What is the best free talking photo AI?
For most people HeyGen's Free plan is the most usable free talking photo tool, with 3 videos a month of up to 1 minute each[1], but its videos carry a watermark[4]. Hedra suits short expressive or singing clips, and LivePortrait is the free open-source option if you can run code yourself[16].
How we judged them
A free talking photo tool is only useful if the free tier makes something you can post. We compared each one on four things, using only what each company states on its own pages as of 2026-10-18:
- Length: how long a free clip can be.
- Watermark: whether free downloads are marked.
- Sign-up: whether you need an account or a card.
- Commercial use: whether the free tier allows it, where the company says.
The free tiers compared
| Tool | Free tier | Watermark and use |
|---|---|---|
| HeyGen [1] | 3 videos a month, up to 1 minute each, no card needed | Logo watermark by default; removal needs a paid plan |
| Hedra [6] | Free plan, clips capped at 20 seconds (per its in-app plan copy) | All Free videos watermarked; no commercial use |
| Krea [9] | 100 compute units a day, no card; the Free plan card lists all lip-sync models | Watermark and licence not stated clearly on the plan cards |
| Vidnoz [10] | Free plan, up to 3 minutes per video at 720p | Talking photo needs an account to create and download |
| D-ID [17] | Free trial after creating an account | Trial videos get a full-screen watermark |
| LivePortrait [14] | Open-source code under the MIT License, plus a hosted demo | You run it yourself; no plan tiers |
HeyGen's watermark is per its help center[4]. Vidnoz's account requirement is from its talking-photo page[11]. D-ID's watermark rule is from its FAQ[13]. LivePortrait's licence file is MIT[16]; we could not verify the licence of every model it depends on.
Which one to pick
- You want a talking photo with a typed script and many languages: HeyGen. Its photo avatar page markets over 175 languages and dialects[5], and the 1-minute free cap is long enough for a real post[1].
- You want an expressive face or a singing clip: Hedra. It describes Character 3 as an audio-to-video model for talking and singing[8]. Keep free clips under 20 seconds.
- You want to compare several lip-sync models without paying: Krea's daily free units cover its lip-sync models on the Free plan[9].
- You can run Python or a hosted notebook and want no account at all: LivePortrait. Its README links a Hugging Face Space demo[15]; we could not verify that Space's usage limits.
- You need a free clip with no sign-up: we could not verify any talking-photo tool that renders and downloads without an account as of 2026-10-18. Vidnoz's image-to-video page says no login is needed[12], but that is its image-to-video tool, not its talking-photo tool.
What free tiers will not give you
Clean, commercial-ready files are the usual paid line. Hedra lists commercial use, no watermark and voice cloning as paid perks[7], and HeyGen lists watermark removal on its paid Creator and Pro plans[2]. If you plan to post for a brand or a client, budget for one month of a paid plan rather than fighting the free watermark. For the general case of avatar presenters rather than photos, see free AI avatar video generators.
Where Cast fits
- You want a short clip of a character who talks in a scene and you do not have a photo to start from: describe the character and pick one of four drawn looks.
- You want to try something free on a phone with no sign-up: Cast's first 5-second video needs no account, and Cast adds no visible watermark of its own.
Where it does not:
- Making your own photo talk. Cast cannot take an uploaded photo; characters are drawn from a text description.
- Lip sync to a voice recording you upload.
- Talking clips longer than 15 seconds.
People also ask
- Is there a talking photo AI with no watermark for free?
- Among the tools we checked, LivePortrait is the one you run yourself under an MIT code licence[16]. HeyGen, Hedra and D-ID all say their free or trial videos are watermarked[4][6][13].
- Which free talking photo AI makes the longest clips?
- Vidnoz states up to 3 minutes per video on its Free plan[10], and HeyGen allows up to 1 minute on its Free plan[1]. Hedra's Free plan is capped at 20 seconds per its in-app plan copy[6].
Sources
Claims about other companies come from their own pages, read on the date shown. Products and prices change; check the link for the current version.
- HeyGen: pricing (read )
- HeyGen: pricing (paid plan features) (read )
- HeyGen: pricing (Enterprise features) (read )
- HeyGen help: how to remove the HeyGen watermark (read )
- HeyGen: photo avatar (read )
- Hedra: plans (read )
- Hedra: plans (paid perks) (read )
- Hedra: models (read )
- Krea: pricing (read )
- Vidnoz: pricing (read )
- Vidnoz: talking head (read )
- Vidnoz: image to video (read )
- D-ID: FAQs (read )
- LivePortrait on GitHub (read )
- LivePortrait README: Hugging Face Space (read )
- LivePortrait: LICENSE (read )
- D-ID: home page (read )
Related answers
- How do I make a photo talk with AI?
- Is there a free AI avatar video generator?
- How does lip sync work in AI video generators?
- Is HeyGen free?
- What is Hedra AI and is it free?
- Can I use AI-generated videos commercially?
Part of the Cast answers library (Talking photos and avatars).
Updated