Case study · September 2026

What the most-liked Claude motion graphics have in common

We measured every cut, color and sound in the 63 Claude Opus 5.5 motion graphics that passed 100 likes on X. The craft didn't predict the likes. Reach did.

Save the headline stat as an image

3.5×

median likes when the post names Claude

42%

of all likes went to 10 showreels

0 of 12

craft measures that predicted likes

Made with Revid · 30s · 1 prompt · 131 credits

We gave Revid the link to this page. It made the video.

Revid read the study, picked the takeaways, wrote the script, voiced it and animated every scene. Remake it with the same settings, or point the brief at your own article.
The prompt
https://www.revid.ai/case-studies/claude-opus-motion-graphics Create a less-than-one-minute video that shows the best takeaways from this article. It will be displayed on top of the article as a summary of the article, start with a great hook. Use the exact numbers from the article.

Motion graphics · Ultra · Minimal & type style · 16:9 · Alex voice · No captions · Startup Energy music

The videos

All 63, sorted your way

Hover to play. Click one to watch it and reuse its prompt.

Reach

Likes followed views, and views followed the word Claude

Each frame is a video, placed by views and likes (log scale). Hover one to see it.

Swipe sideways to see the whole chart

1003001k3k10k1k3k10k30k100k300k1M3Mviews on Xlikes

Names Claude vs does not

621

Names Claude (42)

3.5×

median likes

175

Does not (21)

Showreels vs everything else

792

Showreels (10)

3.2×

median likes

245

Others (53)

  • All of the 10 most-liked posts named Claude. Their like rate was the same as the rest; they just reached 2.7× as many people.
  • The showreels share a recipe: 8 of 10 run exactly 15.1 seconds, 9 sit on a dark background, and all use music with no voice.

Craft

No craft measure predicted likes

Correlation of each measure with likes, with its 95% interval. Every line crosses zero.

Swipe sideways to see the whole chart

-0.6-0.30.0+0.3+0.6Length+0.03Cuts per 10 s+0.10Time of first cut-0.01Mean saturation+0.05Motion intensity+0.13Integrated loudness+0.14Average shot length-0.06Share of light frames+0.06Colorfulness+0.06Motion in first 3 s-0.02Brightness in first 2 s-0.12Day posted+0.16correlation with likes (Spearman rho, 95% interval)
  • Every video here already cleared a quality bar. Past that bar, distribution decided, not the edit.

Anatomy

The typical hit: 29 seconds, 3 cuts, music, no voice

The editing rhythm of the 10 most-liked videos. Each tick is a hard cut.

  • Openings are calm: the first cut comes at 3.9 seconds, and 59% show readable text within two seconds.
  • 78% are landscape and 71% use music only, with no voiceover.

Color

Black, white, grey and one accent

The palette of the whole week, then every video's five dominant colors, most-liked first.

#e6e8e6 · 43%#141318 · 37%#726d68 · 13%#d78d45 · 5%#493cca · 3%

Hover a strip to see the video and its colors.

  • Dark and light backgrounds got the same likes (346 vs 280), but dark videos cut 2.8× as often.

What to copy

Making one this week

1

Say what made it

When a model is the story, name it. The views follow the search.

2

Steal the 15-second reel

Dark background, one accent color, music, a logo at the end.

3

Edit less

The typical hit cut once every nine seconds. Spend the time on the idea.

Made with Revid

Make one from any of these 63 videos

Open a video in the gallery and press Use as prompt. Revid builds your version with your text, brand and music.

Free to start · no credit card

How we measured this

  • Sample: the 63 motion graphics videos about or made with Claude Opus 5.5 posted on X from September 23 to 27, 2026 (UTC, plus two posts late on September 22) that passed 100 likes, from our launch-week library. Likes, views, reposts and replies are X's public counts read on September 28, 2026, so later posts had less time to collect likes. Two posts show very few views for their likes; dropping them moves the median like rate from 0.95% to 0.93%.
  • Cuts: ffmpeg scene detection at 30 fps and 320 px wide, a cut when the scene score is at least 0.30, detections under 0.3 seconds apart merged. The threshold was calibrated by eye on contact sheets of six videos: it counts hard cuts and full-frame flashes, and mostly ignores morphs, dissolves, word swaps on a static background and split-screen panel swaps. Cut counts are therefore conservative; 21% of videos have no hard cut at all. Two long posts are measured on their first 175 seconds.
  • Color: frames sampled twice a second at 96 px, k-means with five clusters in CIELAB. Motion is the mean absolute difference between greyscale frames at 10 fps. Loudness is EBU R128 integrated loudness (median -14.2 LUFS). Voice was detected with a speech-recognition model plus voice activity detection on the first 90 seconds (at least 12 words and 4 seconds of speech); two videos whose only vocals are a song were set to music by hand. "Opens on text" was checked by eye on four frames per video.
  • Statistics: Spearman correlations with bootstrap 95% intervals; group comparisons with permutation tests and Benjamini-Hochberg corrected q values. With 63 videos, a single video can move a group median, and groups under five videos are not compared. Every video here was already a hit, so the study compares hits with bigger hits and says nothing about what separates a hit from a video nobody saw. Follower counts are not in the data.
  • Titles, formats and "made with" are editorial labels from the gallery.

