8 short, free text lessons on how to use each Klaket tool and which AI engine to choose for what. You can read them without an account. Each lesson ends with a button to the tool.
350 free credits, with access to every tool. Plans from $19/month, in US dollars. Pay by card.
01
Short text lessons
No enrolment and no account: read the one you need, in any order.
02
Which engine for what
Compare the video engines by length, speech, real faces, credits and strength.
03
The studio's own numbers
The lengths and credits in these lessons are the ones the app uses.
Lesson 1
Your first video from a product photo
All you need is a product photo and a sentence saying what you want to see. Klaket handles the script, the scenes, the voice and the captions.
Pick 1 to 3 sharp, well-lit photos with a readable label and logo: one product per photo, whole and front-on, on a clean background.
Avoid photos with text on top, collages or several variants together.
In the product description, give the benefits, who it's for, the price and the offer, without promising what you can't deliver.
In Create video, write your request and choose the quality and the length, from 4 to 30 s. If you're not sure, leave the engines on Automatic.
Turn on “I want to see the images before the video is made” to see the scene images before Klaket makes the video. With it off, Klaket does everything on its own and the video appears in Videos.
Before generating, the plan shows the engines and the total in credits. If the generation fails, the credits come back on their own.
Each engine has a strength and a price. On Automatic, Klaket picks from your request: whether there's a person, whether they talk to camera, the language and the length. On Manual, you pick the engine and see the total right away.
Length: up to 15 s fits almost every engine. Over 15 s, up to 30 s, only Seedance 2.5 and Wan 3.0. Each video is one continuous take.
A person talking to camera: in any language, Seedance 2.0 and Seedance 2.5; English only, Kling 3 and Wan 3.0. Seedance 1.5 Pro, Grok Imagine, and PixVerse v6 have no speech of their own: use a voice-over or no voice.
The exact face of a persona, an influencer or your own: Kling 3, Wan 3.0, Seedance 1.5 Pro, Grok Imagine, and PixVerse v6. Seedance 2.0 and Seedance 2.5 don't accept photos with real faces: the person is created from the description.
Lots of videos at low cost: Seedance 1.5 Pro, Grok Imagine, and PixVerse v6. A 10-second video with voice-over costs from 249 credits.
Quality: Standard (720p) costs less and is good for testing ideas; High (1080p) is for the final version and paid ads.
The video engines, one by one
Seedance 2.0
Recommended
ByteDance
Choose it when
You want cinematic footage and a person talking to camera in a language other than English, up to 15 s.
Up to 15 s
Speaks any language
No real faces
Upscaled 1080p
Cost
48 cr/s at 720p · 54 cr/s at 1080p
10-second video with voice-over: 639 credits
Seedance 1.5 Pro
ByteDance
Choose it when
You want lots of voice-over videos at the lowest cost, up to 12 s.
Up to 12 s
No built-in speech
Accepts real faces
Native 1080p
Cost
9 cr/s at 720p · 19 cr/s at 1080p
10-second video with voice-over: 249 credits
Grok Imagine
xAI
Choose it when
You want voice-over videos at a low cost, up to 15 s.
Up to 15 s
No built-in speech
Accepts real faces
Upscaled 1080p
Cost
11 cr/s at 720p · 16 cr/s at 1080p
10-second video with voice-over: 269 credits
PixVerse v6
PixVerse
Choose it when
You want volume with a sharper image and native 1080p, up to 15 s.
Up to 15 s
No built-in speech
Accepts real faces
Native 1080p
Cost
14 cr/s at 720p · 28 cr/s at 1080p
10-second video with voice-over: 299 credits
Kling 3
Kuaishou
Choose it when
You want the exact face of a persona or influencer and the product true to your photos, up to 15 s. It's what Automatic uses for most short videos.
