Image

The file is decoded once and cached as a Metal texture (the cache is keyed by path, so many nodes can share one file at no extra cost). Each frame the GPU draws that texture onto a quad in the node's offscreen output texture, applying all transforms in Grafit space — origin (0,0) at the viewport center, shorter viewport edge mapped to [-1, +1], positive Rotation clockwise. The transform order is: size the quad (per the mode below) → Scale → Rotation → translate by X/Y.

Sizing modes:

In the fragment stage, Flop mirrors the image horizontally (left ↔ right) and Flip mirrors it vertically (top ↔ bottom); both can combine. Opacity multiplies the alpha, the quad's outer edge gets an antialiased fade driven by the global AA multiplier, and the final color is premultiplied (rgb *= a) as required for generator textures. Transparent PNGs and other alpha formats keep their transparency.

Choosing the file: the Path port has no default. With nothing connected, use the Browse… button on the Path row of the inspector — it opens a file picker, shows the chosen file's pixel dimensions, and stores a security-scoped bookmark so the sandboxed app can still open the file after the project is saved and reloaded. Alternatively connect any string (e.g. from Constant or String Combine) to Path for dynamic switching; a new path swaps the texture on the fly.

The Width/Height outputs report the quad's size in Grafit units for the current mode — Fit mode returns the viewport-fit size, True Size mode the pixel-converted native size. Scale is deliberately excluded from these outputs so that wiring them into a Canvas node's Width/Height doesn't double-apply the scale. With no file loaded, both report 2.0. When Enable is false, the output texture is cleared to transparent.

Interactions: in Fit mode a landscape image on a landscape viewport typically covers it fully; a portrait image on a landscape viewport touches top and bottom and leaves transparent pillars — position the gaps with X or cover them with a layer below. Opacity combines with the source's own alpha channel rather than replacing it.

• Background plate — Image (Fit mode) on Canvas layer 1, with generative layers (Particle System, L-System, Magic Storm) composited above.
• Logo bug / watermark — True Size on for pixel-crisp rendering, positioned into a corner with X/Y, Opacity around 0.5.
• Slideshow — Counter or Timer → Switch selecting between path strings → Path for timed or manually punched image changes.
• Audio-reactive artwork — Audio FFT → Remap → Scale for cover art that breathes with the music; add LFO → Rotation for slow drift.
• Texture source for filters — Image output → Invert Image, Gaussian Blur, Image Adjustment, or Image Transform to build a processing chain before the Canvas.
• Photo as mask — feed a high-contrast image into a Canvas node's Mask input (luminance mode) to reveal another layer through the picture's bright areas.
• Crossfade between two images — two Image nodes on separate Canvas layers, with one LFO (or a fader value) driving one Opacity directly and the other through Math (1 − x).

• A generator alone shows nothing: connect Image to a Canvas node. Blend modes, masking, and 3D placement happen there.
• Wire this node's Width/Height outputs into the Canvas node's Width/Height to make the canvas quad match the picture exactly — they intentionally exclude Scale, so no double-scaling.
• After save/load, disk access relies on the stored security-scoped bookmark; if you move or rename the file on disk, re-pick it with Browse….
• Dynamic Path switching is instant for cached files, but the first load of a new file decodes on demand — preload big files by switching to them once off-air.
• Very large sources (multi-thousand-pixel photos) consume GPU memory at full resolution; downscale assets to roughly the output size for lean projects.
• The edge antialiasing on the quad border follows the global AA multiplier; with AA off, edges (especially rotated ones) will alias.
• Related nodes: Video Player (moving footage), Camera Input (live sources), Image Adjustment / Image Transform / Invert Image (processing), Text as Image (typography).