#define MEMALIGN(_SIZE,_ALIGN)        (((_SIZE) + ((_ALIGN) - 1)) & ~((_ALIGN) - 1))    // Memory align (copied from IM_ALIGN() macro).

// WebGPU data
struct RenderResources
{
    WGPUTexture         FontTexture = nullptr;          // Font texture
    WGPUTextureView     FontTextureView = nullptr;      // Texture view for font texture
    WGPUSampler         Sampler = nullptr;              // Sampler for the font texture
    WGPUBuffer          Uniforms = nullptr;             // Shader uniforms
    WGPUBindGroup       CommonBindGroup = nullptr;      // Resources bind-group to bind the common resources to pipeline
    ImGuiStorage        ImageBindGroups;                // Resources bind-group to bind the font/image resources to pipeline (this is a key->value map)
    WGPUBindGroup       ImageBindGroup = nullptr;       // Default font-resource of Dear ImGui
    WGPUBindGroupLayout ImageBindGroupLayout = nullptr; // Cache layout used for the image bind group. Avoids allocating unnecessary JS objects when working with WebASM
};

struct FrameResources
{
    WGPUBuffer  IndexBuffer;
    WGPUBuffer  VertexBuffer;
    ImDrawIdx*  IndexBufferHost;
    ImDrawVert* VertexBufferHost;
    int         IndexBufferSize;
    int         VertexBufferSize;
};

struct Uniforms
{
    float MVP[4][4];
    float Gamma;
};

struct ImGui_ImplWGPU_Data
{
    ImGui_ImplWGPU_InitInfo initInfo;
    WGPUDevice              wgpuDevice = nullptr;
    WGPUQueue               defaultQueue = nullptr;
    WGPUTextureFormat       renderTargetFormat = WGPUTextureFormat_Undefined;
    WGPUTextureFormat       depthStencilFormat = WGPUTextureFormat_Undefined;
    WGPURenderPipeline      pipelineState = nullptr;

    RenderResources         renderResources;
    FrameResources*         pFrameResources = nullptr;
    unsigned int            numFramesInFlight = 0;
    unsigned int            frameIndex = UINT_MAX;
};

// Backend data stored in io.BackendRendererUserData to allow support for multiple Dear ImGui contexts
// It is STRONGLY preferred that you use docking branch with multi-viewports (== single Dear ImGui context + multiple windows) instead of multiple Dear ImGui contexts.
static ImGui_ImplWGPU_Data* ImGui_ImplWGPU_GetBackendData()
{
    return ImGui::GetCurrentContext() ? (ImGui_ImplWGPU_Data*)ImGui::GetIO().BackendRendererUserData : nullptr;
}

//-----------------------------------------------------------------------------
// SHADERS
//-----------------------------------------------------------------------------

static const char __shader_vert_wgsl[] = R"(
struct VertexInput {
    @location(0) position: vec2<f32>,
    @location(1) uv: vec2<f32>,
    @location(2) color: vec4<f32>,
};

struct VertexOutput {
    @builtin(position) position: vec4<f32>,
    @location(0) color: vec4<f32>,
    @location(1) uv: vec2<f32>,
};

struct Uniforms {
    mvp: mat4x4<f32>,
    gamma: f32,
};

@group(0) @binding(0) var<uniform> uniforms: Uniforms;

@vertex
fn main(in: VertexInput) -> VertexOutput {
    var out: VertexOutput;
    out.position = uniforms.mvp * vec4<f32>(in.position, 0.0, 1.0);
    out.color = in.color;
    out.uv = in.uv;
    return out;
}
)";

static const char __shader_frag_wgsl[] = R"(
struct VertexOutput {
    @builtin(position) position: vec4<f32>,
    @location(0) color: vec4<f32>,
    @location(1) uv: vec2<f32>,
};

struct Uniforms {
    mvp: mat4x4<f32>,
    gamma: f32,
};

@group(0) @binding(0) var<uniform> uniforms: Uniforms;
@group(0) @binding(1) var s: sampler;
@group(1) @binding(0) var t: texture_2d<f32>;

@fragment
fn main(in: VertexOutput) -> @location(0) vec4<f32> {
    let color = in.color * textureSample(t, s, in.uv);
    let corrected_color = pow(color.rgb, vec3<f32>(uniforms.gamma));
    return vec4<f32>(corrected_color, color.a);
}
)";

