Rendering Pipeline logo

Rendering Pipeline

Community
jame581
rendering-pipeline

Use when choosing a renderer or planning a project's look and frame budget — renderer per platform, lighting strategy and light limits, anti-aliasing, 3D resolution scaling (FSR, MetalFX), SubViewports and render targets, shader stutter, graphics settings menus, and rendering stats

Overview

Publisherjame581
RepositoryGodotPrompter
Skill namerendering-pipeline
Stars
896
Forks
60
Bundled files
2
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 2 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by jame581 on GitHub. Read the source before you install it.

Installation

Install the Rendering Pipeline AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/jame581/GodotPrompter/tree/master/skills/rendering-pipeline
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/jame581/GodotPrompter.git /tmp/GodotPrompter
mkdir -p .claude/skills
cp -r /tmp/GodotPrompter/skills/rendering-pipeline .claude/skills/rendering-pipeline
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Rendering Pipeline in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Rendering Pipeline on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Rendering Pipeline is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Rendering Pipeline in Godot 4.3+

Project-level rendering decisions: which renderer runs where, how the lighting is built for it, how edges are smoothed, what resolution 3D renders at, and how to measure the result. All examples target Godot 4.3+ with no deprecated APIs. GDScript is shown first, then C#.

This skill picks the strategy; 3d-essentials holds the per-node recipes (materials, light properties, GI setup, fog, decals).

Related skills: 3d-essentials for materials, lights, environment and GI node setup, 2d-essentials for 2D lights and shadows, shader-basics for post-processing, compositor effects and the Shader Baker, godot-optimization for profiling CPU/GPU bottlenecks, web-export for the Compatibility-only browser target.


1. Pick the Renderer First

The renderer decides which lighting features exist at all, so choose it before building the look.

Rendererrendering_methodUse forMissing compared to Forward+
Forward+forward_plusDesktop and console, high-end visuals—
MobilemobilePhones, tablets, standalone XR, low-end desktopSDFGI, VoxelGI, SSAO/SSIL/SSR, volumetric fog, TAA, FSR
Compatibilitygl_compatibilityWeb (the only option there), very old GPUsEverything Mobile lacks (though SSAO arrives in 4.6), plus FXAA/SMAA and MetalFX; glow is a simplified version

Set Project Settings → Rendering → Renderer → Rendering Method. Platform overrides already exist: rendering/renderer/rendering_method.mobile defaults to mobile and rendering_method.web to gl_compatibility, so one project can run Forward+ on desktop and Mobile on phones — but then test the look on both, because effects silently disappear on the lighter renderer.

A settings menu must hide options the running renderer cannot honor. The later sections use this helper:

gdscript
func is_forward_plus() -> bool:
    # Godot 4.4+: prefer RenderingServer.get_current_rendering_method(), which also sees
    # the --rendering-method CLI flag and the automatic fallback on unsupported hardware.
    return ProjectSettings.get_setting_with_override("rendering/renderer/rendering_method") == "forward_plus"
csharp
public static bool IsForwardPlus()
{
    // Godot 4.4+: prefer RenderingServer.GetCurrentRenderingMethod(), which also sees
    // the --rendering-method CLI flag and the automatic fallback on unsupported hardware.
    return ProjectSettings.GetSettingWithOverride("rendering/renderer/rendering_method").AsString() == "forward_plus";
}

Short feature table: see the renderer comparison in 3d-essentials (references/renderer-comparison.md).


2. Lighting Strategy

Decide where light comes from before placing lights. Baked light is nearly free at runtime on every renderer; real-time GI is Forward+ only.

