当我从内存(包含从tga文件加载的RGBA像素数据)将图像缓冲区写回到tga文件时,有时会观察到奇怪的行为。

图像缓冲区是从TGA文件中加载的,该文件的算法被盗并从此线程改编而来:

Loading a tga/bmp file in C++/OpenGL

我从此地址窃取并改编了该写作方法:

http://www.paulbourke.net/dataformats/tga/

这是一个最小的可编译示例,其中tileA.tga已正确存储为tileA_new.tga,但又损坏了TileB_new.tga(图形看起来很奇怪,带有错误的像素)!为什么TileB_new.tga损坏了?

两个源tga文件是不同的,但是可以在gimp和irfanview中正确查看它们,并且我仔细检查了负载算法。之所以有效,是因为当我将两个加载的图块的图像缓冲区(使用OpenGL)渲染到屏幕上时,它们看起来都正确!但是将原始缓冲区写入磁盘的行为却有所不同,为什么?我在十六进制编辑器中比较了源tga文件的标题,但它们是相等的。同样,写入的tga文件具有相同的标头。
我看到的是,tileB.tga的大小是tileA.tga的5倍,但这似乎是正确的,因为gimp / irfanview显示了它们的正确性。也许您可以看到我在这里犯的错误?

//包含两个tga文件的小型visual-studio-project项目,可在此处下载
https://www.file-upload.net/download-13208817/StackOverflowTGA.zip.html

最小示例:

#include <vector>
#include <fstream>

//special-sausage for microsoft
#ifdef _MSC_VER
#pragma warning(disable:4996)
#endif

//=========================================
// Code for loading a TGA-file
//=========================================
typedef union PixelInfo
{
    std::uint32_t Colour;
    struct
    {
        std::uint8_t R, G, B, A;
    };
} *PPixelInfo;

class Tga
{
private:
    std::vector<std::uint8_t> Pixels;
    bool ImageCompressed;
    std::uint32_t width, height, size, BitsPerPixel;

public:
    Tga(const char* FilePath);
    std::vector<std::uint8_t> GetPixels() { return this->Pixels; }
    std::uint32_t GetWidth() const { return this->width; }
    std::uint32_t GetHeight() const { return this->height; }
    std::uint32_t GetBitsPerPixel() const { return this->BitsPerPixel; }
    bool HasAlphaChannel() { return BitsPerPixel == 32; }
};

Tga::Tga(const char* FilePath)
{
    std::fstream hFile(FilePath, std::ios::in | std::ios::binary);
    if (!hFile.is_open()) { throw std::invalid_argument("File Not Found."); }

    std::uint8_t Header[18] = { 0 };
    std::vector<std::uint8_t> ImageData;
    static std::uint8_t DeCompressed[12] = { 0x0, 0x0, 0x2, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 };
    static std::uint8_t IsCompressed[12] = { 0x0, 0x0, 0xA, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0 };

    hFile.read(reinterpret_cast<char*>(&Header), sizeof(Header));

    if (!std::memcmp(DeCompressed, &Header, sizeof(DeCompressed)))
    {
        BitsPerPixel = Header[16];
        width = Header[13] * 256 + Header[12];
        height = Header[15] * 256 + Header[14];
        size = ((width * BitsPerPixel + 31) / 32) * 4 * height;

        if ((BitsPerPixel != 24) && (BitsPerPixel != 32))
        {
            hFile.close();
            throw std::invalid_argument("Invalid File Format. Required: 24 or 32 Bit Image.");
        }

        ImageData.resize(size);
        ImageCompressed = false;
        hFile.read(reinterpret_cast<char*>(ImageData.data()), size);
    }
    else if (!std::memcmp(IsCompressed, &Header, sizeof(IsCompressed)))
    {
        BitsPerPixel = Header[16];
        width = Header[13] * 256 + Header[12];
        height = Header[15] * 256 + Header[14];
        size = ((width * BitsPerPixel + 31) / 32) * 4 * height;

        if ((BitsPerPixel != 24) && (BitsPerPixel != 32))
        {
            hFile.close();
            throw std::invalid_argument("Invalid File Format. Required: 24 or 32 Bit Image.");
        }

        PixelInfo Pixel = { 0 };
        int CurrentByte = 0;
        std::size_t CurrentPixel = 0;
        ImageCompressed = true;
        std::uint8_t ChunkHeader = { 0 };
        int BytesPerPixel = (BitsPerPixel / 8);
        ImageData.resize(static_cast<size_t>(width) * static_cast<size_t>(height) * sizeof(PixelInfo));

        do
        {
            hFile.read(reinterpret_cast<char*>(&ChunkHeader), sizeof(ChunkHeader));

            if (ChunkHeader < 128)
            {
                ++ChunkHeader;
                for (int I = 0; I < ChunkHeader; ++I, ++CurrentPixel)
                {
                    hFile.read(reinterpret_cast<char*>(&Pixel), BytesPerPixel);

