How To Create VOX STYLE Animation With Opus 5.5 | IN 10 MINUTES

Summary
Mark Ai Guy presents a tutorial demonstrating how to automate the end-to-end creation of Vox-style animated documentary videos using Anthropic's Claude Opus 5.5 integrated with Higgsfield AI via Model Context Protocol (MCP). The presenter demonstrates an agentic workflow where a 33-page master instruction prompt guides Claude to autonomously handle scripting, image generation, animation, voiceover, video assembly, title/description generation, and thumbnail creation.

---

What is shown

• [00:06] Flashback to previous manual workflow video in an editing timeline and script document.
• [01:33] Sample demonstration video about D.B. Cooper created using the automated workflow.
• [02:22] Higgsfield AI interface: navigating to the MCP tab, selecting Claude, and connecting the Higgsfield MCP connector URL into Claude's connector settings.
• [03:09] Selecting the Opus 5.5 model in the Claude interface.
• [03:24] Downloading the 33-page engine document (Higgsfield VOX Documentary Agent V1 - Full Autonomous Video Production Engine) from a Google Drive link.
• [03:59] Uploading the engine PDF into Claude Opus 5.5, which prompts the user with five setup questions: niche, topic, length, format, and voice.
• [04:39] Entering prompt parameters: Crime and documentary, Tsutomu Yamaguchi (survivor of both Hiroshima and Nagasaki bombings), 30 seconds, 16:9 aspect ratio, and the "Mark" voice.
• [04:58] Claude executing Higgsfield MCP tool calls behind the scenes (image generation, Seedance image-to-video, voiceover generation, and video compilation).
• [05:55] Claude outputting the completed MP4 video file link (Yamaguchi_VOX_30s_FINAL_24fps.mp4) and technical summary (13 scenes, Seedance 2.5 model, 1080p, 24fps).
• [06:10] Requesting and generating a matching YouTube thumbnail ("SURVIVED TWICE"), YouTube titles, and a video description.
• [08:03] Full playback of the completed 30-second Vox-style animation about Tsutomu Yamaguchi.

---

Claims & numbers

• The presenter claims his previous tutorial got over 277,000 views, 12,000 likes, and nearly 2,000 comments in one month (displayed on screen as 279K views) [00:13].
• The presenter claims Claude Opus 5.5 combined with Higgsfield MCP completely eliminates manual multi-step image generation, animation prompting, and video stitching [00:40].
• The presenter states the entire generation process takes approximately 5 to 10 minutes depending on complexity [05:48].
• The demo video generated contains 13 source clips and images, timed to ~30 seconds in 1080p at 24 fps [05:55].

---

Notable quotes

• [00:40] "With the release of Opus 5.5, I was able to create a prompt that can automate the entire Vox-style animation workflow for you."
• [00:48] "All you have to do is paste in the prompt, answer a few questions, and the engine handles the rest."
• [07:08] "We started with nothing but an idea. Then the engine generated all the scenes, it created the images, it animated everything, it handled the voice, it generated the finished video..."

---

Assessment
This is a practical software workflow tutorial demonstrating third-party MCP tool-calling in Claude Opus 5.5. The waiting periods (5–10 minutes) are edited out for pacing, but the setup, Claude prompt interaction, tool logs, thumbnail creation, and final rendered video playback are shown in full.

Described by gemini-3.8-flash on 2026-09-29 from the video's audio and frames.