SceneForward+MobileCompatibility
Static level, fixed sunLightmapGI + real-time sun shadowsLightmapGILightmapGI
Day/night cycle, large outdoorSDFGISky ambient + DirectionalLight3DSky ambient + DirectionalLight3D
Small dynamic interiorsVoxelGILightmapGI + ReflectionProbesLightmapGI + ReflectionProbes
Procedural levels (nothing to bake)SDFGIAmbient + a few shadowed lightsAmbient + few unshadowed lights

Light budgets

RendererLimitSetting
Forward+512 clustered elements (omni + spot lights + decals + reflection probes) in viewrendering/limits/cluster_builder/max_clustered_elements
Mobile8 omni + 8 spot lights per mesh— split large meshes so lights don't compete
Compatibility8 omni + 8 spot lights per object; 32 renderable lights per type (omni, spot) per framerendering/limits/opengl/max_lights_per_object, max_renderable_lights

Rules that hold everywhere:

  • Shadows are the expensive part, not lights. Keep one shadowed DirectionalLight3D; turn shadow_enabled off on fill lights and small props.
  • Bake what doesn't move. Set static lights' light_bake_mode to Static so the lightmap carries them and they cost nothing at runtime. Baking needs a Forward+ or Mobile capable editor; the baked result plays on all three renderers.
  • Fade distant lights with distance_fade_enabled instead of leaving dozens active.

Light and GI node setup: 3d-essentials (Sections 3 and 5).


3. Anti-Aliasing

MethodCostSmoothsRenderersNotes
MSAA 2×/4×/8× (msaa_3d)Medium–highGeometry edgesAll three2× or 4× is usually the sweet spot; no help with shader/texture shimmer
FXAA (screen_space_aa)LowEverything, blurryForward+, MobileCheapest whole-image option
SMAA (Godot 4.5+)Low–mediumEverything, sharper than FXAAForward+, MobileTune rendering/anti_aliasing/quality/smaa_edge_detection_threshold
TAA (use_taa)MediumEverything incl. specular shimmerForward+ onlyGhosting on particles and skinned meshes
FSR 2 at scale 1.0MediumLike TAA, sharperForward+ onlySet scaling_3d_mode to FSR 2 and scaling_3d_scale = 1.0

The project settings (rendering/anti_aliasing/quality/*) are read once at startup. To change AA from a settings menu, set the root viewport:

gdscript
func apply_antialiasing(msaa: Viewport.MSAA, use_fxaa: bool, use_taa: bool) -> void:
    var vp := get_viewport()
    vp.msaa_3d = msaa
    vp.screen_space_aa = Viewport.SCREEN_SPACE_AA_FXAA if use_fxaa else Viewport.SCREEN_SPACE_AA_DISABLED
    vp.use_taa = use_taa and is_forward_plus()
csharp
public void ApplyAntialiasing(Viewport.Msaa msaa, bool useFxaa, bool useTaa)
{
    Viewport vp = GetViewport();
    vp.Msaa3D = msaa;
    vp.ScreenSpaceAA = useFxaa ? Viewport.ScreenSpaceAAEnum.Fxaa : Viewport.ScreenSpaceAAEnum.Disabled;
    vp.UseTaa = useTaa && IsForwardPlus();
}

Setting an unsupported mode does not crash; Godot ignores it and logs a warning (an editor error on 4.3–4.5). Gate the options in the UI instead. 2D has its own msaa_2d, which smooths 2D geometry edges such as lines and polygons.


4. 3D Resolution Scaling

Render 3D below native resolution and upscale it — the biggest single GPU saving on fill-rate-bound scenes. 2D and UI stay at native resolution.

scaling_3d_modeRenderersNotes
BilinearAllAlso the only mode that supersamples (scale > 1.0)
FSR 1Forward+Spatial upscaler; rendering/scaling_3d/fsr_sharpness (default 0.2, lower = sharper)
FSR 2Forward+Temporal; replaces TAA; best quality per pixel
MetalFX Spatial / Temporal (Godot 4.4+)Apple Metal: spatial on Forward+/Mobile, temporal on Forward+Falls back to FSR when unavailable
Nearest (Godot 4.7+)AllCrisp pixel look; use integer divisors (0.5, 0.333, 0.25)