static void SafeRelease(ImDrawIdx*& res)
{
    if (res)
        delete[] res;
    res = nullptr;
}
static void SafeRelease(ImDrawVert*& res)
{
    if (res)
        delete[] res;
    res = nullptr;
}
static void SafeRelease(WGPUBindGroupLayout& res)
{
    if (res)
        wgpuBindGroupLayoutRelease(res);
    res = nullptr;
}
static void SafeRelease(WGPUBindGroup& res)
{
    if (res)
        wgpuBindGroupRelease(res);
    res = nullptr;
}
static void SafeRelease(WGPUBuffer& res)
{
    if (res)
        wgpuBufferRelease(res);
    res = nullptr;
}
static void SafeRelease(WGPUPipelineLayout& res)
{
    if (res)
        wgpuPipelineLayoutRelease(res);
    res = nullptr;
}
static void SafeRelease(WGPURenderPipeline& res)
{
    if (res)
        wgpuRenderPipelineRelease(res);
    res = nullptr;
}
static void SafeRelease(WGPUSampler& res)
{
    if (res)
        wgpuSamplerRelease(res);
    res = nullptr;
}
static void SafeRelease(WGPUShaderModule& res)
{
    if (res)
        wgpuShaderModuleRelease(res);
    res = nullptr;
}
static void SafeRelease(WGPUTextureView& res)
{
    if (res)
        wgpuTextureViewRelease(res);
    res = nullptr;
}
static void SafeRelease(WGPUTexture& res)
{
    if (res)
        wgpuTextureRelease(res);
    res = nullptr;
}

static void SafeRelease(RenderResources& res)
{
    SafeRelease(res.FontTexture);
    SafeRelease(res.FontTextureView);
    SafeRelease(res.Sampler);
    SafeRelease(res.Uniforms);
    SafeRelease(res.CommonBindGroup);
    SafeRelease(res.ImageBindGroup);
    SafeRelease(res.ImageBindGroupLayout);
};

static void SafeRelease(FrameResources& res)
{
    SafeRelease(res.IndexBuffer);
    SafeRelease(res.VertexBuffer);
    SafeRelease(res.IndexBufferHost);
    SafeRelease(res.VertexBufferHost);
}

static WGPUProgrammableStageDescriptor ImGui_ImplWGPU_CreateShaderModule(const char* wgsl_source)
{
    ImGui_ImplWGPU_Data* bd = ImGui_ImplWGPU_GetBackendData();

#ifdef IMGUI_IMPL_WEBGPU_BACKEND_DAWN
	WGPUShaderSourceWGSL wgsl_desc = {};
    wgsl_desc.chain.sType = WGPUSType_ShaderSourceWGSL;
	wgsl_desc.code = { wgsl_source, WGPU_STRLEN };
#else
	WGPUShaderModuleWGSLDescriptor wgsl_desc = {};
    wgsl_desc.chain.sType = WGPUSType_ShaderModuleWGSLDescriptor;
	wgsl_desc.code = wgsl_source;
#endif

    WGPUShaderModuleDescriptor desc = {};
    desc.nextInChain = reinterpret_cast<WGPUChainedStruct*>(&wgsl_desc);

    WGPUProgrammableStageDescriptor stage_desc = {};
    stage_desc.module = wgpuDeviceCreateShaderModule(bd->wgpuDevice, &desc);
    stage_desc.entryPoint = "main";
    return stage_desc;
}

static WGPUBindGroup ImGui_ImplWGPU_CreateImageBindGroup(WGPUBindGroupLayout layout, WGPUTextureView texture)
{
    ImGui_ImplWGPU_Data* bd = ImGui_ImplWGPU_GetBackendData();
    WGPUBindGroupEntry image_bg_entries[] = { { nullptr, 0, 0, 0, 0, 0, texture } };

    WGPUBindGroupDescriptor image_bg_descriptor = {};
    image_bg_descriptor.layout = layout;
    image_bg_descriptor.entryCount = sizeof(image_bg_entries) / sizeof(WGPUBindGroupEntry);
    image_bg_descriptor.entries = image_bg_entries;
    return wgpuDeviceCreateBindGroup(bd->wgpuDevice, &image_bg_descriptor);
}

static void ImGui_ImplWGPU_SetupRenderState(ImDrawData* draw_data, WGPURenderPassEncoder ctx, FrameResources* fr)
{
    ImGui_ImplWGPU_Data* bd = ImGui_ImplWGPU_GetBackendData();

