Struct dssim::Dssim

pub struct Dssim { /* fields omitted */ }

Configuration for the comparison

Methods

impl Dssim

fn new() -> Dssim

Create new context for comparisons

fn set_scales(&mut self, scales: &[f64])

Set how many scales will be used, and weights of each scale

fn set_save_ssim_maps(&mut self, num_scales: u8)

Set how many scales will be kept for saving

fn create_image<InBitmap, OutBitmap>(
    &self,
    src_img: &InBitmap
) -> Option<DssimImage<f32>> where
    InBitmap: ToLABBitmap + Send + Sync + Downsample<Output = OutBitmap>,
    OutBitmap: ToLABBitmap + Send + Sync + Downsample<Output = OutBitmap>, 

The input image is defined using the imgref crate, and the pixel type can be:

• ImgVec<RGBAPLU> — RGBA premultiplied alpha, float scaled to 0..1
• ImgVec<RGBLU> — RGBA float scaled to 0..1
• ImgVec<f32> — linear light grayscale, float scaled to 0..1

And there's ToRGBAPLU (.to_rgbaplu()) helper to convert the input pixels from [RGBA<u8>], [RGBA<u16>], [RGB<u8>], or RGB<u16>. See main.rs for example how it's done.

You can implement ToLABBitmap and Downsample traits on your own image type.

fn compare(
    &mut self,
    original_image: &DssimImage<f32>,
    modified_image: DssimImage<f32>
) -> (Val, Vec<SsimMap>)

Compare original with another image. See create_image

The SsimMaps are returned only if you've enabled them first.

Val is a fancy wrapper for f64