Advanced generative-art tooling built on the HashLips Art Engine ecosystem.
Orion Art Engine is a Node.js CLI for generating layered PNG and GIF collections and their metadata. It extends the familiar HashLips workflow with tools for projects that need more control over trait compatibility, rarity, metadata, generation stages, and collection recovery.
This repository is a fork of the original HashLips Art Engine. The goal of Orion Art Engine is to preserve the straightforward layer-based workflow while expanding it for more complex production requirements.
Development history: part of this repository's Git history was reconstructed after the original Git metadata was lost. Reconstruction commits are grouped by feature and use the dates on which the recovery was performed. The repository remains a fork so the original HashLips project and its history are clearly attributed.
The original HashLips Art Engine provides a simple way to generate layered NFT collections. During real collection development, I needed more granular control over how traits interact, how rarity is defined, how metadata is produced, and how large generation runs can be inspected or resumed.
Orion Art Engine grew around those requirements.
Rather than treating generation as a single random image-composition step, the engine adds tooling around the entire collection-generation workflow: trait constraints, deterministic counts, metadata inspection, staged rendering, animated assets, recovery, and export utilities.
- Compatibility rules for preventing invalid trait combinations.
- Forced combinations for traits that must appear together.
- Conditional parent relationships between traits.
- Subtrait controls with independent blend, opacity, and Z-index options.
- Z-index overrides for more flexible image composition.
- Optional exclusion of selected layers or traits from metadata.
- Traditional filename-based rarity weights.
- Named rarity weights.
- Exact trait counts when a trait must appear a specific number of times instead of relying only on RNG.
- Rarity analysis generated from collection metadata.
- Metadata generation separated from image rendering.
- Review rarity and compatibility results before rendering the full image collection.
- Regenerate images from cached generation data without rebuilding metadata.
- Resume larger collections in stages.
- Optional duplicate generation.
- Configurable DPI and per-layer-configuration collection settings.
- Generate PNG collections.
- Generate animated GIF collections.
- Mix PNG and GIF traits in the same composition.
- Configure animation frame count, delay, quality, and repetition.
- Metadata output for Ethereum, Solana, and SEI workflows.
- Metadata generation for pre-rendered 1-of-1 collections.
- OpenSea Drop CSV output.
- Extra collection attributes and randomized stat blocks.
- Metadata cleanup, attribute removal/renaming, sorting, and migration utilities.
- IPFS upload workflow through Thirdweb storage.
- Cache generation data for later image regeneration.
- Backup and restore collection state.
- Archive previous collections and reset the engine for a new generation run.
- JavaScript / Node.js
- Canvas-based image composition
- GIF processing
- JSON metadata generation
- CLI workflows
- IPFS / Thirdweb integration
Use Node.js 20 LTS.
Native image rendering depends on the canvas package, so Linux users may also need the system dependencies required by node-canvas.
npm install- Replace the example folders in
layerswith your own layer folders. - Configure collection metadata and
layerConfigurationsinsrc/config.js. - List layers from back to front in
layersOrder. - Run the generator:
npm run generate- Follow the CLI prompts to configure compatibility rules when needed.
- Review the generated metadata and rarity breakdown.
- Approve image generation once the collection rules look correct.
Metadata is generated before images. This makes it possible to catch incorrect rarity or compatibility rules before spending time rendering an entire collection.
The latest generation data is cached, allowing images to be regenerated without recreating metadata:
npm run generate_photos{
growEditionSizeTo: 100,
namePrefix: collectionName,
description,
layersOrder: [
{ name: "Background" },
{ name: "Body" },
{ name: "Outfit" },
{ name: "Face" },
{
name: "Action",
options: {
conditionalOn: ["Body", "Outfit"],
},
},
],
}The CLI includes a compatibility wizard for defining relationships between traits without manually editing generation logic.
Two important rules are supported:
- Incompatibility: a selected trait cannot appear with one or more other traits.
- Forced combination: a selected trait must generate together with another specific trait.
Compatibility configuration can also be edited directly through compatibility/compatibility.json for advanced use cases.
In addition to probability-based rarity, traits can be configured around exact counts. This is useful when a collection requires a trait to appear a known number of times rather than approximately reaching that distribution through random generation.
GIF and PNG traits can be used together. Animation settings are configured in src/config.js:
const gif = {
numberOfFrames: 60,
repeat: 0,
quality: 100,
delay: 40,
};Temporary frames generated during processing can be cleaned after the generation run.
For collections where artwork is already rendered and only metadata is required, enable the 1-of-1 workflow and place the assets in:
./layers/oneOfOne
The engine can then generate metadata without recomposing the artwork.
The engine includes an upload workflow using Thirdweb storage. Configure your Thirdweb secret key in .env and use:
npm run upload images
npm run upload metadataAfter uploading images, update the collection baseUri with the returned CID before generating/uploading the final metadata.
The project includes utilities for common collection maintenance tasks, including:
cleanMetadataremoveAttributesrenameAttributesgenerateOldDnarecreateAndSortMetadatararityFromMetadatacreateOpenseaCSV
Orion Art Engine originated from my work in the HashLips/Web3 ecosystem and from the requirements I encountered while working with generative NFT collections.
The original HashLips Art Engine remains the foundation of this fork and is credited accordingly. Orion Art Engine represents the additional tooling and workflow experimentation developed around that foundation.
MIT. See LICENSE for details.