Category

Text to video generators, and the four different things they mean by it

Four categories of software are sold under one phrase, and they do not produce the same kind of picture, cost the same money, or fail in the same way. Below: the same paragraph put through all four, and what each one gives back.

Published 13 September 2026 · 10 min read · prices carried from this site’s vendor-page checks on 10 September 2026

The short version

The single most useful question to ask of any of them: how much of what appears on screen is decided by what I typed? In avatar tools, the words are; the picture is fixed. In stock tools, neither really is. In generative tools, almost nothing is. In motion graphics, both are, which is why it is the only category where a rewrite reliably changes the film.

“Text to video” means four different products

The phrase describes an input, not an output, which is why it has ended up attached to four unrelated things. Every one of them takes writing at one end. What comes out at the other is the part nobody puts in the marketing copy.

CategoryWhat appears on screenTypical entry priceFails when
AvatarA synthetic presenter reading your script$29/mo (Synthesia, HeyGen)Nobody needed a presenter
Stock footageLibrary clips matched to your sentences$17–$29/mo (Pictory, InVideo AI)The subject is your own product
GenerativeInvented footage, a few seconds at a timeVaries wildly; credit-meteredYou need a specific thing shown
Motion graphicsThe subject drawn: charts, steps, interfaces€3 per video (ExplainVids)You wanted a human face
The four categories sold as text to video. Entry prices carried from this site's vendor-page checks on 6 and 10 September 2026.

For the eleven-tool shortlist rather than the four-category map, the explainer video software comparison puts them in one table with published prices and a cost-per-video column.

The same paragraph, four ways

Take one ordinary piece of product writing — the kind actually sitting in a doc somewhere waiting to become a video:

“Our onboarding takes four steps. Most customers finish in under three minutes, and support tickets in the first week dropped by 40% after we shortened it.”

There are three concrete things in that sentence: a four-step sequence, a duration, and a percentage. What each category does with them is the whole difference between the products.

CategoryWhat you get backWhat happened to the 40%
AvatarA presenter in a studio set, speaking the sentence. Competent, clear, and visually identical to every other avatar video.Spoken aloud. Never shown.
Stock footageSomeone smiling at a laptop, a stock team high-fiving, a stock hand on a stock phone. Pleasant, generic, about nothing.Spoken aloud, possibly captioned over unrelated footage.
GenerativeSix seconds of striking, slightly dreamlike footage that is not your product and cannot be made to be.Gone. The model does not draw figures.
Motion graphicsFour numbered steps appearing in sequence, a timer, and a bar falling from 100 to 60 as the words are spoken.Drawn as a falling bar, on the syllable it is said.
One paragraph through four categories of text to video generator.

This is the test worth running before paying for anything. Write one paragraph containing a number you care about, put it through a free tier of each category, and see which one puts the number on screen. Three of the four will say it and not show it.

Avatar tools: a presenter reading your text

Synthesia and HeyGen are the category, both at $29 a month to start. You type a script, pick a presenter, and get a person on screen delivering it, with lip sync good enough that the average viewer does not question it and dubbing into dozens of languages that genuinely is the best in the market.

Right when

Wrong when

If you are already paying for one of these and the videos feel competent but flat, Synthesia alternatives grouped by why you are leaving covers the four exits, including the one that leaves the category altogether.

Stock-footage tools: your text over other people’s clips

Pictory and InVideo AI, $17–$29 a month. These split your writing into sentences, search a stock library for something plausibly related to each, and lay the clips end to end under a synthetic voice. For turning a blog post into something postable, they are genuinely the fastest route that exists.

The limit is structural rather than a quality problem: the pictures come from a library, so they can only ever be about your topic. A tool searching for “onboarding” returns a person at a desk. It cannot return your onboarding, because no stock library has footage of your product in it.

A quick way to tell whether this category fits: if you could swap your company name out of the script and the video would need no new footage, stock is fine. If the video is supposed to show what you specifically built, it is not.

Generative video: invented footage, and the length problem

Sora, Runway and Veo invent footage frame by frame from a description. At their best the results are extraordinary, and they are the only category here that can show you something that has never been filmed.

Two things keep them out of explainer work for now. The first is length: these models produce clips measured in seconds, so a ninety-second explainer becomes an editing project stitching many generations together — which is the timeline you were trying to avoid. The second is control. You cannot tell a generative model to put your pricing table on screen, spell a product name correctly, or draw a bar chart that matches your actual figures. It produces something in the neighbourhood of the description, and the neighbourhood is not close enough when the video contains claims.

For a mood piece or a title sequence, excellent. For a video whose job is to explain a specific thing accurately, the category is not there yet, and a per-second credit meter makes finding that out expensive.

Motion graphics: the text drawn as the thing it describes

The fourth category takes the same script and, instead of finding someone to read it or something to illustrate it, draws the subject. A four-step process becomes four steps appearing in sequence. A percentage becomes a bar that moves. An interface becomes an interface.

This is the oldest form of explainer video and, until recently, the most expensive: it is what agencies charge $3,000 to $15,000 for a minute of, because historically it meant a designer in After Effects moving keyframes until the animation agreed with the voiceover. What changed is that the agreement between animation and voice — the part that ate the hours — can now be measured rather than nudged.