Up to 15 s
Speaks English only
Accepts real faces
Native 1080p
Cost
26 cr/s at 720p · 35 cr/s at 1080p
10-second video with voice-over: 419 credits
Wan 3.0
Alibaba
Choose it when
You need a long video (up to 30 s) at a good price, with the persona's face; built-in speech in English only.
Up to 30 s
Speaks English only
Accepts real faces
Native 1080p
Cost
32 cr/s at 720p · 63 cr/s at 1080p
10-second video with voice-over: 479 credits
Seedance 2.5
ByteDance
Choose it when
You need more than 15 s (up to 30 s) with cinematic footage and speech in any language. It's the most expensive: save it for the final cut.
Up to 30 s
Speaks any language
No real faces
Upscaled 1080p
Cost
133 cr/s at 720p · 139 cr/s at 1080p
10-second video with voice-over: 1,489 credits
What Automatic picks
In Create video, with the engines on Automatic, Klaket picks like this. On Manual you can choose another one.
Request
Engine
Sound
Person talking to camera, in English · 10 s
Kling 3
talks to camera
Person talking to camera, in another language · 10 s
Seedance 2.0
talks to camera
Voice-over · 10 s
Kling 3
with voice-over
No voice · 10 s
Kling 3
no voice
Long video with voice-over · 30 s
Wan 3.0
with voice-over
Long video, person talking in another language · 30 s
In Create video, you write what you want and Klaket reads from your text whether there's a person, whether they speak and in what language, then writes the script from there.
Write as if to a director: who appears, where, what they do and what they should say at the end.
An example: “A girl shows the serum in the bathroom and says why she loves it.”
Say whether the person talks to camera or you prefer a voice-over, and in what language. In English you have more engines to choose from.
Write the offer and the price the way they should appear: Klaket uses them as written.
Each video is one take, with no cuts in the middle: ask for one clear scene rather than a lot at once.
Test the idea with a short video in Standard (720p) and make the long or High (1080p) version only of what works.
All three tools start from a video that already exists, but they change different things. Choose by what you want to change, and film with that in mind.
You want a different person in the video: Swap person.
You filmed the video and the person must stay untouched: Klaket Swap.
You want to change the outfit or the setting, or the video runs up to 30 s: Swap object.
Swap person
The person from a photo, real or AI-made, takes someone's place in the video, with the same movements, the same setting and the same sound.
A video of 3 to 30 s; on Max, from 5 s.
A sharp, front-facing photo with one person; for Motion, full body, because the outfit in the photo comes along too.
Dance, sport or fast movement: Motion, ideally with the person large in frame at least once. Person far from the camera or 1080p: Normal. Videos with cuts or up to 4 people: Max.
Use someone different from the person in the video: if they look very alike, the swap barely shows.
Minors, public figures and nudity are refused automatically.
Klaket Swap
Film a video on your phone and swap the car, the product in your hand or the colour of your clothes. The person and the background stay exactly as you filmed them, and so does the sound.
A video of 3 to 15 s, uploaded from your gallery, not a copy forwarded through WhatsApp.
One swap at a time, with a specific request: “swap the car for an orange Porsche 911”.
How you film matters most: see the guide below these blocks.
Swap object
Changes an object, the outfit or the setting in a video. The people, the movements and the sound stay the same.
A video of 3 to 30 s.
Say what goes out and what comes in: “swap the water bottle for the serum in image 1”.
Add up to 3 reference images (the product, the outfit, the place) and refer to them by number.
One swap at a time and a specific request give better results.
How to film your video for Klaket Swap
For the most realistic result
Klaket does the swap, but the result depends a lot on how you film. These steps make the difference, depending on the kind of video:
Car or large object
Keep the camera still: rest your phone on something or use a tripod.
The object is fully in frame and clearly visible.
Nothing in front of it: railings, nets or fences make the swap harder.
Make sure nobody else walks in front of the camera.
Daylight, no backlight.
Your hand on an object
Steering wheel, bottle, phone: touch it lightly with your fingertips instead of gripping it with your whole hand.
Keep your hand still.
