Quick Verdict
Candy AI now has several ways to make a fictional AI companion move rather than remain a static chat character. The most visible current options are Scene Video, image-to-video, and Live Action Beta v2.
On Candy AI character pages checked in September 2026, Scene Video generates a 5-second version of the current chat scene for 12 tokens. Other beta video interfaces have shown 5-second and 10-second request options with different token pricing, so there is not one permanent universal cost for every Candy AI video workflow.
Live Action works differently. It gives supported characters selectable actions and uses an XP progression system. Candy AI says users earn XP by chatting and unlock additional actions as they level up, while spam messages do not earn XP.
The best way to use Candy AI video is to keep each clip simple. One character, one clear action, one setting and one camera idea usually work better than trying to fit an entire multi-scene story into a few seconds.
What Is the Candy AI Video Generator?
The Candy AI video generator is part of Candy AI's wider adult companion platform.
Instead of exporting a character picture to a separate animation website, Candy AI lets users move from chat or a generated image into video inside the same ecosystem.
Current video-related experiences include:
- Scene Video
- Image-to-video
- Dedicated Video Mode in chat
- Short beta video requests
- Live Action
- XP-unlocked actions
- Character-specific action libraries
This makes video feel connected to the AI companion rather than like an unrelated creative tool.
Candy AI Scene Video
Scene Video is one of the clearest current video features. On supported character pages, Candy AI describes it with the idea:
Do not just read the scene, watch it happen.
The current interface offers a five-second video of the exact moment or scene around the character.
Scene Video, as of 10 Sep 2026
12 tokens for a 5-second clipVisible on multiple character pages. Because Candy AI is still developing video features, treat this as a current interface price rather than a forever-fixed rate.
What 100 tokens covers
8 videos with 4 tokens remainingRoughly 40 seconds of generated Scene Video output at the current format and cost.
How Scene Video fits into chat
Scene Video is useful because the prompt does not need to start from nothing. You already have:
- A character
- A conversation
- A location or scenario
- A current emotional tone
The video tool can use that context to visualise the moment.
For example, imagine your chat has established that the character is sitting in a quiet café during a rainstorm. Instead of writing a completely new video prompt from scratch, Scene Video can be used to animate that current moment. This creates better continuity between text and media.
How to generate a Candy AI Scene Video
-
Choose a supported character
Open a character that has video features available. Not every interface or account state necessarily exposes identical options.
-
Establish a scene in chat
Give the conversation a clear setting. For example: "We are sitting by the window in a small coffee shop while it rains outside."
-
Switch to video mode
Candy AI's current chat interface can warn you when you request video while still in normal mode. Use the dedicated video option when prompted.
-
Select Scene Video
On compatible characters, choose the Scene Video option.
-
Check the token cost
Current supported pages show 12 tokens for a five-second generation. Always verify the number in the live interface before clicking Generate.
-
Generate
Wait for the clip to finish.
-
Review the result
Check face consistency, motion, hands, camera movement, clothing and scene continuity. Do not automatically regenerate if one tiny detail is imperfect — decide whether the problem actually matters first.
Candy AI Image-to-Video
Candy AI also connects its image generator with video. On current Generate Image pages, a generated image can lead into an AI Video beta option.
This workflow is useful because the still image acts as a visual starting point. You already know:
- What the character looks like
- What they are wearing
- Where they are
- How the frame is composed
The video model then has a more focused job: animate the scene.
Still → motion workflow
-
Build the character first
Do not animate a character you have not finalised.
-
Generate a good still image
Check the face, clothing, pose and background.
-
Pick one movement
Examples: turn towards camera, smile, walk two steps forward, look out of the window, raise a coffee cup, move hair away from face.
-
Generate the short video
Use the available video control.
-
Retry only if you know what failed
Random regeneration wastes tokens.
Candy AI Video Prompt Formula
For short video, use:
She slowly turns towards the camera, smiles and takes two steps closer while the camera gently moves backwards, relaxed expression, natural pace, rooftop bar at night.
