What actually helped in an 8-day AI video-editing trial (Jun 2026): editor plugins, agent-made B-roll, Claude + HyperFrames motion graphics, Omni footage edits

Creator-reported: statements attributed to the creator(s) come from discovery's reading of auto-caption transcript excerpts; the curator checked the video metadata (title, date, length) and any cited documentation pages, not the spoken claims. Nothing was tested; promotional or sponsored framing applies where stated. Dan Kieft and team (2026-06-12) sorted tools into money-saving, speed and creativity. Useful: Higgsfield plugins inside Premiere Pro, After Effects and Figma (generate shots, reframe 16:9 to vertical, 4K upscale; background removal "mixed results"); Higgsfield "Supercomputer" agent producing intro scenes and B-roll/infographic "cards" for explainers, with failures retried on request; Claude Code + HyperFrames (HeyGen's Apache-2.0 open-source HTML-to-video renderer "built for agents", R6) to add titles, captions, lists, counters and UI animations around a raw talking-head recording by describing them, then iterating in plain language ("make the transition faster"); Google Omni for swapping objects/characters/environments in real footage while keeping motion (worked better with a reference image; some breakage). Verdict: tools speed editors up rather than replace them. Flag: he says Omni gave him problems in Europe and he used a VPN - circumventing regional availability can breach service terms; record as a limitation (regional availability), not a method. Related build: with Claude Code he made an Apple-style scroll-animated product page from Nano Banana 2 start/end frames animated in Kling 3.0 (3 s segments joined) - he used a real laptop brand's product and logo, a trademark caution for public use (R3). Craft tips from R4/R5: put most effort into the still image (motivated, directional lighting; composition such as subject on the left third), use an LLM to draft multi-shot prompts but rewrite them yourself, and expect text-to-video to vary run to run. Curator qualifications: Creator-reported: statements attributed to the creator(s) come from discovery's reading of auto-caption transcript excerpts; the curator checked the video metadata (title, date, length) and any cited documentation pages, not the spoken claims. Nothing was tested; promotional or sponsored framing applies where stated. All cited documentation excerpts located on the live pages by curator GET; video metadata checked where cited. Creator video (auto-caption transcript excerpts read by discovery); not curator-verified beyond video metadata. Limitations: Creator statements were read by discovery from auto-captions (ASR); not re-verified by the curator. Sponsored context likely (Higgsfield); partial/condensed reads for R3-R5; nothing executed. Evidence status: statements located in the cited sources by a correlated desk review (Claude); not executed; not independently reviewed.

{
  "citations": [
    {
      "url": "https://www.youtube.com/watch?v=9-iSl83dF34",
      "note": "~7:20; ~9:10; ~0:40"
    },
    {
      "url": "https://www.youtube.com/watch?v=0B_xyflXrwc",
      "note": "~7:20; ~9:10; ~0:40"
    }
  ]
}