C, 212 bytes

d;main(){char*t="##3#b#b3#bbb3#b#b##\r#3b1#+3@12b3@1b-3@1_b3b1#,#\r7#_##+51rR04/1b#61rR0,8#2##\r7?#2#+9#`A#9=###9#^?#,8A#_#\r#+:#%b#:=#b#:#%b#,#",p[9];for(gets(p);d=*t++;putchar(d-3))d=d<51?d:(p[d-51]-53)[t+=4];}

A readable version:

d;
main()
{
 char *t = "##3#b#b3#bbb3#b#b##\r"
 "#3b1#+3@12b3@1b-3@1_b3b1#,#\r"
 "7#_##+51rR04/1b#61rR0,8#2##\r"
 "7?#2#+9#`A#9=###9#^?#,8A#_#\r"
 "#+:#%b#:=#b#:#%b#,#",
 p[9]; // 9 bytes is just enough for the input string of length 8
 for (gets(p); d = *t++; putchar(d-3))
 d = d < 51 ? d : (p[d - 51] - 53)[t += 4];
}

I took the idea from the answer by Reto Koradi. There were several fun improvements I did, which may warrant posting a separate answer:

• Converted from function to program (+10)
• Moved newlines into the control string (-7)
• Added 3 to all character codes to have fewer escaped chars like \" (-3)
• Reading from the string with autoincrement; also replaced t[i++] by *t++ (-4)
• Replaced while by for; removed {} (-4)
• Simplified loop termination: reading until 0円 (-9)
• Transformed t[...],t+=4 to (...)[t+=4] to eliminate the comma operator (-1)

Why all that trouble? To share my favorite one, snow ghost:

 _
 /_\
\(. .)/
 ( )
 (___)

C, 212 bytes

d;main(){char*t="##3#b#b3#bbb3#b#b##\r#3b1#+3@12b3@1b-3@1_b3b1#,#\r7#_##+51rR04/1b#61rR0,8#2##\r7?#2#+9#`A#9=###9#^?#,8A#_#\r#+:#%b#:=#b#:#%b#,#",p[9];for(gets(p);d=*t++;putchar(d-3))d=d<51?d:(p[d-51]-53)[t+=4];}

A readable version:

d;
main()
{
 char *t = "##3#b#b3#bbb3#b#b##\r"
 "#3b1#+3@12b3@1b-3@1_b3b1#,#\r"
 "7#_##+51rR04/1b#61rR0,8#2##\r"
 "7?#2#+9#`A#9=###9#^?#,8A#_#\r"
 "#+:#%b#:=#b#:#%b#,#",
 p[9]; // 9 bytes is just enough for the input string of length 8
 for (gets(p); d = *t++; putchar(d-3))
 d = d < 51 ? d : (p[d - 51] - 53)[t += 4];
}

I took the idea from the answer by Reto Koradi. There were several fun improvements I did, which may warrant posting a separate answer:

• Converted from function to program (+10)
• Moved newlines into the control string (-7)
• Added 3 to all character codes to have fewer escaped chars like \" (-3)
• Reading from the string with autoincrement; also replaced t[i++] by *t++ (-4)
• Replaced while by for; removed {} (-4)
• Simplified loop termination: reading until 0円 (-9)
• Transformed t[...],t+=4 to (...)[t+=4] to eliminate the comma operator (-1)

Why all that trouble? To share my favorite one, snow ghost:

 _
 /_\
\(. .)/
 ( )
 (___)

C, 212 bytes

d;main(){char*t="##3#b#b3#bbb3#b#b##\r#3b1#+3@12b3@1b-3@1_b3b1#,#\r7#_##+51rR04/1b#61rR0,8#2##\r7?#2#+9#`A#9=###9#^?#,8A#_#\r#+:#%b#:=#b#:#%b#,#",p[9];for(gets(p);d=*t++;putchar(d-3))d=d<51?d:(p[d-51]-53)[t+=4];}

I took the idea from the answer by Reto Koradi. There were several fun improvements I did, which may warrant posting a separate answer:

• Converted from function to program (+10)
• Moved newlines into the control string (-7)
• Added 3 to all character codes to have fewer escaped chars like \" (-3)
• Reading from the string with autoincrement; also replaced t[i++] by *t++ (-4)
• Replaced while by for; removed {} (-4)
• Simplified loop termination: reading until 0円 (-9)
• Transformed t[...],t+=4 to (...)[t+=4] to eliminate the comma operator (-1)

Why all that trouble? To share my favorite one, snow ghost:

 _
 /_\
\(. .)/
 ( )
 (___)

C, 212 bytes

d;main(){char*t="##3#b#b3#bbb3#b#b##\r#3b1#+3@12b3@1b-3@1_b3b1#,#\r7#_##+51rR04/1b#61rR0,8#2##\r7?#2#+9#`A#9=###9#^?#,8A#_#\r#+:#%b#:=#b#:#%b#,#",p[9];for(gets(p);d=*t++;putchar(d-3))d=d<51?d:(p[d-51]-53)[t+=4];}

A readable version:

d;
main()
{
 char *t = "##3#b#b3#bbb3#b#b##\r"
 "#3b1#+3@12b3@1b-3@1_b3b1#,#\r"
 "7#_##+51rR04/1b#61rR0,8#2##\r"
 "7?#2#+9#`A#9=###9#^?#,8A#_#\r"
 "#+:#%b#:=#b#:#%b#,#",
 p[9]; // 9 bytes is just enough for the input string of length 8
 for (gets(p); d = *t++; putchar(d-3))
 d = d < 51 ? d : (p[d - 51] - 53)[t += 4];
}

I took the idea from the answer by Reto Koradi. There were several fun improvements I did, which may warrant posting a separate answer:

• Converted from function to program (+10)
• Moved newlines into the control string (-7)
• Added 3 to all character codes to have fewer escaped chars like \" (-3)
• Reading from the string with autoincrement; also replaced t[i++] by *t++ (-4)
• Replaced while by for; removed {} (-4)
• Simplified loop termination: reading until 0円 (-9)
• Transformed t[...],t+=4 to (...)[t+=4] to eliminate the comma operator (-1)

Why all that trouble? To share my favorite one, snow ghost:

 _
 /_\
\(. .)/
 ( )
 (___)