Game patch notes video: dim the capture, one caption card per change

To make a game patch notes video, record a short capture of the changed areas, darken it with video filter's dim operation so text stays readable ($0.02, and the check route is free), then burn one caption card per change with video captions cues ($0.20 for a clip up to 60 seconds). Eight changes at five seconds each fits in 40 seconds and costs $0.24 with an optional trim ($0.02) included.

Why dim the capture first?

Busy gameplay behind white text is hard to read. The video filter docs describe dim as a multiplier on luma between 0 and 1: 0.45 makes the picture darker, 1 changes nothing, black stays black and the chroma is not shifted. The filter keeps the geometry, frame rate and audio of the source and returns a new MP4. Source clips are limited to 300 seconds.

Can you check the filter before paying?

Yes. POST /v1/video-filter/check runs the same schema, op and source checks as the encode, boots no encoder and creates no job. It returns valid, diagnostics, and a next_action of submit_video_filter or fix_program_and_recheck. A program that passes can still fail on the worker, so keep the capture under the limit.

import os
import requests

key = os.environ.get("SUME_API_KEY", "")
if not key:
    raise SystemExit("Set SUME_API_KEY first")

resp = requests.post(
    "https://api.sume.com/v1/video-filter/check",
    headers={"Authorization": f"Bearer {key}"},
    json={
        "video_url": "https://media.sume.com/artifacts/artf_demo/patch-capture.mp4",
        "ops": [{"op": "dim", "amount": 0.5}],
    },
    timeout=60,
)
resp.raise_for_status()
body = resp.json()
print(body.get("valid"), body.get("next_action"))

How should the change cards be written?

Cues carry text, start and end in seconds, and no speech-to-text is involved. Short lines render better than sentences, and the design overrides in the captions docs let you change phrasing (max_words, max_chars), placement (anchor_ratio) and colors for one request. Give each change one card: the verb first, then the thing, such as Fixed: ladder clip in the east tunnel.

What are the limits to plan around?

The captions price is stated for clips of at most 60 seconds, so split a long patch (more than a dozen changes) into two videos. If the changelog already exists as text, write the cues from it instead of re-typing. The captioned file comes back as a new video; the dimmed source stays untouched in your workspace.