Back to blog

Motion Graphics vs AI Video Generators: Which Fits Your Brief?

How to choose between code-drawn motion graphics and generative AI video models: exact text, numbers and charts versus realistic people and places, with four example briefs and when to combine both.

Oct 1, 2026Opus 5.5 Video

Sources were checked on 1 October 2026 and are listed at the end. This is not a benchmark: we did not run the same brief through every tool.

Two ways to make a video

Code-drawn motion graphics. A program draws every frame. On this site, Claude Opus 5.5 reads your brief, writes an HTML animation (with Canvas, SVG or libraries such as GSAP, p5.js or three.js), checks preview frames, and the site's renderer draws each frame in a headless browser and encodes an MP4. Claude itself outputs text, not video: Anthropic's models overview lists text and image input and text output. The renderer is part of this site, not an Anthropic video model.

Generative video models. A model produces the frames directly from a text prompt, and depending on the model also from images or a video. Two examples from their official documentation on the date above: Google's Gemini API describes Veo 3.1 as a model for generating video with native audio, with video extension, frame-specific generation and image-based direction, and recommends Gemini Omni Flash as the default model for turning text and images into short videos and refining them over several turns. Runway's API lists several video models, its own and third-party, that take text, image or video input, billed in credits.

Here is what the code-drawn side looks like, using an existing sample from the homepage:

Existing homepage sample, made with this generator's renderer in a Claude Code session (not a customer video). The figures are illustrative, not real results.

  • 15 s
  • 1:1 · 1080×1080
  • 30 fps
  • Synthesized music
Brief: An animated bar chart titled "Weekly active users", January to June: 8.2k, 11.5k, 15.1k, 21.4k, 32.0k, 48.2k. Dark background. Bars grow one per beat and the last bar is orange, while a headline number counts up to 48,200. Then show "+488% in six months", trace a trend line across the bar tops and end on "Up about 43% every month".
Start with this brief

Side by side, with conditions

Code-drawn motion graphics compared with generative video models
Code-drawn motion graphics (this site)Generative video models
Text, numbers and chartsDrawn as text and shapes from the brief, so wording and values are what the code says. Claude can still misread the brief or add text you did not ask for; check every word and number.Depends on the model and the prompt. The official pages we checked make no claim either way, so test with your own text before relying on it.
Realistic people and placesNot supported: the output is illustration, type, shapes and charts.What these models are built for: footage-like scenes, camera moves and natural motion.
Input assetsA text brief; public web pages linked in it can be read for names, wording and images. No file uploads.Text, and depending on the model images or video, such as a reference image or a clip to extend or edit.
Changing the resultEdit the brief and generate again. No timeline or chat-based revision.Depends on the tool: some offer extension, frame control or multi-turn edits; otherwise regenerate.
Checking the outputCompare text, numbers, labels and timing with your source.Check people, hands, physics and continuity, plus any text in the frame.
CostCredits per video, charged as Claude works and capped by length (details below).The provider's own pricing, often per generated second or in credits; see each provider's pricing page.

On this site, a 15-second video starts at 50 credits and costs at most 200; a typical one ends around 40. Credits not used come back and a failed video is refunded. Plans are on the pricing page.

Four briefs and the workflow for each

1. New product benefit cards. A short announcement with the product name, a feature shown in a simplified interface and a button with your call to action. Choose code-drawn motion graphics: the product name, labels and call to action have to be spelled exactly. More on the product launch video maker.

New feature announcement15 s · 16:9

Suggested brief with a made-up product; not rendered yet. Replace the details with yours.

Brief: A 15-second new feature announcement for Pathline. The new feature turns your backlog into a weekly plan. Use a deep navy background with mint accents. 0–4 s: a "New" badge pops in above the name "Pathline". 4–11 s: a simplified, illustrated app window (not a real screenshot) shows the feature working in two steps with short labels, under the headline "Your week, planned in one click". 11–15 s: end card with a button that says "Join the waitlist", held for about two seconds. Crisp interface-style motion, upbeat music.
Start with this brief

2. A short phrase as kinetic typography. Four lines of text landing on the beat in a vertical video. Choose code-drawn motion graphics: the words are the whole video, so line breaks and spelling must follow your text. More on the kinetic typography generator.

Short statement15 s · 9:16

Suggested brief with original example lines; not rendered yet.

Brief: A 15-second kinetic typography video, vertical 9:16, with the text in the middle of the frame and clear of the top and bottom edges, where phone apps often overlay buttons and captions. A short statement, one line per beat of punchy synthesized music, in this order: "Start small." / "Ship often." / "Listen closely." / "Keep going.". Make "Keep going." stand out with color and scale. Use a black background, white type and one electric yellow accent. Heavy sans-serif type; each line enters with a quick, snappy move and holds long enough to read, at least about a second, before the next replaces it; end with all lines together for the last 2 seconds. Show only these words, spelled and broken into lines exactly as written.
Start with this brief

3. A chart of how a number changed. Three values as an animated bar chart. Choose code-drawn motion graphics: each value is copied into the brief and drawn as a labeled number, which you then compare with your source. More on the animated chart video maker.

Bar chart with sample data15 s · 1:1

Sample data (A = 20, B = 35, C = 45 items), not real results; not rendered yet.

Brief: A 15-second animated bar chart, square 1:1. Title: "Sample data". Values in items, in this order: "A": 20 items; "B": 35 items; "C": 45 items. Bars grow one after another from a baseline at zero; label each bar with its name and its value exactly as written, followed by the unit. Plain dark background, white labels and one accent color. Hold the finished chart for the last 3 seconds. Use only these numbers: do not round, convert, total or compare them, and add no percentages, growth rates or other figures.
Start with this brief

4. A person in a real place. For example: "A barista in a sunlit café pours latte art, filmed handheld at counter height, shallow depth of field, eight seconds." Choose a generative video model. This site cannot make realistic people or filmed-looking scenes, so there is no button here; use a video model's own tool and check its documentation for length, resolution and pricing.

Using both in one video

  1. Make the footage, such as the café shot, with a generative video model in that model's own tool.
  2. Make the parts that need exact text with this generator: a title card, a chart, a kinetic typography line or an end card with your call to action. Choose the same format as the footage (16:9, 9:16 or 1:1).
  3. Put them together in a video editor of your choice. This site does not import footage, edit a timeline or export layers, so cutting, transitions and the final mix happen in the editor.

FAQ

Should I use Claude motion graphics or an AI video model for my video?

Decide by what has to be right. If it is text, numbers, a chart or a layout, use code-drawn motion graphics and check the result. If it is a realistic person, place or camera move, use a generative video model. If you need both, make each part with the right tool and combine them in an editor.

Which video workflow fits product text and charts versus realistic footage?

Product text and charts fit a code-drawn workflow, where every word and value comes from your brief. Realistic footage fits a generative video model. Whichever you choose, review the output before publishing: neither approach guarantees a mistake-free result.

Is Claude a video model?

No. Claude models output text. This site has Claude Opus 5.5 write animation code and renders it to video with its own renderer.

Can AI video models show readable text?

It depends on the model and the prompt, and the official pages we checked make no claim either way. If text matters, test with your exact wording, or add the text in a code-drawn or editing step.

Can this site make a video of a real person or place?

No. It makes illustrated motion graphics, typography and charts. Linked public pages can supply names, wording and images, but it does not generate footage-like scenes.

Sources

All checked on 1 October 2026:

Related pages