That gives the model one coherent sequence.
Why simple prompts work better
Video generation is more difficult than image generation. A still image only needs to solve one moment. A video needs to maintain identity, clothing, anatomy, background, lighting, motion, timing and camera continuity across multiple frames.
Every extra action increases the chance of something going wrong.
Much harder
"She stands, walks across the room, changes clothes, opens the door, runs outside, gets in a car and drives away"
Much better
"She stands up slowly and walks towards the door while looking back at the camera."
Short Candy AI clips benefit from short narrative goals.
Candy AI Live Action Beta v2
Live Action is different from a normal video generator. Current supported character pages show Live Action Beta v2.
Instead of writing a full motion prompt, users can choose from predefined actions. Examples visible on current pages include basic actions such as:
Smile for me
Blow a bubble
Characters can also have additional action sets. The exact number and type of available actions depend on the character and access level.
Candy AI Live Action XP
Live Action includes a progression system. Candy AI currently shows:
Character level
Displayed on supported Live Action pages.
XP progress
Earned by chatting with the AI companion.
Actions tied to progression
More actions can become available as your level increases.
Candy AI states this explicitly. It is an important design choice — it encourages actual interaction rather than users sending hundreds of meaningless messages just to unlock actions.
How to earn Candy AI XP
The public interface does not expose a universal mathematical formula for every XP gain, so avoid guides claiming "exactly X XP per sentence" unless Candy AI publishes that figure. The confirmed rule is simpler: chat naturally with your AI companion to progress. As your level increases, more Live Action options can become available.
Scene Video vs Live Action
They sound similar, but they serve different purposes.
| Feature | Scene Video | Live Action |
|---|---|---|
| Main purpose | Animate the current scene | Trigger character actions |
| Input | Scene/context and generation control | Predefined action selection |
| Current status | Video feature / beta workflows | Beta v2 |
| Cost model | Token-based | Depends on action/access |
| Progression | Not the core mechanic | XP and levels |
| Best for | Story moments | Interactive character reactions |
Scroll the table sideways to see all columns.
Use Scene Video when you want to visualise what is happening in the conversation. Use Live Action when you want to directly control a supported action.
Candy AI Video Length and Cost
Video length
Candy AI currently has more than one video interface. Scene Video prominently shows a five-second output. Other beta interfaces have shown choices including 5 seconds and 10 seconds, with token costs displayed before the request.
This is why I would not write "all Candy AI videos are five seconds" or "all Candy AI videos cost 12 tokens". The exact options depend on the workflow and current product version. Always check the live button.
How much does Candy AI video cost?
The strongest current public example is Scene Video: 12 tokens for 5 seconds. Other Candy AI video modes have displayed other prices.
Premium currently gives users 100 tokens per month. If you spent all 100 tokens only on 12-token Scene Videos, you could generate 8 complete videos with four tokens remaining. That would equal roughly 40 seconds of generated Scene Video output.
This simple calculation shows why video users need to understand the token system.
Candy AI Premium price for video users
| Billing term | Price |
|---|---|
| Monthly | $13.99/month |
| 3 months | $8.99/month equivalent, $26.97 billed |
| 12 months | $3.99/month equivalent, $47.88 billed |
The annual plan is advertised as 70% off. Premium unlocks AI video functionality and includes 100 monthly tokens.
However, the subscription does not make video unlimited. A heavy video user can need more tokens beyond the included allowance.
See full Candy AI pricingHow to Get More Value From 100 Tokens
The worst approach is random generation. The better approach is a staged workflow.
-
Stage 1: Chat
Develop the scene using unlimited Premium text messages.
-
Stage 2: Image
Generate the character in the correct setting.
-
Stage 3: Fix the still frame
Do not move forward while the visual identity is wrong.
-
Stage 4: Animate
Choose one short movement.