    // Setup orthographic projection matrix into our constant buffer
    // Our visible imgui space lies from draw_data->DisplayPos (top left) to draw_data->DisplayPos+data_data->DisplaySize (bottom right).
    {
        float L = draw_data->DisplayPos.x;
        float R = draw_data->DisplayPos.x + draw_data->DisplaySize.x;
        float T = draw_data->DisplayPos.y;
        float B = draw_data->DisplayPos.y + draw_data->DisplaySize.y;
        float mvp[4][4] =
        {
            { 2.0f/(R-L),   0.0f,           0.0f,       0.0f },
            { 0.0f,         2.0f/(T-B),     0.0f,       0.0f },
            { 0.0f,         0.0f,           0.5f,       0.0f },
            { (R+L)/(L-R),  (T+B)/(B-T),    0.5f,       1.0f },
        };
        wgpuQueueWriteBuffer(bd->defaultQueue, bd->renderResources.Uniforms, offsetof(Uniforms, MVP), mvp, sizeof(Uniforms::MVP));
        float gamma;
        switch (bd->renderTargetFormat)
        {
        case WGPUTextureFormat_ASTC10x10UnormSrgb:
        case WGPUTextureFormat_ASTC10x5UnormSrgb:
        case WGPUTextureFormat_ASTC10x6UnormSrgb:
        case WGPUTextureFormat_ASTC10x8UnormSrgb:
        case WGPUTextureFormat_ASTC12x10UnormSrgb:
        case WGPUTextureFormat_ASTC12x12UnormSrgb:
        case WGPUTextureFormat_ASTC4x4UnormSrgb:
        case WGPUTextureFormat_ASTC5x5UnormSrgb:
        case WGPUTextureFormat_ASTC6x5UnormSrgb:
        case WGPUTextureFormat_ASTC6x6UnormSrgb:
        case WGPUTextureFormat_ASTC8x5UnormSrgb:
        case WGPUTextureFormat_ASTC8x6UnormSrgb:
        case WGPUTextureFormat_ASTC8x8UnormSrgb:
        case WGPUTextureFormat_BC1RGBAUnormSrgb:
        case WGPUTextureFormat_BC2RGBAUnormSrgb:
        case WGPUTextureFormat_BC3RGBAUnormSrgb:
        case WGPUTextureFormat_BC7RGBAUnormSrgb:
        case WGPUTextureFormat_BGRA8UnormSrgb:
        case WGPUTextureFormat_ETC2RGB8A1UnormSrgb:
        case WGPUTextureFormat_ETC2RGB8UnormSrgb:
        case WGPUTextureFormat_ETC2RGBA8UnormSrgb:
        case WGPUTextureFormat_RGBA8UnormSrgb:
            gamma = 2.2f;
            break;
        default:
            gamma = 1.0f;
        }
        wgpuQueueWriteBuffer(bd->defaultQueue, bd->renderResources.Uniforms, offsetof(Uniforms, Gamma), &gamma, sizeof(Uniforms::Gamma));
    }

    // Setup viewport
    wgpuRenderPassEncoderSetViewport(ctx, 0, 0, draw_data->FramebufferScale.x * draw_data->DisplaySize.x, draw_data->FramebufferScale.y * draw_data->DisplaySize.y, 0, 1);

    // Bind shader and vertex buffers
    wgpuRenderPassEncoderSetVertexBuffer(ctx, 0, fr->VertexBuffer, 0, fr->VertexBufferSize * sizeof(ImDrawVert));
    wgpuRenderPassEncoderSetIndexBuffer(ctx, fr->IndexBuffer, sizeof(ImDrawIdx) == 2 ? WGPUIndexFormat_Uint16 : WGPUIndexFormat_Uint32, 0, fr->IndexBufferSize * sizeof(ImDrawIdx));
    wgpuRenderPassEncoderSetPipeline(ctx, bd->pipelineState);
    wgpuRenderPassEncoderSetBindGroup(ctx, 0, bd->renderResources.CommonBindGroup, 0, nullptr);

    // Setup blend factor
    WGPUColor blend_color = { 0.f, 0.f, 0.f, 0.f };
    wgpuRenderPassEncoderSetBlendConstant(ctx, &blend_color);
}

// Render function
// (this used to be set in io.RenderDrawListsFn and called by ImGui::Render(), but you can now call this directly from your main loop)
void ImGui_ImplWGPU_RenderDrawData(ImDrawData* draw_data, WGPURenderPassEncoder pass_encoder)
{
    // Avoid rendering when minimized
    int fb_width = (int)(draw_data->DisplaySize.x * draw_data->FramebufferScale.x);
    int fb_height = (int)(draw_data->DisplaySize.y * draw_data->FramebufferScale.y);
    if (fb_width <= 0 || fb_height <= 0 || draw_data->CmdListsCount == 0)
        return;

    // FIXME: Assuming that this only gets called once per frame!
    // If not, we can't just re-allocate the IB or VB, we'll have to do a proper allocator.
    ImGui_ImplWGPU_Data* bd = ImGui_ImplWGPU_GetBackendData();
    bd->frameIndex = bd->frameIndex + 1;
    FrameResources* fr = &bd->pFrameResources[bd->frameIndex % bd->numFramesInFlight];

    // Create and grow vertex/index buffers if needed
    if (fr->VertexBuffer == nullptr || fr->VertexBufferSize < draw_data->TotalVtxCount)
    {
        if (fr->VertexBuffer)
        {
            wgpuBufferDestroy(fr->VertexBuffer);
            wgpuBufferRelease(fr->VertexBuffer);
        }
        SafeRelease(fr->VertexBufferHost);
        fr->VertexBufferSize = draw_data->TotalVtxCount + 5000;