                    ImageData[CurrentByte++] = Pixel.B;
                    ImageData[CurrentByte++] = Pixel.G;
                    ImageData[CurrentByte++] = Pixel.R;
                    if (BitsPerPixel > 24) ImageData[CurrentByte++] = Pixel.A;
                }
            }
            else
            {
                ChunkHeader -= 127;
                hFile.read(reinterpret_cast<char*>(&Pixel), BytesPerPixel);

                for (int I = 0; I < ChunkHeader; ++I, ++CurrentPixel)
                {
                    ImageData[CurrentByte++] = Pixel.B;
                    ImageData[CurrentByte++] = Pixel.G;
                    ImageData[CurrentByte++] = Pixel.R;
                    if (BitsPerPixel > 24) ImageData[CurrentByte++] = Pixel.A;
                }
            }
        } while (CurrentPixel < (static_cast<size_t>(width) * static_cast<size_t>(height)));
    }
    else
    {
        hFile.close();
        throw std::invalid_argument("Invalid File Format. Required: 24 or 32 Bit TGA File.");
    }

    hFile.close();
    this->Pixels = ImageData;
}

//=========================================
// code for writing a TGA-file
//=========================================
void writeTGA(const std::string &refFile, Tga &refTGA)
{
    unsigned short width = static_cast<unsigned short>(refTGA.GetWidth());
    unsigned short height = static_cast<unsigned short>(refTGA.GetWidth());
    unsigned char bitsPerPixel = static_cast<unsigned char>(refTGA.GetBitsPerPixel());
    unsigned char bitsAlphaChannel = (bitsPerPixel == 32 ? 8 : 0);

    FILE * fptr = fopen(refFile.c_str(), "w");

    putc(0, fptr);
    putc(0, fptr);
    putc(2, fptr);                          /* uncompressed RGB */
    putc(0, fptr); putc(0, fptr);
    putc(0, fptr); putc(0, fptr);
    putc(0, fptr);
    putc(0, fptr); putc(0, fptr);           /* X origin */
    putc(0, fptr); putc(0, fptr);           /* y origin */
    putc((width & 0x00FF), fptr);
    putc((width & 0xFF00) / 256, fptr);
    putc((height & 0x00FF), fptr);
    putc((height & 0xFF00) / 256, fptr);
    putc(bitsPerPixel, fptr);               /* 24/32 bit bitmap */
    putc(bitsAlphaChannel, fptr);           /* When 32 bit, write 8, else 0 */

    auto pixelData = refTGA.GetPixels();

    for (size_t i = 0; i < static_cast<size_t>(width) * static_cast<size_t>(height) * (bitsPerPixel/8); i += (bitsPerPixel/8))
    {
        unsigned char r = pixelData[i];
        unsigned char g = pixelData[i + 1];
        unsigned char b = pixelData[i + 2];
        unsigned char a = (bitsAlphaChannel == 8 ? pixelData[i + 3] : 0);

        putc(b, fptr);
        putc(g, fptr);
        putc(r, fptr);

        if (bitsAlphaChannel == 8)
            putc(a, fptr);
    }

    fclose(fptr);
}

//=========================================
// main
//=========================================
int main()
{
    Tga oTgaA("tileA.tga");
    writeTGA("tileA_new.tga", oTgaA);  // works correct as aspected

    Tga oTgaB("tileB.tga");
    writeTGA("tileB_new.tga", oTgaB); // graphic-file has artefacts, why?
}

最佳答案

由于我的评论似乎已经解决了问题(请参见上面的^^^^)。我会在这里扩展。

在Windows(至少具有Microsoft CRT)上,以文本模式写入文件和以二进制模式写入之间存在显着差异。

具体来说,与“ \ n”匹配的任何字符都将扩展为两个字符序列“ \ r \ n”。某些特定功能还会在MB和Unicode字符之间应用转换。有关更多信息,请参见fopen上的MSDN文档,网址为:https://msdn.microsoft.com/en-us/library/yeby3zcb.aspx

因此,在读取/写入非文本数据时,请确保将"rb""wb"标志适当地传递给fopen

在Posix系统上,此注意事项不适用,但是仍然要使您的意图清晰明了。

关于c++ - 从内存写入时,有时TGA文件损坏,我们在Stack Overflow上找到一个类似的问题:https://stackoverflow.com/questions/51114208/

10-11 18:06