Enum wgpu_test::image::ComparisonType

pub enum ComparisonType {
    Mean(f32),
    Percentile {
        percentile: f32,
        threshold: f32,
    },
}

The FLIP library generates a per-pixel error map where 0.0 represents “no error” and 1.0 represents “maximum error” between the images. This is then put into a weighted-histogram, which we query to determine if the errors between the test and reference image is high enough to count as “different”.

Error thresholds will be different for every test, but good initial values to look at are in the [0.01, 0.1] range. The larger the area that might have inherent variance, the larger this base value is. Using a high percentile comparison (e.g. 95% or 99%) is good for images that are likely to have a lot of error in a small area when they fail.

Variants

Mean(f32)

If the mean error is greater than the given value, the test will fail.

Percentile

Fields

percentile: f32

threshold: f32

If the given percentile is greater than the given value, the test will fail.

The percentile is given in the range [0, 1].

Trait Implementations

impl Clone for ComparisonType

fn clone(&self) -> ComparisonType

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ComparisonType

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Copy for ComparisonType