Struct Renderer 

pub struct Renderer<'a> { /* private fields */ }

A document’s renderer: the per-document caches (fonts, CMYK table, decoded image samples) that its page renders share. The pipeline renders every page at two scales; hold one Renderer per document across both.

Implementations

impl<'a> Renderer<'a>

pub fn new(meta: &'a PdfMeta) -> Renderer<'a>

A renderer decoding its JPEGs for docling’s bitmap hint (1.0 pixel per PDF unit — the model-input renders).

pub fn with_bitmap_hint(meta: &'a PdfMeta, bitmap_hint: f64) -> Renderer<'a>

A renderer with an explicit bitmap_target_pixels_per_unit: 0.0 decodes every image at full size (the page rasters a caller keeps — render_pages, the VLM input — where docling-parse is asked with 0.0 too).

pub fn render_scaled(&self, index: usize, scale: f64) -> Option<RgbImage>

Render page index at scale pixels per point the way docling-parse sizes its canvas (display orientation, /Rotate applied).

pub fn render(&self, index: usize, width: u32, height: u32) -> Option<RgbImage>

Render page index into exactly width × height display-frame pixels (the crop box stretched onto the canvas, as every renderer here does).

pub fn render_with_hint( &self, index: usize, width: u32, height: u32, bitmap_hint: f64, ) -> Option<RgbImage>

Renderer::render with an explicit bitmap_target_pixels_per_unit for this one render: 1.0 decodes JPEGs the way docling’s model inputs are decoded, 0.0 at full size — what a bitmap kept for OCR wants (the recognizer reads a scan’s 300 dpi, not a quarter of it blitted up; docling’s own OCR runs on its hint-1.0 decode, the deliberate deviation the pdfium raster of a scan already makes).