        WGPUBufferDescriptor vb_desc =
        {
            nullptr,
            "Dear ImGui Vertex buffer",
            WGPUBufferUsage_CopyDst | WGPUBufferUsage_Vertex,
            MEMALIGN(fr->VertexBufferSize * sizeof(ImDrawVert), 4),
            false
        };
        fr->VertexBuffer = wgpuDeviceCreateBuffer(bd->wgpuDevice, &vb_desc);
        if (!fr->VertexBuffer)
            return;

        fr->VertexBufferHost = new ImDrawVert[fr->VertexBufferSize];
    }
    if (fr->IndexBuffer == nullptr || fr->IndexBufferSize < draw_data->TotalIdxCount)
    {
        if (fr->IndexBuffer)
        {
            wgpuBufferDestroy(fr->IndexBuffer);
            wgpuBufferRelease(fr->IndexBuffer);
        }
        SafeRelease(fr->IndexBufferHost);
        fr->IndexBufferSize = draw_data->TotalIdxCount + 10000;

        WGPUBufferDescriptor ib_desc =
        {
            nullptr,
            "Dear ImGui Index buffer",
            WGPUBufferUsage_CopyDst | WGPUBufferUsage_Index,
            MEMALIGN(fr->IndexBufferSize * sizeof(ImDrawIdx), 4),
            false
        };
        fr->IndexBuffer = wgpuDeviceCreateBuffer(bd->wgpuDevice, &ib_desc);
        if (!fr->IndexBuffer)
            return;

        fr->IndexBufferHost = new ImDrawIdx[fr->IndexBufferSize];
    }

    // Upload vertex/index data into a single contiguous GPU buffer
    ImDrawVert* vtx_dst = (ImDrawVert*)fr->VertexBufferHost;
    ImDrawIdx* idx_dst = (ImDrawIdx*)fr->IndexBufferHost;
    for (int n = 0; n < draw_data->CmdListsCount; n++)
    {
        const ImDrawList* draw_list = draw_data->CmdLists[n];
        memcpy(vtx_dst, draw_list->VtxBuffer.Data, draw_list->VtxBuffer.Size * sizeof(ImDrawVert));
        memcpy(idx_dst, draw_list->IdxBuffer.Data, draw_list->IdxBuffer.Size * sizeof(ImDrawIdx));
        vtx_dst += draw_list->VtxBuffer.Size;
        idx_dst += draw_list->IdxBuffer.Size;
    }
    int64_t vb_write_size = MEMALIGN((char*)vtx_dst - (char*)fr->VertexBufferHost, 4);
    int64_t ib_write_size = MEMALIGN((char*)idx_dst - (char*)fr->IndexBufferHost, 4);
    wgpuQueueWriteBuffer(bd->defaultQueue, fr->VertexBuffer, 0, fr->VertexBufferHost, vb_write_size);
    wgpuQueueWriteBuffer(bd->defaultQueue, fr->IndexBuffer,  0, fr->IndexBufferHost,  ib_write_size);

    // Setup desired render state
    ImGui_ImplWGPU_SetupRenderState(draw_data, pass_encoder, fr);

    // Setup render state structure (for callbacks and custom texture bindings)
    ImGuiPlatformIO& platform_io = ImGui::GetPlatformIO();
    ImGui_ImplWGPU_RenderState render_state;
    render_state.Device = bd->wgpuDevice;
    render_state.RenderPassEncoder = pass_encoder;
    platform_io.Renderer_RenderState = &render_state;

    // Render command lists
    // (Because we merged all buffers into a single one, we maintain our own offset into them)
    int global_vtx_offset = 0;
    int global_idx_offset = 0;
    ImVec2 clip_scale = draw_data->FramebufferScale;
    ImVec2 clip_off = draw_data->DisplayPos;
    for (int n = 0; n < draw_data->CmdListsCount; n++)
    {
        const ImDrawList* draw_list = draw_data->CmdLists[n];
        for (int cmd_i = 0; cmd_i < draw_list->CmdBuffer.Size; cmd_i++)
        {
            const ImDrawCmd* pcmd = &draw_list->CmdBuffer[cmd_i];
            if (pcmd->UserCallback != nullptr)
            {
                // User callback, registered via ImDrawList::AddCallback()
                // (ImDrawCallback_ResetRenderState is a special callback value used by the user to request the renderer to reset render state.)
                if (pcmd->UserCallback == ImDrawCallback_ResetRenderState)
                    ImGui_ImplWGPU_SetupRenderState(draw_data, pass_encoder, fr);
                else
                    pcmd->UserCallback(draw_list, pcmd);
            }
            else
            {
                // Bind custom texture
                ImTextureID tex_id = pcmd->GetTexID();
                ImGuiID tex_id_hash = ImHashData(&tex_id, sizeof(tex_id));
                auto bind_group = bd->renderResources.ImageBindGroups.GetVoidPtr(tex_id_hash);
                if (bind_group)
                {
                    wgpuRenderPassEncoderSetBindGroup(pass_encoder, 1, (WGPUBindGroup)bind_group, 0, nullptr);
                }
                else
                {
                    WGPUBindGroup image_bind_group = ImGui_ImplWGPU_CreateImageBindGroup(bd->renderResources.ImageBindGroupLayout, (WGPUTextureView)tex_id);
                    bd->renderResources.ImageBindGroups.SetVoidPtr(tex_id_hash, image_bind_group);
                    wgpuRenderPassEncoderSetBindGroup(pass_encoder, 1, image_bind_group, 0, nullptr);
                }

