Solidity allows implicit as well as explicit conversion. Solidity compiler allows implicit conversion between two data types provided no implicit conversion is possible and there is no loss of information. For example uint8 is convertible to uint16 but int8 is convertible to uint256 as int8 can contain negative value not allowed in uint256.

Explicit Conversion

We can explicitly convert a data type to another using constructor syntax.

int8 y = -3;
uint x = uint(y);
//Now x = 0xfffff..fd == two complement representation of -3 in 256 bit format.

Conversion to smaller type costs higher order bits.

uint32 a = 0x12345678;
uint16 b = uint16(a); // b = 0x5678

Conversion to higher type adds padding bits to the left.

uint16 a = 0x1234;
uint32 b = uint32(a); // b = 0x00001234 

Conversion to smaller byte costs higher order data.

bytes2 a = 0x1234;
bytes1 b = bytes1(a); // b = 0x12

Conversion to larger byte add padding bits to the right.

bytes2 a = 0x1234;
bytes4 b = bytes4(a); // b = 0x12340000

Conversion between fixed size bytes and int is only possible when both are of same size.

bytes2 a = 0x1234;
uint32 b = uint16(a); // b = 0x00001234
uint32 c = uint32(bytes4(a)); // c = 0x12340000
uint8 d = uint8(uint16(a)); // d = 0x34
uint8 e = uint8(bytes1(a)); // e = 0x12

Hexadecimal numbers can be assigned to any integer type if no truncation is needed.

uint8 a = 12; // no error
uint32 b = 1234; // no error
uint16 c = 0x123456; // error, as truncation required to 0x3456