Openclaw Skills for 2D game assets generates cohesive sprite sets, icons, stickers, and transparent cutouts that all share one visual style.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install game-assets-2d
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install game-assets-2d using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is 2D Game Assets?
Openclaw Skills for 2D game assets is a production workflow for generating cohesive sprite sets, die-cut stickers, item icons, parallax layers, and asset sheets. It uses a style LoRA to lock the look across the set, ControlNet Canny to preserve silhouette and structure when variations are needed, and background removal to deliver clean transparent cutouts.
Use it when you need a themed pack that feels like one artist made every piece: one potion kit, one chest family, one weapon lineup, or one icon library. The skill is optimized for game-ready consistency, so the outputs stay legible at small sizes and are ready to composite into engines, editors, and UI pipelines.
2D Game Assets Use Cases
- In Openclaw Skills, generate a complete pack of potions, chests, weapons, or consumables in one coherent style.
- Create sprite icons and die-cut stickers with clean cutouts for inventory or UI use.
- Preserve the silhouette of a source asset while re-skinning it into biome, rarity, or elemental variants.
- Build parallax pieces and themed environment layers that share one aesthetic.
- Batch-produce transparent PNGs and asset sheets for fast game integration.
How 2D Game Assets Works
- Resolve the live model catalog and confirm the base model, style LoRA, Canny ControlNet, and background remover schema before generating.
- Capture the LoRA trigger word and verify the live field shapes for
lora,controlNet, and background removal so the request matches the runtime. - If the asset must keep the same silhouette, preprocess the reference image with Canny edge detection to create a guide image.
- Run synchronous image generation with the base model plus the style LoRA, and add the Canny
controlNetblock when structure needs to stay fixed. - Batch a full set in one call with
numberResults, keeping the base model and LoRA constant while changing only the intended subject detail. - Remove backgrounds last with BiRefNet to produce transparent PNGs, then review the pack for style cohesion, silhouette fidelity, and cutout quality.
2D Game Assets Setup
Openclaw Skills does not require a local package install; setup is about selecting live models and validating schemas before you generate.
- Connect to the Runware-backed environment used by Openclaw Skills.
- Inspect the live catalog and confirm the base model, style LoRA, Canny ControlNet, and BiRefNet background remover are available.
- Verify the exact request schema for each model before you hardcode anything into a prompt or automation.
- Collect any structure reference image you plan to preserve and decide whether transparency is required.
runware-models
runware-run <base-model-air> --inspect
runware-run <style-lora-air> --inspect
runware-run <controlnet-canny-air> --inspect
runware-run <birefnet-air> --inspect
- Reuse the confirmed LoRA trigger word in every prompt for the set.
- For SDXL sprite and sticker work, set
clipSkip: 2when you want a simpler, bolder game-art look.
2D Game Assets Data Schema & Taxonomy
Openclaw Skills organizes each generation around a small set of reusable data layers.
| Layer | Key fields | Purpose |
|---|---|---|
| Request intent | asset_type, set_theme, style, count, canvas_size, transparency |
Defines what to make and how many outputs to generate. |
| Style control | base_model, lora[], trigger_word, lora.weight, clipSkip |
Locks the visual identity across the whole pack. |
| Structure control | reference_asset, controlNet.guideImage, startStep, endStep, weight, controlMode |
Preserves silhouette and proportions when making variants. |
| Cleanup | removeBackground, background_removal_model, alpha output |
Produces transparent PNG cutouts ready for game use. |
| Output group | rendered variants, cleaned PNGs, asset sheet | The final pack of game-ready 2D assets. |
Metadata taxonomy
style consistency: whether every output shares the same LoRA-driven look.silhouette fidelity: how closely Canny-guided variants match the source shape.bg_state: generation usually starts on a flat black or single-color background, then moves to transparent after cleanup.variation axis: the single intended change per image, such as biome, rarity, material, or elemental theme.production readiness: whether the asset stays legible at small in-game sizes and cuts out cleanly without halos.
2D Game Assets Advanced Features
- Schema-first generation: inspect live model schemas before every run so your LoRA and ControlNet fields stay valid.
- Style lock across a full pack: reuse one specialized LoRA and one trigger word so every asset reads as part of the same art direction.
- Structure-preserving variants: Canny edge guidance keeps shape and proportions stable while prompts change the finish.
- Batch rendering:
numberResultslets you generate a whole family of assets in a single call. - Clean delivery pipeline: background removal produces transparent PNGs that are ready for engines, UI systems, and editors.
- SDXL sprite tuning:
clipSkip: 2and a black-background prompt pattern help create simpler, bolder game-ready art. - Single-variable sheet discipline: keep the base model, LoRA, and guide constant while changing only one subject detail per output for maximum set coherence.
SKILL.md
Loading
Related Openclaw Skills
Image to 3D Asset
Openclaw Skills converts photos or prompts into production-ready GLB 3D assets with controllable topology, materials, and polygon budgets.
LLM Agent
Build OpenAI-compatible LLM agents on Runware that can reason, call your functions, and answer with grounded tool results.
Logos and Vectors
Openclaw Skills creates real SVG logos or converts raster images into editable vector art that stays crisp at any size.
Photoreal Stills
An advanced AI generation skill designed to defeat synthetic rendering artifacts by prompting for optical behaviors, real-world camera properties, and intentional imperfections.
Controlled Generation (ControlNet)
Preserve an image’s structure while changing its style, subject, or finish with ControlNet-driven generation.
Composite Scene
Composite Scene merges multiple real images into one coherent AI-generated scene by reconciling placement, scale, lighting, and perspective.
Featured*
