[Feature]: Create a getting-started video series #224

What would you like to happen?

A short video series for someone who already uses Apache Hop and is opening Data Hopper EDW for the first time. Same production as apache/hop#7931 and apache/hop#9081: a manifest, PNG stills, short muted clips, a generated voice-over, and a 1080p MP4. Nothing in this work uploads to YouTube.

The series follows Getting started — building an EDW on the retail-example project. Ten videos, about four minutes each. A later video may point at an earlier one in a single sentence. It does not re-teach it.

The longer tutorials under docs/video-tutorials/ stay. Those are how-to builds. This series is the map.

Series

Each row is a sub-issue of this feature.

How the topics fit

Metrics and analytics operations share video 10. A load writes timings to the OPS database, and Create operational analytics turns that database into a small warehouse of its own. Lineage is the last beat of that video: when a number looks wrong, follow the path.

Sources and source models stay apart. A source is the landing data and its catalog contract. A source model is the picture you draw before Generate Data Vault.

The resource definition group and updating it stay apart. One video is the list. The next video runs it.

These belong in the series and are beats inside the video where the viewer first needs them:

• Data Catalog. Video 03 shows a published source feed. Video 07 shows the target layouts the update writes back.
• Configure EDW setup. Video 02. It creates local-catalog and the shared configurations source-model, data-vault, business-vault, and dimensional. It does not create the group.
• EDW Journey. Video 01 uses the architecture poster as the map of the series. Video 06 opens the same Journey on group retail-sources.
• The four controls. Check model is video 05. Harvest, the schema gate, and data quality are named in video 07 as the wave around the update, one sentence each. The video runs the update.
• Semantic layer and a presentation. The last minute of video 09. The dimensional model is what you load. The semantic layer is how a page reads it.
• Execution map. One sentence in video 10, next to the metrics.
• Hop JDBC. One sentence in video 04. No client demo.
• AI help and the MCP server. Named at the end of video 01 as a later stop (training module 8).

Left out of this series: AdventureWorks, multi-team modeling, masking patterns, identity and survivorship, and a worked Kimball bus matrix. The retail tour and the training modules cover those.

Say resource definition group in full the first time in each video that uses it. Do not say RDG.

For every video

• Write the English narration in manifest.yaml. About 500 to 650 spoken words, in 6 to 8 scenes of 20 to 40 seconds, so the cut lands near four minutes. A longer language uses that same scene length. If a preview passes five minutes, cut the English before translating.
• Capture stills and short muted clips on retail-example, in one Hop theme for the whole series.
• English voice-over first. After that cut has been heard, add Chinese (zh-CN), Spanish (es-ES), Portuguese (pt-PT), French (fr-FR), German (de-DE), and Italian (it-IT). Product words stay in English in every language: Hop, pipeline, workflow, Data Vault, Business Vault, resource definition group, and the extensions .hsm, .hdv, .hbv, .hdm.
• Render at 1080p. Each scene stays up for the longest voice-over among the languages in that render, plus a short pause. The fade sits inside that pause.

Captures

• Capture the Hop window as large as it will go. It does not have to be exactly 1920×1080. The renderer fits each picture into that frame. A capture close to 1080p keeps the labels sharp.
• PNG stills when the viewer should read the screen. Muted MP4 clips, 30 fps, only when something moves.
• No secrets, personal paths, bookmarks, or notification popups.
• Name files by the scene id.
• Do not paint the watermark. The renderer puts Data Hopper at the bottom right.

Renderer

Episode 01 also adds the series notes and the renderer, the same way the Hop pull request landed the tool with the orientation episode.

• docs/video-series/SERIES.md — the rules a later episode needs.
• docs/video-series/glossary.yaml — product words that stay in English, and pronunciations (GUI said as "gooey", extensions said as letters).
• tools/video-series/render.py — check the manifest, speak the narration, time every scene to the longest language, and write the MP4, WebVTT, and one YouTube description per language.
• Generated audio and video stay under output/ and out of git.
• Pick the voice once, from a sample, and set that id on every episode before a publish render. The longer tutorials in this repo use ursa. The Hop series uses orion.

Each episode folder holds manifest.yaml, descriptions.yaml, images/, and clips/.

Sample

retail-example is the set for every episode.