GET IMAGE FROM OV7670 WITH VARIOUS RESOLUTION

In my last research, I was strugling to get basic image from this camera module OV7670 because I cannot use default VGA image to be processed (stored) in my microcontroller, which is STM32F103RET6. The maximum resolution that the microcontroller can process is QVGA resolution.

Base on implementation guide from Omnivision, you just need to set following registers from your microcontroller.
For QVGA resolution.

{0x0c, 0x0c},
{0x3e, 0x11},
{0x70, 0x3a},
{0x71, 0x35},
{0x72, 0x11},
{0x73, 0xf1},
{0xa2, 0x02},

For QCIF resolution.

{0x0c, 0x0c},
{0x3e, 0x11},
{0x70, 0x3a},
{0x71, 0x35},
{0x72, 0x11},
{0x73, 0xf1},
{0xa2, 0x52},

But when you make a trial using above code snippet, you still get inproper QVGA image. To fix this, you need to set windowing register to make sure your QVGA image is a 320x240 image. You need to specify value for HSART, HSTOP, VSTART, and VSTOP.
For QVGA resolution.

{0x32, 0x80},
{0x17, 0x16},//22
{0x18, 0x04},//04
{0x19, 0x02},
{0x1a, 0x7b},
{0x03, 0x06},

For QCIF resolution.

{0x32, 0x80},
{0x17, 0x3a},
{0x18, 0x03},
{0x19, 0x02},
{0x1a, 0x7b},
{0x03, 0x06},

And now, you will get a proper sized image. But wait, in this way, you will still get one or two broken images when you try the camera module on different light condition. This happens because the default camera setting enables Auto Gain Control algorithm. When the camera module automatically adjust gain value against the light intensity, it automatically adds dummy pixels too. To deal with this, we need to limit maximum gain value.

{0x13, 0xe0},
{0x00, 0x00},
{0x14, 0x48},
{0x13, 0xe5},

And change dummy row insertion according to the maximum gain value.

{0xa4, 0x89},

And finally, you will get stable camera operation. You could make some experiments using different AGC method too. Have a good luck.