GLSL: How do I cast a float to an int?

Question

In a GLSL fragment shader I am trying to cast a float into an int. The compiler raises an error:

ERROR: 0:60: '=' :  cannot convert from 'mediump float' to 'highp int'

I tried raising the precision of the int:

mediump float indexf = floor(2.0 * mixer);
highp int index = indexf;

but to no avail.

How do I cast the int properly?

For future reference, you will get better answers if you ask general programming questions like this one at Stack Overflow. stackoverflow.com - depending on community feedback this question might be migrated there anyway. — Ricket, Feb 17 '11 at 18:48
For reinterpret casts, there are now functions such as floatBitsToInt. — jozxyqk, Oct 10 '14 at 9:35

Ricket · Accepted Answer · 2011-02-17 18:47:26Z

42

Try this:

highp int index = int(indexf);

I found it here.

answered Feb 17 '11 at 18:47

Ricket

14.5k5 gold badges62 silver badges82 bronze badges

\$\begingroup\$ Also note that casting a float to an int automatically floors it (at least in any implementation I've ever seen) so your call to floor should be unnecessary. \$\endgroup\$ – Ricket Feb 17 '11 at 20:03
8

\$\begingroup\$ In most languages the rounding mode is truncate / round-to-zero, which is equivalent to floor for positive numbers but not negative numbers. I don't remember if this applies to GLSL, but I'd be surprised if it doesn't. \$\endgroup\$ – user744 Feb 17 '11 at 20:19
\$\begingroup\$ @Joe That's an interesting fact that I never thought about, and I just confirmed it with a Java test case. \$\endgroup\$ – Ricket Feb 18 '11 at 3:05
4

\$\begingroup\$ Is highp important here? (On an Integer) \$\endgroup\$ – Léon Pelletier Jul 5 '15 at 6:11
1

\$\begingroup\$ How cna I convert this to const so that I can sample an array? \$\endgroup\$ – lesolorzanov Feb 7 '19 at 20:34

| show 3 more comments

1 Answer 1