                // Project scissor/clipping rectangles into framebuffer space
                ImVec2 clip_min((pcmd->ClipRect.x - clip_off.x) * clip_scale.x, (pcmd->ClipRect.y - clip_off.y) * clip_scale.y);
                ImVec2 clip_max((pcmd->ClipRect.z - clip_off.x) * clip_scale.x, (pcmd->ClipRect.w - clip_off.y) * clip_scale.y);

                // Clamp to viewport as wgpuRenderPassEncoderSetScissorRect() won't accept values that are off bounds
                if (clip_min.x < 0.0f) { clip_min.x = 0.0f; }
                if (clip_min.y < 0.0f) { clip_min.y = 0.0f; }
                if (clip_max.x > fb_width) { clip_max.x = (float)fb_width; }
                if (clip_max.y > fb_height) { clip_max.y = (float)fb_height; }
                if (clip_max.x <= clip_min.x || clip_max.y <= clip_min.y)
                    continue;

                // Apply scissor/clipping rectangle, Draw
                wgpuRenderPassEncoderSetScissorRect(pass_encoder, (uint32_t)clip_min.x, (uint32_t)clip_min.y, (uint32_t)(clip_max.x - clip_min.x), (uint32_t)(clip_max.y - clip_min.y));
                wgpuRenderPassEncoderDrawIndexed(pass_encoder, pcmd->ElemCount, 1, pcmd->IdxOffset + global_idx_offset, pcmd->VtxOffset + global_vtx_offset, 0);
            }
        }
        global_idx_offset += draw_list->IdxBuffer.Size;
        global_vtx_offset += draw_list->VtxBuffer.Size;
    }
    platform_io.Renderer_RenderState = NULL;
}

static void ImGui_ImplWGPU_CreateFontsTexture()
{
    // Build texture atlas
    ImGui_ImplWGPU_Data* bd = ImGui_ImplWGPU_GetBackendData();
    ImGuiIO& io = ImGui::GetIO();
    unsigned char* pixels;
    int width, height, size_pp;
    io.Fonts->GetTexDataAsRGBA32(&pixels, &width, &height, &size_pp);

    // Upload texture to graphics system
    {
        WGPUTextureDescriptor tex_desc = {};
        tex_desc.label = "Dear ImGui Font Texture";
        tex_desc.dimension = WGPUTextureDimension_2D;
        tex_desc.size.width = width;
        tex_desc.size.height = height;
        tex_desc.size.depthOrArrayLayers = 1;
        tex_desc.sampleCount = 1;
        tex_desc.format = WGPUTextureFormat_RGBA8Unorm;
        tex_desc.mipLevelCount = 1;
        tex_desc.usage = WGPUTextureUsage_CopyDst | WGPUTextureUsage_TextureBinding;
        bd->renderResources.FontTexture = wgpuDeviceCreateTexture(bd->wgpuDevice, &tex_desc);

        WGPUTextureViewDescriptor tex_view_desc = {};
        tex_view_desc.format = WGPUTextureFormat_RGBA8Unorm;
        tex_view_desc.dimension = WGPUTextureViewDimension_2D;
        tex_view_desc.baseMipLevel = 0;
        tex_view_desc.mipLevelCount = 1;
        tex_view_desc.baseArrayLayer = 0;
        tex_view_desc.arrayLayerCount = 1;
        tex_view_desc.aspect = WGPUTextureAspect_All;
        bd->renderResources.FontTextureView = wgpuTextureCreateView(bd->renderResources.FontTexture, &tex_view_desc);
    }

    // Upload texture data
    {
        WGPUImageCopyTexture dst_view = {};
        dst_view.texture = bd->renderResources.FontTexture;
        dst_view.mipLevel = 0;
        dst_view.origin = { 0, 0, 0 };
        dst_view.aspect = WGPUTextureAspect_All;
        WGPUTextureDataLayout layout = {};
        layout.offset = 0;
        layout.bytesPerRow = width * size_pp;
        layout.rowsPerImage = height;
        WGPUExtent3D size = { (uint32_t)width, (uint32_t)height, 1 };
        wgpuQueueWriteTexture(bd->defaultQueue, &dst_view, pixels, (uint32_t)(width * size_pp * height), &layout, &size);
    }