Offer presets rather than a raw slider. These are AMD's FSR 1 scales, which work as well with FSR 2: Ultra Quality 0.77, Quality 0.67, Balanced 0.59, Performance 0.5.

gdscript
const FSR_PRESETS := {"ultra_quality": 0.77, "quality": 0.67, "balanced": 0.59, "performance": 0.5}


func apply_upscaling(preset: String) -> void:
    var vp := get_viewport()
    if preset == "native":
        vp.scaling_3d_mode = Viewport.SCALING_3D_MODE_BILINEAR
        vp.scaling_3d_scale = 1.0
        return
    vp.scaling_3d_mode = Viewport.SCALING_3D_MODE_FSR2 if is_forward_plus() else Viewport.SCALING_3D_MODE_BILINEAR
    vp.scaling_3d_scale = FSR_PRESETS.get(preset, 1.0)
csharp
private static readonly System.Collections.Generic.Dictionary<string, float> FsrPresets = new()
{
    { "ultra_quality", 0.77f }, { "quality", 0.67f }, { "balanced", 0.59f }, { "performance", 0.5f },
};

public void ApplyUpscaling(string preset)
{
    Viewport vp = GetViewport();
    if (preset == "native")
    {
        vp.Scaling3DMode = Viewport.Scaling3DModeEnum.Bilinear;
        vp.Scaling3DScale = 1.0f;
        return;
    }
    vp.Scaling3DMode = IsForwardPlus() ? Viewport.Scaling3DModeEnum.Fsr2 : Viewport.Scaling3DModeEnum.Bilinear;
    vp.Scaling3DScale = FsrPresets.GetValueOrDefault(preset, 1.0f);
}

Lower scales blur textures; on Forward+ and Mobile, Godot lowers the mipmap bias automatically by log2(scale) to compensate (plus -0.5 with TAA).


5. SubViewports and Render Targets

Each active SubViewport is another full render. Set render_target_update_mode: UPDATE_WHEN_VISIBLE (live views), UPDATE_ONCE (snapshots), UPDATE_DISABLED (pause), UPDATE_ALWAYS.

See references/subviewports.md for UI display, per-viewport settings, and icon capture.


6. Shader Compilation Stutter

The first time a material, light combination, or effect is drawn, the GPU pipeline is compiled, which can hitch a frame.

  • Godot 4.4+ compiles pipelines in the background with ubershaders on Forward+ and Mobile, which removes most of this hitching. Projects on 4.3 benefit from upgrading.
  • The pipeline cache (rendering/rendering_device/pipeline_cache/enable, on by default) makes the second launch faster than the first.
  • The Shader Baker (Godot 4.5+, export preset option) shortens the first launch on D3D12 and Metal — see shader-basics and export-pipeline.
  • On Compatibility, and for effects that still hitch, show each material once behind a loading screen (spawn the effect off-camera or at zero scale for a frame).

7. Measuring

Read rendering stats from Performance and switch Viewport.debug_draw to see where the cost is.

gdscript
func _process(_delta: float) -> void:
    $Stats.text = "draw calls %d | primitives %d | VRAM %.0f MB" % [
        Performance.get_monitor(Performance.RENDER_TOTAL_DRAW_CALLS_IN_FRAME),
        Performance.get_monitor(Performance.RENDER_TOTAL_PRIMITIVES_IN_FRAME),
        Performance.get_monitor(Performance.RENDER_VIDEO_MEM_USED) / 1048576.0,
    ]


func _unhandled_input(event: InputEvent) -> void:
    if event.is_action_pressed("debug_overdraw"):
        var vp := get_viewport()
        vp.debug_draw = Viewport.DEBUG_DRAW_DISABLED if vp.debug_draw == Viewport.DEBUG_DRAW_OVERDRAW else Viewport.DEBUG_DRAW_OVERDRAW
csharp
public override void _Process(double delta)
{
    GetNode<Label>("Stats").Text = string.Format("draw calls {0} | primitives {1} | VRAM {2:F0} MB",
        Performance.GetMonitor(Performance.Monitor.RenderTotalDrawCallsInFrame),
        Performance.GetMonitor(Performance.Monitor.RenderTotalPrimitivesInFrame),
        Performance.GetMonitor(Performance.Monitor.RenderVideoMemUsed) / 1048576.0);
}

