2023.04.10 15:23 "[Tiff] CFAREPEATPATTERNDIM and CFAPATTERN", by David C. Partridge

I *think* I have these two worked out:

Reading:

constexpr uint8_t TIFF_CFAPattern_RGGB[] { 0,1,1,2 }; 
constexpr uint8_t TIFF_CFAPattern_BGGR[] { 2,1,1,0 }; 
constexpr uint8_t TIFF_CFAPattern_GRBG[] { 1,0,2,1 }; 
constexpr uint8_t TIFF_CFAPattern_GBRG[] { 1,2,0,1 }; 

// 
/ Attempt to read the CFA from the root IFD if this is a CFA image 
// 
if (PHOTOMETRIC_CFA == photo) 
{ 
       uint16_t cfarepeatpatterndim[2]{ 0 }; 
       if (TIFFGetField(m_tiff, TIFFTAG_CFAREPEATPATTERNDIM, &cfarepeatpatterndim)) 
       { 
             uint16_t 
                    x{ (cfarepeatpatterndim[0]) }, 
                    y{ (cfarepeatpatterndim[1]) }; 
             uint8_t* cfapattern = static_cast<uint8_t*>(_alloca(x * y)); 
             TIFFGetField(m_tiff, TIFFTAG_CFAPATTERN, x*y, cfapattern); 
       } 
}

and Writing:

TIFFSetField(m_tiff, TIFFTAG_PHOTOMETRIC, PHOTOMETRIC_CFA); 
constexpr uint16_t cfapatterndim[2]{ 2,2 }; 
TIFFSetField(m_tiff, TIFFTAG_CFAREPEATPATTERNDIM, cfapatterndim); 
// 
// Note that when writing the CFA pattern, need to specify how many 
// octets are to be written. 
// 
switch (cfatype) 
{ 
case CFATYPE_BGGR: 
TIFFSetField(m_tiff, TIFFTAG_CFAPATTERN, 4, TIFF_CFAPattern_BGGR); 
       break; 
case CFATYPE_GRBG: 
       TIFFSetField(m_tiff, TIFFTAG_CFAPATTERN, 4, TIFF_CFAPattern_GRBG); 
       break; 
case CFATYPE_GBRG: 
       TIFFSetField(m_tiff, TIFFTAG_CFAPATTERN, 4, TIFF_CFAPattern_GBRG); 
       break; 
case CFATYPE_RGGB: 
       TIFFSetField(m_tiff, TIFFTAG_CFAPATTERN, 4, TIFF_CFAPattern_RGGB); 
       break; 
}

I'd be most grateful if someone who knows TIFF a lot better than I would validate that code.

However, I'm still a bit puzzled how I should read/write the data for EXIFTAG_CFAPATTERN which is defined as:

   {EXIFTAG_CFAPATTERN, -1, -1, TIFF_UNDEFINED, 0, TIFF_SETGET_C16_UINT8,  TIFF_SETGET_UNDEFINED, FIELD_CUSTOM, 1, 1, "CFAPattern", NULL}

as the length is a priori unknown AFAICT?

PS could someone please explain how I should interpret the meaning of e.g. TIFF_SETGET_C16_UINT8

Thank you, David