Re: [fxtf-drafts] [filter-effects-1] Is WPT test for accumulate correct? (#371)

> From the spec, it appears that accumulation is just equivalent to replace (
> `Given two filter values Va and Vb, returns the filter value, Vb.`

Yes, this appears to be wrong. I'm not sure where it comes from but perhaps it was just put there until we worked out how to spec this.

> Neither the blur calculation (which looks like explicit addition) or the brightness behavior (which looks like some sort of interpolation?) appears correct if I am reading the spec correctly.

Assuming the above definition is wrong, I think the blur calculation is expected (component-wise addition as typical for accumulation).

For the brightness component, it's based on `brightness(1)` being the zero value. (Or as the spec puts it, "The initial value for interpolation is 1"). Since accumulation is supposed to accumulate the delta from the zero value, the operation here is to take the delta of `-0.7`, add to `-0.9` (i.e. `-1.6`) then apply that delta to the zero value (i.e. `-0.6` which then gets clamped to `0`). I believe that's consistent with what we do for `scale()` for example and I think it's the intuitive result here (building on `brightness(0.1)` will quickly get you into complete darkness).

GitHub Notification of comment by birtles
Please view or discuss this issue at using your GitHub account

Received on Thursday, 12 September 2019 00:31:13 UTC