public override void _UnhandledInput(InputEvent @event)
{
    if (@event.IsActionPressed("debug_overdraw"))
    {
        Viewport vp = GetViewport();
        vp.DebugDraw = vp.DebugDraw == Viewport.DebugDrawEnum.Overdraw ? Viewport.DebugDrawEnum.Disabled : Viewport.DebugDrawEnum.Overdraw;
    }
}

Useful debug_draw modes: OVERDRAW (transparency and particle cost), UNSHADED (is it lighting or geometry?), LIGHTING, WIREFRAME (call RenderingServer.set_debug_generate_wireframes(true) before the meshes load). The editor's Debugger → Visual Profiler splits GPU time per pass. Profiling method: godot-optimization.


8. Graphics Settings Menu

Players expect presets (Low / Medium / High / Ultra) plus individual toggles, saved between sessions. Apply runtime settings to the root viewport, environment toggles to the active Environment, and store the choice in a ConfigFile under user://.

Full preset table, apply code, and persistence in GDScript and C#: references/graphics-settings.md.


9. Common Pitfalls

SymptomCauseFix
Effects vanish on phone or web buildBuilt the look on Forward+, shipped on Mobile/CompatibilityPreview with the target renderer; plan lighting per Section 2
TAA or FSR option does nothingNot Forward+ (warning in output)Gate the option on the running renderer (Section 1 helper)
Changing the AA project setting in game has no effectProject settings are read at startupSet the root Viewport properties
Lights flicker on and off on MobileMore than 8 omni or 8 spot lights touch one meshSplit the mesh, bake static lights, or fade distant lights
Minimap halves the frame rateSubViewport renders full quality every frameLower its size, disable its shadows/effects, update every few frames
SubViewport looks aliased while the game doesn'tSubViewports don't inherit root AA/scalingSet msaa_3d / scaling_3d_* on the SubViewport
Pixel-art 3D looks smearedBilinear upscalingSubViewportContainer.stretch_shrink with nearest texture filter, or Nearest scaling (4.7+)
Hitch the first time an effect playsPipeline compilationUpgrade to 4.4+, pre-warm behind a loading screen

10. Checklist

  • Renderer chosen per target platform; .mobile / .web overrides reviewed
  • The look is checked on the lightest renderer that ships
  • Lighting plan written down: what is baked, what is real-time, which lights cast shadows
  • Light counts within the renderer's budget (512 clustered / 8+8 per mesh on Mobile and Compatibility)
  • AA method matches the renderer; menu hides unsupported options
  • 3D resolution scaling offered as presets (FSR 2 on Forward+, bilinear elsewhere)
  • Every SubViewport has a deliberate update mode, size, and its own AA settings
  • Draw calls, primitives, and VRAM checked with Performance monitors on target hardware
  • Graphics settings saved to user:// and re-applied on startup

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Rendering Pipeline AI skill do?

Use when choosing a renderer or planning a project's look and frame budget — renderer per platform, lighting strategy and light limits, anti-aliasing, 3D resolution scaling (FSR, MetalFX), SubViewports and render targets, shader stutter, graphics settings menus, and rendering stats

Why use Rendering Pipeline on TypingMind?

Because you install it once and use it with any model. Rendering Pipeline is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Rendering Pipeline in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/jame581/GodotPrompter/tree/master/skills/rendering-pipeline. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Rendering Pipeline?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Rendering Pipeline?

As many as you like. As long as a model supports skills, you can use Rendering Pipeline with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Rendering Pipeline AI skill free?

Yes. It is published on GitHub by jame581 under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