Creatify Boreal AI Video Generator
Boreal is Creatify Labs' advertising video model. One prompt returns a finished MP4 with the picture and its dialogue, ambience, music and sound effects already in sync, anywhere from 1 to 20 seconds.
Text to Video
Made with Boreal
Four clips generated on Veevid, no editing and no post-production. Turn the sound on: every one came back with its audio already in place.
A UGC ad from one prompt
8sSelfie framing, a spoken line and room tone, generated together. The prompt that made this clip is printed in full further down the page.
A spokesperson who hits the script
8sThe line written in [SPEECH] is the line you hear. Office ambience underneath, no music, locked-off camera.
Sound design without dialogue
8sLeave [SPEECH] empty and Boreal scores the shot instead: beans into the hopper, a burr grinder, bass underneath.
Ten seconds, still one pass
10sLonger reads stay in sync. Duration is any whole number of seconds up to 20, not a set of fixed tiers.
How to Write a Boreal Prompt
Boreal reads a prompt split into four labelled sections. Most models take one block of text and have to guess which parts are camera direction and which are dialogue. Boreal does not have to guess, and that is the single biggest lever you have over the result.
[VISUAL]What the camera sees and does: subject, setting, framing, lighting and the camera move. This is the only section that describes pictures, so every visual instruction belongs here.
[SPEECH]The exact words spoken on camera, in quotation marks. Boreal generates the voice and the mouth movement from this text, so write the line you want to hear rather than a description of it.
[SOUNDS]Everything you hear that is not the spoken line: room tone, foley, ambience and music. It is generated in the same pass as the picture, so it lands in sync without a separate audio step.
[TEXT]On-screen text such as a caption, a price or an end card. Small lettering is the model's weakest area, so keep it short or state that you want none.
Three prompts you can copy
Each of these produced the clip beside it, generated on Veevid at 720p. Copy one, change the product, and run it.
UGC testimonial, 9:16
9:16 · 8s · 720p · 16 credits
[VISUAL] Selfie-style handheld shot of a woman in her late twenties sitting on a sofa in a sunlit living room, talking straight to the camera. Around the third second she raises a small matte-white skincare bottle to chest height in her right hand, and it stays in that same right hand, fully in frame, for the rest of the shot. Natural window light, slight handheld sway, shallow depth of field. [SPEECH] "Okay, three weeks in. I stopped using everything else, and this is the only thing that stuck." [SOUNDS] Quiet room tone, faint street noise through the window, no music. [TEXT] None.
The spoken line runs about 16 words across 8 seconds, which is a comfortable pace. Two details worth copying: the bottle's label comes back as illegible squiggles, which is the small-text limitation in action and why the prompt never asks for a brand name on the product; and the prompt names which hand holds the bottle and says it stays there. Without that sentence the first run had the bottle jump between hands halfway through — naming the hand is the cheapest fix for prop continuity.
Mistakes that cost you a render
Putting dialogue inside [VISUAL]
The visual section drives the picture, so a line written there tends to be acted out rather than spoken. Move every spoken word into [SPEECH].
Describing the speech instead of writing it
'She explains the benefits' is a description, and Boreal may read it aloud word for word. Write the actual sentence you want to hear.
Packing a 5 second clip with three separate actions
Boreal is strongest on short clips and focused scenes. Keep it to one action and one camera move. If you need three beats, generate three clips and cut them together.
Asking for a logo or small print in [TEXT]
Logos, fine lettering and app interfaces drift between runs. Keep on-screen text to a few large words, or write 'None.' and add the real text in your editor afterwards.
Writing the prompt in another language
English is the only language Creatify has validated. Others are untested rather than blocked. The generator has a Translate button that converts your prompt to English before you spend credits.
What a Boreal Video Costs
Boreal is billed per second of finished video. Length and resolution are the only two things that move the price: reference images and driving audio are not charged extra, because all four ways of using the model run through one endpoint at one rate.