    // Create the associated sampler
    // (Bilinear sampling is required by default. Set 'io.Fonts->Flags |= ImFontAtlasFlags_NoBakedLines' or 'style.AntiAliasedLinesUseTex = false' to allow point/nearest sampling)
    {
        WGPUSamplerDescriptor sampler_desc = {};
        sampler_desc.minFilter = WGPUFilterMode_Linear;
        sampler_desc.magFilter = WGPUFilterMode_Linear;
        sampler_desc.mipmapFilter = WGPUMipmapFilterMode_Linear;
        sampler_desc.addressModeU = WGPUAddressMode_ClampToEdge;
        sampler_desc.addressModeV = WGPUAddressMode_ClampToEdge;
        sampler_desc.addressModeW = WGPUAddressMode_ClampToEdge;
        sampler_desc.maxAnisotropy = 1;
        bd->renderResources.Sampler = wgpuDeviceCreateSampler(bd->wgpuDevice, &sampler_desc);
    }

    // Store our identifier
    static_assert(sizeof(ImTextureID) >= sizeof(bd->renderResources.FontTexture), "Can't pack descriptor handle into TexID, 32-bit not supported yet.");
    io.Fonts->SetTexID((ImTextureID)bd->renderResources.FontTextureView);
}

static void ImGui_ImplWGPU_CreateUniformBuffer()
{
    ImGui_ImplWGPU_Data* bd = ImGui_ImplWGPU_GetBackendData();
    WGPUBufferDescriptor ub_desc =
    {
        nullptr,
        "Dear ImGui Uniform buffer",
        WGPUBufferUsage_CopyDst | WGPUBufferUsage_Uniform,
        MEMALIGN(sizeof(Uniforms), 16),
        false
    };
    bd->renderResources.Uniforms = wgpuDeviceCreateBuffer(bd->wgpuDevice, &ub_desc);
}

bool ImGui_ImplWGPU_CreateDeviceObjects()
{
    ImGui_ImplWGPU_Data* bd = ImGui_ImplWGPU_GetBackendData();
    if (!bd->wgpuDevice)
        return false;
    if (bd->pipelineState)
        ImGui_ImplWGPU_InvalidateDeviceObjects();

    // Create render pipeline
    WGPURenderPipelineDescriptor graphics_pipeline_desc = {};
    graphics_pipeline_desc.primitive.topology = WGPUPrimitiveTopology_TriangleList;
    graphics_pipeline_desc.primitive.stripIndexFormat = WGPUIndexFormat_Undefined;
    graphics_pipeline_desc.primitive.frontFace = WGPUFrontFace_CW;
    graphics_pipeline_desc.primitive.cullMode = WGPUCullMode_None;
    graphics_pipeline_desc.multisample = bd->initInfo.PipelineMultisampleState;

    // Bind group layouts
    WGPUBindGroupLayoutEntry common_bg_layout_entries[2] = {};
    common_bg_layout_entries[0].binding = 0;
    common_bg_layout_entries[0].visibility = WGPUShaderStage_Vertex | WGPUShaderStage_Fragment;
    common_bg_layout_entries[0].buffer.type = WGPUBufferBindingType_Uniform;
    common_bg_layout_entries[1].binding = 1;
    common_bg_layout_entries[1].visibility = WGPUShaderStage_Fragment;
    common_bg_layout_entries[1].sampler.type = WGPUSamplerBindingType_Filtering;

    WGPUBindGroupLayoutEntry image_bg_layout_entries[1] = {};
    image_bg_layout_entries[0].binding = 0;
    image_bg_layout_entries[0].visibility = WGPUShaderStage_Fragment;
    image_bg_layout_entries[0].texture.sampleType = WGPUTextureSampleType_Float;
    image_bg_layout_entries[0].texture.viewDimension = WGPUTextureViewDimension_2D;

    WGPUBindGroupLayoutDescriptor common_bg_layout_desc = {};
    common_bg_layout_desc.entryCount = 2;
    common_bg_layout_desc.entries = common_bg_layout_entries;

    WGPUBindGroupLayoutDescriptor image_bg_layout_desc = {};
    image_bg_layout_desc.entryCount = 1;
    image_bg_layout_desc.entries = image_bg_layout_entries;

    WGPUBindGroupLayout bg_layouts[2];
    bg_layouts[0] = wgpuDeviceCreateBindGroupLayout(bd->wgpuDevice, &common_bg_layout_desc);
    bg_layouts[1] = wgpuDeviceCreateBindGroupLayout(bd->wgpuDevice, &image_bg_layout_desc);

    WGPUPipelineLayoutDescriptor layout_desc = {};
    layout_desc.bindGroupLayoutCount = 2;
    layout_desc.bindGroupLayouts = bg_layouts;
    graphics_pipeline_desc.layout = wgpuDeviceCreatePipelineLayout(bd->wgpuDevice, &layout_desc);

    // Create the vertex shader
    WGPUProgrammableStageDescriptor vertex_shader_desc = ImGui_ImplWGPU_CreateShaderModule(__shader_vert_wgsl);
    graphics_pipeline_desc.vertex.module = vertex_shader_desc.module;
    graphics_pipeline_desc.vertex.entryPoint = vertex_shader_desc.entryPoint;

