- 5 to 15 seconds
Clip length
- 480p, 768p, 1080p, 2K
Resolution
- Text, first frame, references
Modes
- Generated with every clip
Audio
- #6 on Artificial Analysis, Nov 2026
Text-to-video rank
About HeyGen Video
HeyGen Video is HeyGen's general-purpose video generation model, built on MiniMax H3 and post-trained by HeyGen. Masonry runs it through three fal endpoints. Text-to-video turns a written brief into a clip. Image-to-video uses your image as the first frame. Reference-to-video builds a new scene that keeps the products, people, or places from your reference images, so a real item can appear in a shot that was never filmed. Every clip comes with sound generated alongside the picture, and fal describes each output as having its own sound rather than offering an audio switch.
The range of output sizes is what sets it apart for business teams. Clips run any whole-second length from 5 to 15 seconds, and resolution goes from 480p for quick drafts through 768p and 1080p to 2K for a sharper final. Text-to-video frames cover 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16, and reference-to-video adds an adaptive setting that follows your references. On the Artificial Analysis video leaderboards in November 2026, HeyGen Video ranked sixth for text-to-video with an Elo of 1153 and fifth for image-to-video with 1123, just ahead of the MiniMax H3 Max route Masonry also offers, though inside the boards' confidence ranges.
Prompts behind these HeyGen Video videos
Actual prompts from HeyGen Video renders by the Masonry community. Copy any prompt, then remix it into your own creation.
Why teams choose HeyGen Video
Choose HeyGen Video for short, sound-on clips where a reference has to carry a real product or person into a new scene, especially when you want a 2K render at the end of a cheap 480p exploration. It is the closest alternative to MiniMax H3 Max, since both come from the H3 family, and on the Artificial Analysis boards HeyGen's post-trained version scored slightly higher for both text-to-video and image-to-video. For clips longer than 15 seconds, or for editing footage you already have, use Kling 4.0, and for 4K output compare Gemini Omni Flash 1.1, keeping in mind that Google lists its 1080p and 4K as upscaled. As with any generative video, judge the model on your own product: watch the full clip and check labels and shapes frame by frame.
What HeyGen Video can do
The capabilities that set HeyGen Video apart and earn its place in a brief
Built on MiniMax H3, Post-Trained by HeyGen
HeyGen Video starts from MiniMax's H3 video model and adds HeyGen's own post-training. It is a scene and product video model, not an avatar tool, so it generates whole shots rather than a talking head over a template.
Reference-to-Video for Real Products
Supply reference images of a product, a person, or a location and the model builds a new scene that keeps them. fal's endpoint takes up to nine reference images, and also accepts reference video and audio clips.
Sound in Every Clip
Ambience, effects, and speech are generated with the picture in the same pass. There is no audio toggle, so describe what should be heard, or mute the track if you plan to add a voiceover.
480p to 2K Output
Draft cheaply at 480p or 768p, then render the keeper at 1080p or 2K. 2K gives more headroom for cropping a landscape master into other placements.
Clips From 5 to 15 Seconds
Choose any whole-second duration from 5 to 15 seconds, enough for a social hook, a product reveal, or a short scene with a line of dialogue.
Six Frames Plus Adaptive
Generate in 21:9, 16:9, 4:3, 1:1, 3:4, or 9:16. Reference-to-video can also adapt the frame to your references, and image-to-video follows the shape of your first frame.
Where teams reach for HeyGen Video
- Product scenes without a shootPut an approved product photo into a kitchen, a gym bag, or a desk setup and get a short clip with matching ambient sound.
- Vertical social adsGenerate 9:16 clips for Reels, TikTok, and Shorts with sound already in the file, then add the offer and CTA as overlays.
- Animating a key visualUse a campaign image as the first frame so the clip opens on artwork your team has already approved.
- Draft-to-final workflowsExplore directions at 480p, pick a winner, and rerender it at 1080p or 2K for delivery.
- Location and lifestyle B-rollBuild short establishing shots for landing pages, product pages, and presentations, in ultrawide 21:9 when the placement calls for it.
- Spokesperson momentsCombine a reference photo of a consenting presenter with a product reference for a short line to camera, then review the speech before it ships.
Frequently asked questions
What teams need to know about creating with HeyGen Video in Masonry
What is HeyGen Video?
