Resolume Forum

Ah, I see what happened. I had

Int In => Power => 1/x.

The result of an Integer negative exponent is rounded to zero as the result of Power is cast to Int. Then the 1/x gives inf.

Working as designed.

David

in Wire, the Power node returns inf when given a negative exponent.

Negative exponents are a thing. 2 ^ (-1) = 1/2

please fix

I was having trouble, running into a bug where subsequent passes could not read from the target of pass 0. Making target 0 "PERSISTENT": true fixed that problem. more investigations to come...

$WIDTH is always (for every pass) the width of the input texture so we have to use $WIDTH/2 and $WIDTH/4 to do the subsequent downsampling.

This would be great to get fixed.

Because of https://resolume.com/forum/viewtopic.php?p=103010 I do not have a workaround. This second problem makes it impossible to use more than one persistent buffer in multiple passes. I found a nice trick for saving high precision floats in a four component ...

when buffers are marked PERSISTENT, they no longer respect the FLOAT designation, that is, they become 8-bit per component buffers even when FLOAT: true is specified.

In this patch normalized fragment coordinates are stored in a buffer. If they are stored at 32-bit floats, there is sufficient ...

has anyone succeeded in reading multiple buffers in different passes? I don't believe it works.

multipass buffer access seems broken

after initializing contents of two buffers, first pass can read first buffer, but second pass gets pixels from the first buffer, not the second as expected.


/*{
"CREDIT": "dtristram",
"DESCRIPTION": "illustrates bug in multipass support, note image 1 ...

I think what is happening is I am rerunning my shader in the same Wire session, and FRAMEINDEX is just starting off at some big number.

I'm having a little trouble using FRAMEINDEX in an isf shader.

If I test the value like
if (FRAMEINDEX < 10) { ...
then I do not enter this code unless I test against a huge number like 10000000.

but code like this gives expected result though:

vec4 color = vec4( fract( FRAMEINDEX / 1000 ...