    // Vertex input configuration
    WGPUVertexAttribute attribute_desc[] =
    {
        { WGPUVertexFormat_Float32x2, (uint64_t)offsetof(ImDrawVert, pos), 0 },
        { WGPUVertexFormat_Float32x2, (uint64_t)offsetof(ImDrawVert, uv),  1 },
        { WGPUVertexFormat_Unorm8x4,  (uint64_t)offsetof(ImDrawVert, col), 2 },
    };

    WGPUVertexBufferLayout buffer_layouts[1];
    buffer_layouts[0].arrayStride = sizeof(ImDrawVert);
    buffer_layouts[0].stepMode = WGPUVertexStepMode_Vertex;
    buffer_layouts[0].attributeCount = 3;
    buffer_layouts[0].attributes = attribute_desc;

    graphics_pipeline_desc.vertex.bufferCount = 1;
    graphics_pipeline_desc.vertex.buffers = buffer_layouts;

    // Create the pixel shader
    WGPUProgrammableStageDescriptor pixel_shader_desc = ImGui_ImplWGPU_CreateShaderModule(__shader_frag_wgsl);

    // Create the blending setup
    WGPUBlendState blend_state = {};
    blend_state.alpha.operation = WGPUBlendOperation_Add;
    blend_state.alpha.srcFactor = WGPUBlendFactor_One;
    blend_state.alpha.dstFactor = WGPUBlendFactor_OneMinusSrcAlpha;
    blend_state.color.operation = WGPUBlendOperation_Add;
    blend_state.color.srcFactor = WGPUBlendFactor_SrcAlpha;
    blend_state.color.dstFactor = WGPUBlendFactor_OneMinusSrcAlpha;

    WGPUColorTargetState color_state = {};
    color_state.format = bd->renderTargetFormat;
    color_state.blend = &blend_state;
    color_state.writeMask = WGPUColorWriteMask_All;

    WGPUFragmentState fragment_state = {};
    fragment_state.module = pixel_shader_desc.module;
    fragment_state.entryPoint = pixel_shader_desc.entryPoint;
    fragment_state.targetCount = 1;
    fragment_state.targets = &color_state;

    graphics_pipeline_desc.fragment = &fragment_state;

    // Create depth-stencil State
    WGPUDepthStencilState depth_stencil_state = {};
    depth_stencil_state.format = bd->depthStencilFormat;
#ifdef IMGUI_IMPL_WEBGPU_BACKEND_DAWN
    depth_stencil_state.depthWriteEnabled = WGPUOptionalBool_False;
#else
    depth_stencil_state.depthWriteEnabled = false;
#endif
    depth_stencil_state.depthCompare = WGPUCompareFunction_Always;
    depth_stencil_state.stencilFront.compare = WGPUCompareFunction_Always;
    depth_stencil_state.stencilFront.failOp = WGPUStencilOperation_Keep;
    depth_stencil_state.stencilFront.depthFailOp = WGPUStencilOperation_Keep;
    depth_stencil_state.stencilFront.passOp = WGPUStencilOperation_Keep;
    depth_stencil_state.stencilBack.compare = WGPUCompareFunction_Always;
    depth_stencil_state.stencilBack.failOp = WGPUStencilOperation_Keep;
    depth_stencil_state.stencilBack.depthFailOp = WGPUStencilOperation_Keep;
    depth_stencil_state.stencilBack.passOp = WGPUStencilOperation_Keep;

    // Configure disabled depth-stencil state
    graphics_pipeline_desc.depthStencil = (bd->depthStencilFormat == WGPUTextureFormat_Undefined) ? nullptr :  &depth_stencil_state;

    bd->pipelineState = wgpuDeviceCreateRenderPipeline(bd->wgpuDevice, &graphics_pipeline_desc);

    ImGui_ImplWGPU_CreateFontsTexture();
    ImGui_ImplWGPU_CreateUniformBuffer();

    // Create resource bind group
    WGPUBindGroupEntry common_bg_entries[] =
    {
        { nullptr, 0, bd->renderResources.Uniforms, 0, MEMALIGN(sizeof(Uniforms), 16), 0, 0 },
        { nullptr, 1, 0, 0, 0, bd->renderResources.Sampler, 0 },
    };

    WGPUBindGroupDescriptor common_bg_descriptor = {};
    common_bg_descriptor.layout = bg_layouts[0];
    common_bg_descriptor.entryCount = sizeof(common_bg_entries) / sizeof(WGPUBindGroupEntry);
    common_bg_descriptor.entries = common_bg_entries;
    bd->renderResources.CommonBindGroup = wgpuDeviceCreateBindGroup(bd->wgpuDevice, &common_bg_descriptor);

    WGPUBindGroup image_bind_group = ImGui_ImplWGPU_CreateImageBindGroup(bg_layouts[1], bd->renderResources.FontTextureView);
    bd->renderResources.ImageBindGroup = image_bind_group;
    bd->renderResources.ImageBindGroupLayout = bg_layouts[1];
    bd->renderResources.ImageBindGroups.SetVoidPtr(ImHashData(&bd->renderResources.FontTextureView, sizeof(ImTextureID)), image_bind_group);

