skills.rest
NVIDIA
NVIDIAOFFICIAL
Explore all skills by this author →
WebsiteGitHub

jetson-video-recipe

Converts Jetson encoder workload intent into validated NVENC and PyNvVideoCodec recipe artifacts.

3.2k|370|Updated Feb 25, 2026
Search Tags:#video encoding#Recipe#h264#jetson#NVENC#pynvvideocodec#rate control
One-click install
npx skills add https://github.com/NVIDIA/skills --skill jetson-video-recipe
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: jetson-video-recipe
Source: https://github.com/NVIDIA/skills/tree/main/skills/jetson-video-recipe
Command: npx skills add https://github.com/NVIDIA/skills --skill jetson-video-recipe

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Turning a video encoding use case into correct NVENC or PyNvVideoCodec settings on NVIDIA Jetson requires deep knowledge of presets, rate control, profiles, and surface limitations. This Skill converts a plain-language encoder intent into one deterministic, schema-validated recipe with exact native and Python projections, without running any media operations.

Core Features & Use Cases

  • Deterministic Recipe Planning: Resolves use case, codec, resolution, frame rate, rate control, bitrate, latency, format, and profile into a canonical schema-2 nvcodec-recipe JSON with explicit defaults and rationale.
  • Dual Surface Projection: Projects the same intent to native Video Codec SDK (AppEncCuda CLI) and PyNvVideoCodec 2.1 sample configurations, reporting per-surface projection losses instead of silently dropping controls.
  • Replay Validation and Live Checks: Replays recipe derivation for deterministic validation and optionally classifies live compatibility against a setup environment artifact.
  • Use Case: Ask for an H.264 1080p60 6 Mbps CBR low-latency live-streaming recipe and receive one validated artifact with native and PyNvVideoCodec projections, defaulted values, and assumptions, ready for handoff to execution or benchmark workflows.

Quick Start

Ask your agent to plan an H.264 1080p60 6 Mbps CBR low-latency live-streaming recipe for both native Video Codec SDK and PyNvVideoCodec without running it.

Frequently Asked Questions about jetson-video-recipe

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create an NVENC encoder recipe for Jetson?▼

Describe your use case, codec, resolution, frame rate, and rate control intent, and the skill plans one schema-2 nvcodec-recipe via its recipe_model.py engine. It validates the recipe deterministically and projects it to both native AppEncCuda and PyNvVideoCodec configurations.

How do I configure low-latency H.264 live streaming with PyNvVideoCodec?▼

Specify a live_streaming use case with your resolution, frame rate, and bitrate; the catalog resolves low-latency tuning, CBR, GOP, B-frame, and VBV defaults. The same intent is projected to the PyNvVideoCodec 2.1 basic encode sample configuration without executing any encode.

Does PyNvVideoCodec 2.1 support H.264 High profile?▼

The released PyNvVideoCodec 2.1 sample parser cannot express a named profile, so an explicit High profile is preserved in the native projection but reported as a structured projection loss on the Python side. It is never silently dropped or downgraded.

Can I combine CQ with an average bitrate in an encoder recipe?▼

No. CQ is a VBR target-quality control and conflicts with an average bitrate in the shared recipe contract, so the skill explains the conflict and asks you to choose one. It never reinterprets the bitrate as a cap or silently discards CQ.

Does this skill run or benchmark video encodes?▼

No. Planning and validation are media-free and never launch AppEncCuda, PyNvVideoCodec samples, or benchmarks. Execution is handed to jetson-video-pipeline and measurement to jetson-video-benchmark via the validated recipe artifact.

Related Skills

Explore More Software Engineering Skills→
drewid74drewid74
8

grafana-prometheus-monitoring

Provision Grafana dashboards, datasources, and Prometheus SLO alerts as code.

Community
Advanced
hummbl-devhummbl-dev

de14-variable-control-and-isolation

Apply DE14 methodology to isolate single variables for causal impact analysis.

Community
Basic
GeneralReasoningGeneralReasoning
4

FFmpeg Format Conversion

Convert media files between containers and codecs using FFmpeg.

Official
Intermediate
fractaryfractary

doc-writer

Create or update documentation by rendering type-specific templates with runtime context.

Official
Advanced