HeyGen Video is HeyGen's video generation model, built on MiniMax H3 and post-trained by HeyGen. In Masonry it makes 5 to 15 second clips with sound from a text prompt, from an image used as the first frame, or from reference images of products, people, or places.
Is HeyGen Video an avatar model?
No. HeyGen is known for avatar and translation tools, but HeyGen Video is a general text-, image-, and reference-to-video model. It generates complete scenes, which can include a person speaking, rather than animating a presenter over a template.
How is HeyGen Video related to MiniMax H3?
fal describes HeyGen Video as built on MiniMax H3 and post-trained by HeyGen. On the Artificial Analysis leaderboards in November 2026, HeyGen Video scored 1153 for text-to-video and 1123 for image-to-video, against 1148 and 1116 for the MiniMax H3 Max route, which is within the boards' confidence ranges.
How long can a HeyGen Video clip be?
Any whole-second length from 5 to 15 seconds in every mode, with 5 seconds as the default on fal.
What resolutions does HeyGen Video support?
480p, 768p, 1080p, and 2K. Use 480p or 768p for fast, low-cost drafts and 1080p or 2K for the version you deliver.
Which aspect ratios does HeyGen Video support?
Text-to-video supports 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16. Reference-to-video supports the same six plus an adaptive setting that follows your references. Image-to-video follows the shape of the first frame you supply.
Does HeyGen Video generate sound?
Yes. Every clip is generated with its own sound, and there is no setting to turn it off. Describe the ambience, effects, or dialogue you want, or mute the clip in your editor before adding a voiceover.
How many reference images can HeyGen Video use?
fal's reference-to-video endpoint accepts up to nine reference images, plus up to three reference videos and three audio clips. Tell the model in the prompt what each reference is and what it should do in the scene.
Can HeyGen Video animate my own image?
Yes. Image-to-video uses your image as the first frame, so an approved product shot or campaign visual opens the clip. Describe the motion and the sound rather than what the image already shows.
How did HeyGen Video rank on Artificial Analysis?
In November 2026 it was sixth on the Artificial Analysis text-to-video leaderboard with an Elo of 1153 and fifth on the image-to-video leaderboard with 1123. Several neighboring models were within the same confidence range, so test it on your own brief before committing.
How much does HeyGen Video cost in Masonry?
Masonry bills each clip in credits by resolution and length, so 1080p and 2K renders cost more than 480p and 768p drafts. The editor shows the exact credit cost before you generate.
Can HeyGen Video output be used commercially?
fal marks all three HeyGen Video endpoints for commercial use. Review the current fal and HeyGen terms, and get consent before using a real person's likeness or voice in a reference.
How does my team use HeyGen Video in Masonry?
Open a Masonry canvas, pick HeyGen Video from the model selector, and describe the video you need: a product shot, an ad creative, a social post. Masonry generates it, then you refine, edit, and combine HeyGen Video with other models in one workspace.
Is HeyGen Video free to try?
New Masonry accounts get 50 free credits to try GPT Image 2.5 Flare. HeyGen Video is included in every paid plan, and the editor shows the exact credit cost before you generate.
How do I write good prompts for HeyGen Video?
Write one scene per clip: the subject, what happens, the camera move, the setting, and the sound. For reference-to-video, say what each reference is, such as "the bottle from the first image" or "the woman from the second image", and what it should do in the scene. For image-to-video, describe the motion from the first frame rather than re-describing the image. Because sound is always generated, name the ambience you want, quote any line of dialogue, or ask for "no speech" when the clip will carry a voiceover. See the prompt gallery on this page for real HeyGen Video prompts you can copy and adapt.
Who makes HeyGen Video?
HeyGen Video is built by HeyGen. Inside Masonry it runs alongside 50+ image and video models, so your team can pick the right one for each brief without switching tools.
Can I see examples made with HeyGen Video?
Yes, the prompt gallery on this page shows real videos teams have generated with HeyGen Video in Masonry, each paired with the exact prompt you can copy and adapt for your own brand.
Start creating with HeyGen Video
Generate, edit, and compare across 50+ models in one workspace.
Guides for HeyGen Video
Prompt walkthroughs and examples from the Masonry blog
Explore more AI models
Compare HeyGen Video with other models teams run in Masonry