    SafeRelease(vertex_shader_desc.module);
    SafeRelease(pixel_shader_desc.module);
    SafeRelease(graphics_pipeline_desc.layout);
    SafeRelease(bg_layouts[0]);

    return true;
}

void ImGui_ImplWGPU_InvalidateDeviceObjects()
{
    ImGui_ImplWGPU_Data* bd = ImGui_ImplWGPU_GetBackendData();
    if (!bd->wgpuDevice)
        return;

    SafeRelease(bd->pipelineState);
    SafeRelease(bd->renderResources);

    ImGuiIO& io = ImGui::GetIO();
    io.Fonts->SetTexID(0); // We copied g_pFontTextureView to io.Fonts->TexID so let's clear that as well.

    for (unsigned int i = 0; i < bd->numFramesInFlight; i++)
        SafeRelease(bd->pFrameResources[i]);
}

bool ImGui_ImplWGPU_Init(ImGui_ImplWGPU_InitInfo* init_info)
{
    ImGuiIO& io = ImGui::GetIO();
    IMGUI_CHECKVERSION();
    IM_ASSERT(io.BackendRendererUserData == nullptr && "Already initialized a renderer backend!");

    // Setup backend capabilities flags
    ImGui_ImplWGPU_Data* bd = IM_NEW(ImGui_ImplWGPU_Data)();
    io.BackendRendererUserData = (void*)bd;
#if defined(__EMSCRIPTEN__)
    io.BackendRendererName = "imgui_impl_webgpu_emscripten";
#elif defined(IMGUI_IMPL_WEBGPU_BACKEND_DAWN)
    io.BackendRendererName = "imgui_impl_webgpu_dawn";
#elif defined(IMGUI_IMPL_WEBGPU_BACKEND_WGPU)
    io.BackendRendererName = "imgui_impl_webgpu_wgpu";
#else
    io.BackendRendererName = "imgui_impl_webgpu";
#endif
    io.BackendFlags |= ImGuiBackendFlags_RendererHasVtxOffset;  // We can honor the ImDrawCmd::VtxOffset field, allowing for large meshes.

    bd->initInfo = *init_info;
    bd->wgpuDevice = init_info->Device;
    bd->defaultQueue = wgpuDeviceGetQueue(bd->wgpuDevice);
    bd->renderTargetFormat = init_info->RenderTargetFormat;
    bd->depthStencilFormat = init_info->DepthStencilFormat;
    bd->numFramesInFlight = init_info->NumFramesInFlight;
    bd->frameIndex = UINT_MAX;

    bd->renderResources.FontTexture = nullptr;
    bd->renderResources.FontTextureView = nullptr;
    bd->renderResources.Sampler = nullptr;
    bd->renderResources.Uniforms = nullptr;
    bd->renderResources.CommonBindGroup = nullptr;
    bd->renderResources.ImageBindGroups.Data.reserve(100);
    bd->renderResources.ImageBindGroup = nullptr;
    bd->renderResources.ImageBindGroupLayout = nullptr;

    // Create buffers with a default size (they will later be grown as needed)
    bd->pFrameResources = new FrameResources[bd->numFramesInFlight];
    for (unsigned int i = 0; i < bd->numFramesInFlight; i++)
    {
        FrameResources* fr = &bd->pFrameResources[i];
        fr->IndexBuffer = nullptr;
        fr->VertexBuffer = nullptr;
        fr->IndexBufferHost = nullptr;
        fr->VertexBufferHost = nullptr;
        fr->IndexBufferSize = 10000;
        fr->VertexBufferSize = 5000;
    }

    return true;
}

void ImGui_ImplWGPU_Shutdown()
{
    ImGui_ImplWGPU_Data* bd = ImGui_ImplWGPU_GetBackendData();
    IM_ASSERT(bd != nullptr && "No renderer backend to shutdown, or already shutdown?");
    ImGuiIO& io = ImGui::GetIO();

    ImGui_ImplWGPU_InvalidateDeviceObjects();
    delete[] bd->pFrameResources;
    bd->pFrameResources = nullptr;
    wgpuQueueRelease(bd->defaultQueue);
    bd->wgpuDevice = nullptr;
    bd->numFramesInFlight = 0;
    bd->frameIndex = UINT_MAX;

    io.BackendRendererName = nullptr;
    io.BackendRendererUserData = nullptr;
    io.BackendFlags &= ~ImGuiBackendFlags_RendererHasVtxOffset;
    IM_DELETE(bd);
}

void ImGui_ImplWGPU_NewFrame()
{
    ImGui_ImplWGPU_Data* bd = ImGui_ImplWGPU_GetBackendData();
    if (!bd->pipelineState)
        ImGui_ImplWGPU_CreateDeviceObjects();
}

//-----------------------------------------------------------------------------

#endif // #ifndef IMGUI_DISABLE