The trade is the mirror image of the avatar category: there is no human face, so if the point of the video was a person addressing the viewer, this is the wrong instrument.

What it costs per finished video

Monthly prices hide the number that matters. A $29 subscription is $29 a video if you make one that month and $2.42 a video if you make twelve — and almost nobody makes twelve.

Videos a monthSynthesia $29HeyGen $29Vyond $99ExplainVids
1$29.00$29.00$99.00€3.00
4$7.25$7.25$24.75€3.00
12$2.42$2.42$8.25€2.92
Cost of one finished 60-second video at different volumes. Subscription figures are list monthly price divided by videos made; carried from this site's checks on 6 September 2026.

The honest reading of that table: at twelve videos a month a subscription wins on unit price, and at one or four it does not. Most people asking what a text to video generator costs are making one. The full breakdown of what an explainer video costs runs the same arithmetic against freelancers and agencies too.

Which one to use for which job

What you are makingThe category that fits
Compliance or safety trainingAvatar. A person addressing the viewer is the form.
The same video in eight languagesAvatar. Dubbing is the category’s real advantage.
A blog post, on LinkedIn todayStock footage. Fastest path from writing to postable.
A title sequence or a mood pieceGenerative. The only one that invents imagery.
A product, a process, or figuresMotion graphics. The subject gets drawn.
A landing page heroMotion graphics. The page is the subject — and a URL can be the whole input.
Matching the job to the category rather than to the marketing copy.

Where ExplainVids fits, and where it does not

We are the fourth category and nothing else. You type what the video should explain — or paste a URL and we read the page — and about two minutes later there is a narrated film with motion graphics, a real voice, captions lit word by word, vertical or widescreen, for €3 with no subscription. Your figures are drawn as charts because you gave them to us, not invented because a model guessed.

Pick us when

Pick something else when

The fastest way to find out which category you actually needed is to put one paragraph with a real number in it through this and watch where the number ends up. Three euros, about two minutes.

Make one and see

Questions

What is a text to video generator?

A tool that takes writing and returns a finished video without an editing timeline. The phrase covers four incompatible products: avatar tools that put a synthetic presenter on screen reading your words, stock-footage tools that match library clips to your sentences, generative models that invent new footage from a description, and motion graphics tools that draw the subject your text describes. All four are sold as “text to video” and only one of them will be what you meant.

Which text to video generator is best?

It depends entirely on what the picture is supposed to show. If a person addressing the viewer is the point, Synthesia or HeyGen at $29 a month. If you have a blog post and want it on LinkedIn by lunchtime, Pictory or InVideo AI at $17–$29 a month. If you want a few seconds of footage that never existed, that is generative video and a different category again. If the text describes a product, a process or a set of numbers, you want motion graphics — the subject drawn rather than a person describing it.

Is there a free text to video generator?

There are free tiers, and they share the same three limits: a watermark, a length cap, or a count cap — usually two of the three. Synthesia’s free Basic tier is about 10 minutes a month of credits, HeyGen’s is three videos a month at up to a minute each and watermarked, Colossyan’s Starter is free with 20 minutes a month. If the video is going on a homepage or into an ad, none of them apply, because the watermark is the product.

Can a text to video generator use my own script?

Every category accepts a script, but they do different things with it. Avatar tools read it out verbatim, so the script is the whole input. Stock tools chop it into sentences and search a library for each one, which means the writing controls the pictures only indirectly. Generative tools treat it as a prompt and largely rewrite what you asked for. Motion graphics tools sit in between: the script sets the narration and the narration sets the timing, so a rewritten line moves everything after it.

What is the difference between text to video and AI video generation?

“AI video generation” usually means the generative category specifically — a model inventing footage frame by frame from a description, as Sora, Runway and Veo do. “Text to video” is the older and broader phrase and includes the three assembly-based categories, which do not invent footage at all: they perform it, retrieve it, or draw it. The distinction matters because the generative category is the only one where you cannot control what appears.

How long does a text to video generator take?

Minutes, in every category, which is the whole point of the format. The honest number is not the render time but the number of attempts: a tool that returns a video in ninety seconds but needs six goes to get the pictures right has cost you an afternoon. Ask how much of the output is determined by your input, because that is what decides how many attempts it takes.

How these prices were checked

No price on this page was re-read for it. Every figure is carried from the two articles on this site that did read them off the vendor’s own pricing page — the software comparison on 6 September 2026 and the Synthesia piece on 10 September 2026. Restating a number from memory is how one site ends up quoting two different prices for one plan, so the figures have exactly one home and this page borrows them.

Generative video is deliberately left without a price. Those tools meter by credit against clip length and resolution, and any single monthly figure would misrepresent them. These plans change often: if you are reading well after 13 September 2026, check the one you are about to buy.

Related: explainer video styles once you know it is motion graphics you want, corporate explainer videos if this is for training or internal comms, fifteen examples worth studying for what good looks like, and the explainer video generator itself if you would rather just try one.

← All articles