Why 3D Games Look Bad in Godot

Summary

The video tests whether Godot itself makes 3D games look bad by improving a simple village scene without replacing its models. The presenter argues that default engine settings are only a starting point: lighting, materials, atmosphere, vegetation, and careful optimization can make the same assets look far more convincing.

Main visual improvements

• Lighting and tone mapping: The presenter changes the sun’s direction and warmth to give buildings stronger shape and longer shadows. A blue sky adds ambient light to shadowed areas, while AgX tone mapping softens bright highlights.
• Contact and bounced light: SSAO adds darkening where objects meet, helping barrels and stones feel grounded. SDFGI approximates bounced light, including subtle color reflected from nearby grass. The video cautions that screen-space effects can shift as the camera moves, so they should be tested while moving through the scene.
• Atmospheric depth: Fog reduces the contrast of distant trees, separating the background from nearby objects. Too much ground haze can obscure paths, so atmosphere should not compromise gameplay readability.
• Vegetation: A shader adjusts leaf normals so foliage reads more like a unified tree crown than flat cards, and adds slight wind movement. Grass uses vertex colors to keep roots relatively still while bending its tips, with noise adding variation.
• Ground and density: A masked ground material blends green shades with worn dirt around paths and doors. The presenter adds about 20,000 grass and flower instances using Godot’s MultiMesh system instead of creating an individual scene node for each. Visibility range and shadows still need limits.
• Character and finishing touches: A character-only fill light improves visibility in shadows, while lanterns and windows add localized warm light. Anti-aliasing, a subtle vignette, faint noise, and dust provide minor finishing effects. A narrower field of view is considered but left off for the main comparison because it changes how much of the world players can see.

Performance and optimization

The completed version initially ran at roughly 50 FPS on an RTX 5080. The presenter identifies the sun-and-shadow setup as the largest performance cost in testing, but recommends reducing shadow quality rather than removing the sun’s directional lighting altogether. Other possible adjustments include lowering anti-aliasing samples and shortening the grass draw distance.

With selected settings reduced, the demo reached the monitor’s 75 FPS limit. The video emphasizes that this is not a maximum-frame-rate result, but an example of how tuning expensive parts of an effect can preserve its most visible benefits.

Key takeaway

The final comparison removes the sun setup while keeping the other improvements. The scene retains its fog, vegetation, and shaders, but loses much of its lighting direction and contrast. The presenter’s conclusion is that effects work together and should be chosen to solve specific visual problems—not simply enabled at maximum settings.

The experiment demonstrates what can be achieved in this scene, but does not prove that every Godot game will look good automatically.

Tutorial and analysis covered

1. Adjust directional sunlight, sky lighting, and tone mapping.
2. Add ambient occlusion, bounced light, and fog while checking visual behavior in motion.
3. Improve foliage and grass with shaders and material masks.
4. Populate a scene efficiently with MultiMesh and control rendering costs.
5. Profile visual settings and optimize selectively rather than disabling whole effects.

Main speaker and source

A single narrator and developer from the Coding Quests channel presents the scene-building demonstration and performance analysis.