Create Professional 3D Motion Graphics with Claude, Omni 1.1 Flash, and ElevenLabs
Workflow Overview
Learn how to produce high-end 3D motion graphics and commercial animations without touching Blender, After Effects, or complex software. This complete workflow guides you through reverse-engineering reference videos in Claude, rendering seamless clips with Google Flow's Omni 1.1 Flash, generating timed voiceovers with ElevenLabs, and transforming custom scripts into original animated assets.
🛠️ Tools & Resources Used
- Claude Desktop (Multimodal LLM) - Analyzes reference videos scene-by-scene, extracts exact narration scripts with timestamps, and drafts detailed motion graphics prompts for custom scripts.
- Google Flow (Omni 1.1 Flash) (AI Video Generator) - Renders smooth 3D motion graphics, animated typography, and dynamic transitions with embedded sound effects.
- ElevenLabs (AI Voiceover) - Generates natural, studio-quality speech timed to fit the target video duration.
- Premiere Pro / CapCut (Video Editor) - Stitches multi-part generations, synchronizes narration tracks, layers cinematic background music, and refines audio design.
⏱️ Quick Workflow Summary
- Step 1: Upload your reference video into Claude and run the master analysis prompt to generate shot-by-shot generation prompts.
- Step 2: Generate multi-part video segments in Google Flow using Omni 1.1 Flash at 9:16.
- Step 3: Extract the narration script via Claude and synthesize voiceover in ElevenLabs.
- Step 4: Assemble video clips, align audio pacing, and layer sound effects in Premiere Pro or CapCut.
- Step 5: Scale the workflow to produce 100% original 3D motion graphics from custom scripts.
📝 Step-by-Step Tutorial
Step 1: Deconstruct Reference Video with Claude Desktop [Timestamp: 00:00:24]
Open Claude Desktop and upload the reference motion graphics video that represents the visual style, typography, and animation quality you want to achieve. Paste Prompt 1 from the Master Prompt package. Claude analyzes the video scene-by-scene, identifying on-screen text, 3D assets, camera moves, and color palettes, then outputs structured, ready-to-use image-to-video generation prompts mapped to exact timestamps.
Prompt / Settings:
1. Analyze this video scene by scene and create detailed AI video-generation prompts for each shot so I can recreate it as closely as possible in Google Flow. Describe the camera angle and movement, subject/object actions, composition, environment, lighting, colors, facial expressions, motion graphics, visual effects, transitions, animation style, and cinematic style. Include the exact on-screen text, typography, and its animation wherever visible. Keep recurring characters, objects, colors, and visual elements consistent across scenes. Make each prompt detailed, accurate, and ready to copy directly into an AI video generator. Do not summarize or creatively reinterpret the video—focus on accurately recreating what is actually shown.
2. Could you extract the voiceover from this video? I need the exact narration so I can use ElevenLabs to generate the voiceover separately and easily add it to the clips generated by Flow.
3. You already know what kind of animations I'm aiming for. Now I want to make motion graphics in this style using my own script and audio, without copying any video. The 3D assets, icons, and elements should adapt to the script. Keep the animation style unique and different, skip the voiceover, but definitely add high-quality sound effects.
[ Add Your Script ]
Step 2: Render 3D Motion Graphics Segments in Google Flow [Timestamp: 00:00:56]
Open Google Flow and create a new project. Configure your project settings before generating: select the Omni 1.1 Flash model, choose a 9:16 vertical aspect ratio, and set the generation count to 1x to conserve credits. Since Omni Flash renders up to 10 seconds per clip and our sequence is 23 seconds, copy Claude's scene prompts and render the video in three sequential parts. Append a short note requesting ambient sound effects at the end of each prompt to let the model generate native audio alongside the motion.
Prompt / Settings:
Model: Omni 1.1 Flash
Aspect Ratio: 9:16
Generation Count: 1x
Duration: Custom (up to 10s per clip)
Audio Note: Include high-quality UI motion and ambient whoosh sound effects synchronized with elements
Step 3: Extract Script and Generate Voiceover in ElevenLabs [Timestamp: 00:01:58]
Return to Claude and run Prompt 2 from the Master Prompt to extract the verbatim narration script along with pacing marks. Next, navigate to ElevenLabs Text-to-Speech, paste the extracted script, and choose an engaging narrator voice (such as 'Adam'). Test the playback speed to ensure the voiceover duration matches your total visual sequence (e.g., exactly 23 seconds) before downloading the final audio file.
Step 4: Assemble Visuals, Sync Narration, and Polish Audio [Timestamp: 00:02:45]
Import your rendered Omni Flash clips and ElevenLabs narration into Premiere Pro or CapCut. Place the clips sequentially along the timeline and align them to the voiceover beats. Adjust clip speed or trim minor pauses if the spoken words need tighter pacing. Enhance the video by layering subtle transition whooshes, UI pop sound effects, and low-volume background music (-20 dB to -24 dB).
Step 5: Generate Completely Original Motion Graphics from Custom Scripts [Timestamp: 00:04:21]
To build brand-new motion graphics without copying an existing video, keep your established Claude conversation active so the model remembers the visual aesthetic. Paste Prompt 3 along with your custom narration script. Claude generates brand-new visual scenes, custom 3D icons, and animated layout prompts tailored specifically to your content. Copy these prompts into Google Flow, render your new clips using Omni 1.1 Flash, and assemble your original video.
💡 Key Takeaways & Pro Tips
- Segment Long Videos: Omni 1.1 Flash generates up to 10 seconds per output. For standard 20-30 second reels or shorts, break your prompt into 2 to 3 chronological segments and stitch them seamlessly in your video editor.
- Voice Matching Matters: Not every voice reads at the same cadence. If your voiceover runs too long or too short, swap voices (e.g., testing 'Adam' in ElevenLabs) or adjust speed settings by 5-10% to achieve frame-accurate synchronization.
- Retain Chat Context for Infinite Consistency: Keeping your creative session inside a single Claude thread ensures that new scripts naturally inherit the sophisticated lighting, bevels, typography, and animation style learned from earlier analysis.
📌 Original Source & Attribution
- Source Video: How to Make a Professional 3D Motion Graphics Video with Omni 1.1 Flash
- Original URL: https://www.youtube.com/watch?v=C9loLbFNveU
- Credit: Aira Tech