Keep the object clearly visible, barely covered by your hand.
Story style (handheld camera)
Slow movements, no sudden turns.
Keep the object in frame the whole time.
Don't let your face cover the object.
For every video
3 to 15 seconds, in 1080p or 4K.
One swap at a time, with a specific request: "swap the car for an orange Porsche 911".
Upload the original from your gallery, not a copy forwarded on WhatsApp.
Pick the AI model, the length, the quality and the format, and write your prompt: video or image, with no steps in between. Your text goes to the model exactly as written.
Use it when you know exactly what you want to see. For an ad with a script written by Klaket, use Create video.
Follow the order: who or what → where → the action → style and lighting → camera movement. Up to 2,000 characters, in any language.
Be specific: “30 ml amber bottle on a marble counter” instead of “a nice product”. One action per video: requests with a lot going on come out worse.
For readable text in an image, put it in quotes and pick GPT Image 2.5. Grok Imagine and Seedream 5.0 Pro rewrite the prompt internally: the result can drift a little from what you wrote.
Upload a start image so the video begins exactly on it, and ask for up to 4 images at once to keep the best.
“Improve” (2 credits) rewrites the prompt with more detail, and you can change it before generating.
Everything is paid for in credits, and you see the cost before generating. The free plan gives you 350 credits when you sign up, with every tool and every AI engine; free-plan videos are watermarked.
Video from 175 credits; Freestyle from 14; a batch of posts, 40; a landing page, 130. Importing a product costs nothing.
Per second of video: Swap person from 24 credits, Swap object from 36 and Klaket Swap from 91.
If a generation fails, the credits come back on their own.
Plan credits renew every month and whatever is left doesn't carry over. Credits from extra packs have no expiry date and are only used after your plan credits.
To make them go further, test with short videos in Standard (720p) and make the long or High (1080p) version only of what works.
See and approve each scene before credits go into the video.
Automatic refunds
If a generation fails, your credits come back on their own.
Secure payment
Payments processed by Stripe. Cancel anytime.
Labelled as AI
The preview carries an AI label, as the EU AI Act requires.
Find my planTell us how much you want to make per month and we'll work out the content and the right plan.
Frequently asked questions
Are the courses really free?
Yes. The lessons are public: you don't need an account or a card to read them. To try the tools, create an account. The free plan gives you 350 credits when you sign up, with access to every tool and every AI engine (videos are watermarked). Paid plans start at $19/month, in US dollars, paid by card.
Do the lessons include videos?
No. They're text lessons, to read at your own pace and come back to when you need them.
I know nothing about AI. Which engine do I choose?
Leave the engines on Automatic: you write what you want and Klaket picks from your request. Before generating, you see the engines and the total in credits. Lesson 2 explains when it's worth choosing yourself.
How long are the videos, and which models make them?
Each video is one continuous take of 4 to 30 seconds, depending on the engine, in 720p or 1080p, made with Seedance 2.0, Seedance 1.5 Pro, Grok Imagine, PixVerse v6, Kling 3, Wan 3.0, and Seedance 2.5. The recommended model is Seedance 2.0. For something longer, make another video with the same product and the same person and join them in CapCut, Instagram or TikTok.
Can I use what I make in ads?
Yes. As long as you follow our Terms and Acceptable Use Policy, you can use the videos, posts and landing pages you create commercially: in ads, on social and in your store. Videos on the free plan carry the Klaket watermark. Because outputs are AI-generated, other people may get similar results.
What if a generation fails?
If it fails for a technical reason, the credits for that generation go back to your account automatically, whether it's a video, a batch of posts or a page. For videos, you see and approve the scene before the video is generated: each scene costs 107 credits, and the video is only charged once you approve it.
Which languages are supported?
US English, Spain Spanish, Latin American Spanish, Brazilian Portuguese, European Portuguese, French and German. Scripts, posts and pages are written from scratch in each language, and the person in the video can speak any of them, lip-synced.