Skip to main content
Agent skill

Video To 4k

Upscales a single low-resolution video to 4K with Video2X (Real-ESRGAN), then encodes to unified 3840×2160 H.265 Main10 40Mbps + AAC 320kbps MP4. Source frame rate is preserved (no fps=60 duplication, no RIFE). Already-4K sources skip Video2X and go straight to FFmpeg. Use when the user wants video-to-4k, 4K upscale, UHD export, Video2X, Real-ESRGAN, H.265 Main10, or convert an SD/HD clip to 4K.

Get started

Get the complete skill folder, including its scripts and reference files.

When to use this skill

Use Video To 4k when an AI agent needs a reusable SKILL.md workflow for this job: Upscales a single low-resolution video to 4K with Video2X (Real-ESRGAN), then encodes to unified 3840×2160 H.265 Main10 40Mbps + AAC 320kbps MP4. Source frame rate is preserved (no fps=60 duplication, no RIFE). Already-4K sources skip Video2X and go straight to FFmpeg. Use when the user wants video-to-4k, 4K upscale, UHD export, Video2X, Real-ESRGAN, H.265 Main10, or convert an SD/HD clip to 4K.

When not to use it

Skip Video To 4k when the task is outside the coding category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.

How to install

Where will you use it?

Install scope
Claude Code folder setup
mkdir -p ~/.claude/skills/video-to-4k
# Save SKILL.md to ~/.claude/skills/video-to-4k/SKILL.md
  1. Claude Code personal install: place the skill and its bundled files at ~/.claude/skills/video-to-4k/SKILL.md. For a shared project, commit the folder at .claude/skills/video-to-4k/ instead.
  2. Current Claude Code detects SKILL.md changes in watched directories during the session. If you just created a top-level skills directory, run /reload-skills. Check that the skill is listed before testing its trigger.
  3. For Claude chat or Cowork in the Desktop app, manage uploaded skills in Customize > Skills; copying files into a local Claude Code folder is not a Desktop install. Other agents may use different install locations.

Full install guide for Claude, Cursor, and Codex →

Try a starter prompt

After installing the skill, paste this starter prompt into your agent to verify it triggers on a small task.

Starter prompt
Use the Video To 4k skill for this task: Use Video To 4k when this documented workflow matches the task. Confirm any required inputs or files before making changes.

What supports this recommendation?

70/100 provisional content score. This measures documentation, practical detail and source signals. It is not a measured task success rate or a security certification.

Documentation
77/100
Practical detail
68/100
Source evidence
61/100
Source signals
70/100
Model compatibility evidence
  • claudeuntested

    No model-specific signal or recorded compatibility test was found.

  • chatgptuntested

    No model-specific signal or recorded compatibility test was found.

  • geminiuntested

    No model-specific signal or recorded compatibility test was found.

  • copilotuntested

    No model-specific signal or recorded compatibility test was found.

  • llamauntested

    No model-specific signal or recorded compatibility test was found.

  • perplexityuntested

    No model-specific signal or recorded compatibility test was found.

  • mistraluntested

    No model-specific signal or recorded compatibility test was found.

  • grokuntested

    No model-specific signal or recorded compatibility test was found.

Inferred compatibility is based on content signals. Check the app, its available tools and setup requirements before running a skill.

What this skill does

# Video to 4K

Convert a supported video to a **unified 4K master**:

| Spec | Value | |------|-------| | Resolution | 3840×2160 | | Frame rate | **Source fps kept** (no `fps=` filter, no RIFE) | | Video | H.265 Main10 (`libx265`, `yuv420p10le`) | | Video bitrate | 40 Mbps | | Audio | AAC 320 kbps | | Container | MP4 (`.mp4`, `hvc1` tag) |

## Pipeline

1. **Probe** with ffprobe (width, height, fps, audio). 2. **Already 4K?** (`width ≥ 3840` and `height ≥ 2160`) → skip Video2X. 3. **Below 4K** → [Video2X](https://github.com/k4yt3x/video2x) Real-ESRGAN upscale (scale 2 or 4) to intermediate under `video-to-4k/upscaled/`. 4. **Always** FFmpeg final encode → `video-to-4k/` at the unified specs above.

## Rules

When this skill applies, read and follow [skill-dependency-manager](../skill-dependency-manager.md) — run scripts as documented, install missing tools into `.dependency/`.