720p
2credits/s
20 credits for 10s
1080p
6credits/s
60 credits for 10s
2K
24credits/s
240 credits for 10s
Against the other native-audio models on Veevid
| Model | Tier | Credits/s | 10 seconds |
|---|---|---|---|
| Boreal | 720p | 2 | 20 |
| Grok Imagine | 720p | 3 | 30 |
| MiniMax H3 | 768p | 8 | 80 |
| MiniMax H3 Max | 768p | 16 | 160 |
| Wan 3.0 | 720p | 16 | 160 |
| Kling 3.0 Standard | audio on | 25 | 250 |
| Seedance 2.5 | 720p | 63 | 630 |
Every rate above is the per-second price with audio at each model's 720p-class tier, read from the same pricing code the generator charges from. One caveat worth stating plainly: Grok Imagine's 480p tier is 1.6 credits per second, which is below Boreal's 720p rate. Boreal is the cheapest per second at 720p, not the cheapest option on the site outright.
What People Use Boreal For
Creatify post-trained it on advertising footage, so it is pointed at three jobs rather than general filmmaking.
UGC-style ads
Creator-style clips shot on an implied phone camera, with a spoken line and honest room tone. The format that performs on paid social, generated in the vertical ratio it will actually run in.
- At 2 credits a second, ten 8-second hook variants cost 160 credits — less than one 8-second clip on Seedance 2.5.
- 9:16 and 1:1 are first-class ratios, not centre-crops of a landscape render.
- No creator booking, no studio time and no separate voiceover session.
Product videos
Animate a product photo, or describe the shot from scratch. The sound pass carries the foley that makes a product clip feel filmed rather than rendered.
- Upload a product photo and Boreal animates it, following the image for composition and appearance.
- Leave [SPEECH] empty and the audio pass goes to foley and music instead of a voice.
- 720p for volume testing, 1080p or 2K for the cut you actually ship.
Presenter and spokesperson clips
A person delivering a script to camera, for explainers, announcements and landing page intros. Write the line in [SPEECH] and Boreal speaks it.
- The words in [SPEECH] are the words you hear, so the script is the deliverable.
- Upload a narration track and Boreal keeps your audio instead of generating a voice.
- Up to 20 seconds, which covers most social slots and ad units.
Where Boreal Falls Down
Worth reading before you spend credits. These come from Creatify's own model documentation and its published evaluation, not from our marketing copy.
Logos, small text and hands drift
Reference images guide appearance and composition, but exact identity, logos, fine lettering, hands and complex interactions can all vary between runs. Do not promise anyone that a product label will come back readable.
Check the lip sync before you publish
Creatify's documentation asks you to review generated dialogue and lip sync before publishing. Treat every talking clip as a draft that needs one human viewing, not a finished asset.
Single-person talking clips are its weakest case
In Creatify's own blind evaluation, a person talking straight to camera is where Boreal gains least over its untrained base model. It still works, but this is the shot type most likely to need a second take.
Short and simple beats long and busy
Creatify states the model is strongest on short clips and focused scenes. A 20 second shot with several moving elements is the hardest thing you can ask of it, so test before you commit a campaign to it.
English is the only validated language
The prompt schema names English as the validated language. Others are untested rather than blocked, so results are less predictable. The generator can translate your prompt before it runs.
Every output is marked as AI-generated
Boreal always writes C2PA provenance metadata and an imperceptible watermark into the MP4. Neither is visible in the picture, but both are detectable, which matters when a platform or a client asks how a clip was made.
How to Generate a Video with Boreal
Three steps from a written brief to a finished clip with sound.
Write the four sections
Split your brief into [VISUAL], [SPEECH], [SOUNDS] and [TEXT]. Add a reference image if you are animating a product or a person, or a driving audio track if you already have the narration.
Set length, ratio and resolution
Pick any whole number of seconds from 1 to 20, a ratio from auto through 9:16, and 720p, 1080p or 2K. The credit cost updates as you change them.
Generate and check the audio
At 2 credits a second you can afford several takes. Play each one with the sound on and check the lip sync before you publish.
Creatify Boreal FAQ
The questions worth answering before you spend credits.
Where to Use Boreal
The model runs in two generation modes on Veevid, and two nearby models are worth comparing against.
Render your first Boreal ad
Write the four sections, pick a length, and get back a clip with its dialogue and sound already in place. A 10 second 720p take costs 20 credits.
Try Boreal