Comparisons

The Switch draws the world at 1600×900; here it can go up to 3840×2160, 5.8 times the pixels. The menus and text are drawn at the same size on top. Same moment, only the resolution changed (anti-aliasing off in both); the 1600×900 shot is stretched to 4K the way a 4K screen would show it.

The world goes on at the sides while the menus stay 16:9 in the middle, where they are on the Switch. Full-screen layers reach the edges; pictures made for 16:9 keep black bars. 21:9 at 4K height (5120×2160).

A bigger shadow map gives sharper, more detailed shadow edges, most visibly far from the camera. The bamboo forest with soft shadows off, so only the map size changes.

Meant to be used with soft shadows (PCSS): a sharp 4096 map alone shows every leaf as a hard dot; soft shadows on top keep the detail and soften it with distance, as in the next section.

Percentage-closer soft shadows behave more like real ones. Shown with the shadow map at 4096.

• Sharp at the contact: feet, poles, the base of a wall.
• Softer farther away: tree canopies become dappled shade, not hard dots.
• Less shimmer on distant shadows while the camera moves.

The shadow map can't see small, close detail. Contact shadows fill it in from the picture itself: a short ray from each point toward the sun finds what's in the way.

• Small objects cast their shadow: the taps over the schoolyard sink.
• Depth in foliage: hedge and flower-bed leaves shade each other.
• Thin lines under ledges: balconies, a window's awning, a wall's cap.

Applied to the 3D scene before the menus are drawn, so text stays sharp. TAA also stops thin things (fences, nets, wires) from shimmering while the camera moves. Yo-kai World, zoomed in 3×.

Without it the ground blurs a short way ahead. Anisotropic filtering keeps the road, pavement and walls sharp into the distance, even at a low angle. Yo-kai World, zoomed in 2×.

The world is drawn smaller than the resolution and rebuilt at full size from the frames before; the menus are drawn afterwards at full size. DLSS for the cleanest picture, FSR for the most frames. Each image: the whole native frame on top, the 1:1 crops underneath.