- Run `convert.py` through **`.dependency/python/python`**. Never use host `python` / `ffmpeg` / `video2x`. - **Never overwrite sources.** Outputs go under `video-to-4k/` (final) and `video-to-4k/upscaled/` (Video2X intermediate). - Use the bundled script — do not hand-write equivalent `video2x` / `ffmpeg` commands. - **One file per run** — pass `--video` with a single file; repeat for each clip in a batch.

## Setup (first run)

### FFmpeg + Python

Populate `ffmpeg` and `python` under `.dependency/` per skill-dependency-manager.

### Video2X (CLI)

Download the **CLI** release (not the Qt6 installer) from [video2x releases](https://github.com/k4yt3x/video2x/releases/latest) and extract to `.dependency/video2x/`:

| Platform | Release asset | |----------|---------------| | Windows | `video2x-windows-amd64.zip` → `video2x.exe` | | Linux | `Video2X-x86_64.AppImage` (or distro package) → register as `bin` |

Register in `.dependency/manifest.json`:

```json "video2x": { "populated": true, "bin": ".dependency/video2x/video2x.exe" } ```

Use `video2x` (no `.exe`) or the AppImage path on Unix. Requires a **Vulkan** GPU and AVX2 CPU. Docs: [Command Line](https://docs.video2x.org/running/command-line.html).

## Quick Start

```bash .dependency/python/python .ai/video-to-4k/convert.py --video path/to/clip.mp4 ```

Example:

``` assets/video/clip.mp4 → assets/video/video-to-4k/upscaled/clip.mkv (Video2X, only if below 4K) → assets/video/video-to-4k/clip.mp4 (final master) ```

Already-4K source → only `video-to-4k/clip.mp4` (FFmpeg only).

Anime / cartoon content (Real-ESRGAN anime model):

```bash .dependency/python/python .ai/video-to-4k/convert.py --video clip.mp4 --anime ```

## Defaults

| Setting | Default | Notes | |---------|---------|-------| | Upscaler | Real-ESRGAN `realesrgan-plus` | General / live-action | | `--anime` | `realesr-animevideov3` | Anime / illustration | | Scale | Auto 2 or 4 | Smallest factor that meets or exceeds 4K | | Intermediate | High-bitrate HEVC MKV in `video-to-4k/upscaled/` | Kept for re-encode; delete manually or `--clean-upscaled` | | Final | Always re-encode | Even if source already matches specs |

## Common Flags

`--video` · `-o` / `--output` · `--anime` · `--gpu` · `--clean-upscaled`

```bash .dependency/python/python .ai/video-to-4k/convert.py --video clip.mp4 .dependency/python/python .ai/video-to-4k/convert.py --video clip.mp4 --gpu 0 --clean-upscaled .dependency/python/python .ai/video-to-4k/convert.py --video clip.mp4 -o out/clip.mp4 ```

**Never overwrite source files.** Input must be a single video file (`--video`), not a directory. Supported inputs: `.mp4`, `.mkv`, `.mov`, `.avi`, `.webm`, `.wmv`, `.flv`, `.m4v`, `.mpeg`, `.mpg`, `.ts`, `.mts`, `.m2ts`, `.3gp`, `.ogv`.

## Agent Notes

1. Use the bundled script only. 2. Missing Python / FFmpeg / Video2X → populate `.dependency/`, set `populated: true`, retry the same command. 3. Tell the user where `video-to-4k/` (and `video-to-4k/upscaled/` if used) outputs are; they swap assets manually. 4. Default model is **general** (`realesrgan-plus`). Pass `--anime` for anime/cartoon. 5. Do **not** force fps. Keep the source frame rate. Need 60fps interpolation → **video-to-60fps** first (RIFE at source resolution). 6. Pipeline / FFmpeg / Video2X details: [reference.md](reference.md)

## Tests

From repo root:

```bash .dependency/python/python .ai/video-to-4k/test_convert.py ```

Manual CLI examples: [cli/video-to-4k.md](../../../cli/video-to-4k.md)

Intended uses

  • Use Video To 4k when this documented workflow matches the task.

Implementation notes

  • Review the source instructions and adapt inputs before running the workflow.

Try it, then tell us how it went

A copied prompt does not tell us whether the task worked. Report your result after checking the output. We collect your tool and outcome, never your files or prompt.