Node Library
MOD ships 130+ processing nodes (and growing) spanning image, texture, 3D/photogrammetry, video, audio, AI, and architecture/engineering/construction (AEC) workloads. Drag them onto the Workflow Builder canvas, or discover and run them over the API.
How a node works
Every node is a containerized capability with:
- Input / output / configuration schemas — what it consumes, produces, and the knobs it exposes (visible in the node's panel and via
GET /node_definitions/{id}). - A credit cost (
cost_in_credits) — typically 0–3 credits per run depending on the node. - Hardware variants (
gpu_variants) — many nodes ship CUDA / ROCm / Intel / NPU / CPU builds, and MOD routes the run to matching hardware; CPU-only nodes run anywhere. - Batch mode — most processing nodes accept a folder and fan out across its files.
# Discover the available nodes (and a node's full schema) over the API.
# $AUTH is "Authorization: Bearer <token>" — see /api-reference#authentication
curl -s https://dev-api.modtechlabs.com/node_definitions/ -H "$AUTH" | jq '.[].name'
curl -s https://dev-api.modtechlabs.com/node_definitions/123 -H "$AUTH" \
| jq '{name, type, cost_in_credits, input_schema, output_schema}'
Inputs
Every workflow starts with an input node that points at one of your files:
- Image Input, Video Input, Audio Input, Mesh Input, Text Input
- NeRF Input, Texture Set Input, Resource Input, Script Input
Image processing
- Resize Image — scale to target dimensions
- Blur Detection — detect and score image blur
- Blur Cleanup — AI-powered deblurring
- Blur Image — CPU, GPU, and OpenCV variants
- Background Extraction — remove backgrounds via ML segmentation
- Transparent Background — alpha-matte cutout
- Exposure Balance — normalize exposure across image sets
- HDR Converter — tone-map / convert HDR imagery
- Inpainting — fill masked regions (incl. depth-guided iterative inpaint)
Texture generation (PBR)
Generate PBR material maps from photos — many have a fast classical path and an ML path:
- Albedo (+ AlbedoML) — base color
- Normal — normal maps
- Roughness, Metalness, Specular
- Ambient Occlusion (+ AmbientOcclusionML)
- Cavity (+ CavityML) — fine surface detail
- Height — height/displacement maps
3D & photogrammetry
- Colmap — structure-from-motion / dense reconstruction
- DepthAnything — monocular depth estimation
- Gaussian splatting — Instant Splat, Open Splat, SuperSplat, gaussian-pro
- Radiance fields / NeRF — rad-foam, sparse-craft, and related splat/NeRF nodes
- Mesh Converter — convert between OBJ, FBX, GLB, PLY, USD
- Decimation — reduce polygon count while preserving shape
- Place Mesh on Depth, SLDPRT / IFC / USD converters
Video
- Video Converter — transcode between MP4, MOV, AVI (FFmpeg CPU/GPU)
- Video Split — split into frames or segments
- Video Object Remover / Video Painter — inpaint moving objects
Audio & speech
- Audio Converter — MP3 / WAV / FLAC
- ChatterboxTTS, OpenVoice — text-to-speech / voice
AI & ML
- Claude AI Process — LLM-assisted analysis
- Chatbot — conversational node
- CNN Image Classifier — image classification
- SAM2 Mask — Segment Anything 2 masking
- Anomalib — visual anomaly detection
- Comic suite — description / LLM extract / pricing
AEC (architecture, engineering, construction)
A domain pack for built-environment data:
- IFC to GLB Converter — BIM model conversion
- Excel Parser, Reference Table Loader, Multi-Source Reconciliation
- Compliance Window Check, Labor Variance Check
- Exception Report Generator, Provenance Stamp
Export & utilities
- Unity Export, Unreal Export, USD Converter, GLB Combine
- Zip Download / Zip Folder — package outputs
- Blender Script — run Blender Python
- Script Input — run a custom script
- Accelerator Benchmark — CPU/NPU/CUDA/ROCm/Intel compute benchmark
Don't see what you need? Build your own with the custom container node wizard, or browse community nodes in the Marketplace. The list above is a representative slice — the full, live catalog is in the Workflow Builder's node sidebar and via
GET /node-definitions/.