Data pulled September 28, 2026. The index grows every day; numbers are refreshed on a schedule and the date above always tells you which snapshot you are reading.

Cite this study

Revid (2026). What the Most-Liked Claude Opus 5.5 Motion Graphics Have in Common: 63 Videos, Every Cut Measured. https://www.revid.ai/case-studies/claude-opus-motion-graphics

More from the Revid index

🎵Case study

11

AI music video generators scored on 10 checks

Best AI music video generators

We scored 11 AI music video generators on 10 things a musician needs, from full-song output to lip sync and publishing, with a source for every claim. Plus what ChatGPT and Google AI Overviews recommend, and why the most-named tool cannot take a song.

Read the study
📼Case study

1.55×

median views for AI-made nostalgia vs real footage

The 80s AI video trend

Google searches for 80s AI video prompts went from almost nothing to 1,945 impressions in a day in September 2026. We read that spike against 1,676 80s TikToks: formats, views, likes, and how AI-made nostalgia performs against real footage.

Read the study
⏱️Case study

713k

AI video clips measured across 9 model families

AI video model index

Production numbers from 713,649 AI video clips generated on Revid in 90 days: share by model, time to generate, rejection rates, clip lengths and aspect ratios for Seedance 2.5, MiniMax Hailuo H3, PixVerse, Kling, Veo 3.1 and Sora 2.

Read the study
💘Case study

7×

more views for couples content than crypto

Love beats money

We measured the average views of 46 hashtag themes across 2.4 million TikToks. Couples content averages 7x the views of crypto. Here is the full ranking of what TikTok rewards, from romance to real estate.

Read the study
🫣Case study

2.2×

more views for handheld footage than selfie videos

Your face costs views

Selfie-style videos average half the views of handheld footage, faceless videos beat person-focused ones, and animals beat everyone. The data on whether you need to show your face on TikTok.

Read the study
🤖Case study

2×

average views for AI-generated TikToks vs real footage

AI slop vs real footage

In our index, AI-generated TikToks averaged twice the views of real footage since May 2025. On X, Instagram and Threads, AI visuals are 5x more common in the lowest engagement tier than in the viral tier. Both are true.

Read the study
🩺Case study

3,000

viral health TikToks audited from their transcripts

TikTok health claims

An audit of the 3,000 most-viewed English health TikToks in our index: how many make a health claim, how many back it with anything, how many sell a product, and which claims went most viral.

Read the study
FAQs

Frequently Asked Questions

What do the most-liked Claude Opus 5.5 motion graphics have in common?
Among the 63 Claude Opus 5.5 motion graphics that passed 100 likes on X in launch week, the typical video runs 29 seconds in 16:9, has 3 hard cuts (about 1.1 per 10 seconds), waits 3.9 seconds for its first cut, carries music with no voiceover and shows readable text within two seconds. But none of those craft measures predicted likes. What separated the hits was reach: posts that named Claude got 3.5× the median likes, and 15-second showreels took 42% of all likes with 10 videos.
Do faster cuts get more likes on motion graphics videos?
Not in this data. The correlation between cuts per 10 seconds and likes was 0.10, with a 95% interval that includes zero. The top 10 videos did cut faster (2.2 vs 1.0 per 10 seconds), but on 10 videos that gap is not statistically reliable, and showreels, which cut fastest, won through reach rather than a higher like rate.
How long should a motion graphics video be on X?
Length did not matter for likes (correlation 0.03, interval includes zero). The median video was 29 seconds and 25% of the 63 were exactly 15.1 seconds, the length of the showreel format that won the week.
Should AI motion graphics have a voiceover?
Most of the successful ones did not: 71% carried music or sound design only and 25% had a voiceover or dialogue. Voice videos had a median of 228 likes against 373 for music-only, a difference too small to mean anything at this sample size.
How was this measured?
We downloaded every video, detected hard cuts with ffmpeg scene detection (score at least 0.30, calibrated by hand on six videos), clustered colors in CIELAB over frames sampled twice a second, measured motion as the mean frame difference, loudness with EBU R128, and speech with a speech-recognition model. Likes and views are X's public counts on September 28, 2026. Correlations are Spearman with bootstrap intervals; group differences carry Benjamini-Hochberg corrected q values.

Still have questions?

Contact us

Find detailed answers to 100+ questions about features, tools, and workflows

or check our markdown version optimized for LLMs