-
Stage 5: Regenerate only with a reason
Know exactly what you want to improve.
This makes every video token more productive.
Video and phone calls compete for the same balance
Video is not the only feature competing for your token balance. Candy AI's AI Phone Call currently shows beta pricing of 3 tokens per minute.
A user who loves both video and calls is much more likely to need top-ups.
Best Candy AI Video Prompts
For realistic characters
She looks up from her coffee, smiles at the camera and gently moves the cup to the side, natural motion, soft afternoon window light.
She walks slowly towards the camera on a rain-covered city street while neon lights reflect on the pavement, relaxed confident expression.
She turns from the balcony view to face the camera, hair moving slightly in the breeze, calm smile, sunrise light.
She sits back on the sofa, crosses one leg naturally and looks towards the camera with a playful expression, warm evening light.
Each prompt has one primary movement.
For anime characters
She taps the arcade button, looks over her shoulder towards the camera and grins, energetic anime motion, neon arcade lights.
She turns towards the city skyline, then looks back at the camera as her hair moves in the wind, slow cinematic anime movement.
She lifts her mug slightly, smiles and leans closer to the table, soft anime animation, warm café lighting.
Again, simple action wins.
How to Improve Consistency in Video
Start from a good image
Image-to-video has a stronger foundation when the still frame already looks correct.
Avoid extreme camera motion
Large spins and sudden perspective changes make identity harder to maintain.
Keep the face visible
Heavy obstruction gives the model fewer stable facial cues.
Do not change appearance mid-clip
Use separate generations for major outfit or hair changes.
Use one character first
Master single-character scenes before attempting complicated multi-person sequences.
Common Candy AI Video Problems
The face changes
Reduce extreme head turns and camera movement. Start from a stable image when possible.
Hands look strange
Simplify hand actions and avoid detailed finger movements when they are not essential.
Background warps
Use a simpler scene with fewer moving objects.
Motion is too fast
Add words such as "slowly", "gently" or "natural pace".
Character does too much
Remove secondary actions.
Clip feels static
Add one explicit movement and one subtle camera direction.
Result does not match the chat
Restate the key scene detail before generating or use a still image that reflects the intended setting.
Video mode error
Candy AI can warn users that a request containing "video" needs to be made in video mode. Switch modes rather than repeatedly sending the same text request.
Is Candy AI Video Free? And Privacy
Candy AI is free to start, but its current Premium sales pages list video generation as a subscription benefit, and token-based video costs apply within the paid ecosystem.
Do not assume that free signup means unlimited free AI video. Use free access to inspect the interface, then check the live Premium and token requirements attached to your account.
Candy AI video privacy
Candy AI's privacy policy says user inputs and generated outputs can be processed as part of providing the service. Its moderation policies apply to video as well as text and images.
Child exploitation, content resembling minors, non-consensual intimate imagery, harmful real-person deepfakes, deceptive impersonation, and certain illegal and harmful content. The safest approach is to create fictional adult characters rather than trying to use Candy AI as a real-person deepfake tool.
Is Candy AI Live Action Worth Using?
Live Action is worth trying if you like interaction and progression. Its appeal is not only visual quality.
The level and XP system gives users a reason to keep talking to the same companion and access more actions over time.
If you only want a standard prompt-to-video generator with no relationship mechanics, Live Action may feel less important.
Candy AI is a companion platform first. The feature is designed around that identity.
Best Candy AI Video Workflow
My recommended process is:
- Choose or create one character.
- Build a short scene through chat.
- Generate a still image if visual precision matters.
- Correct the image before video.
- Pick one clear action.
- Check the token price.
- Generate a short clip.
- Review face, motion and background.
- Regenerate only if you can name the problem.
- Use Live Action for quick interactive movements that fit the character.
This saves tokens and produces more coherent media.
Candy AI Video Generator FAQ
Yes. Candy AI supports Scene Video, image-to-video and other beta video workflows.
The current Scene Video option on supported pages generates a five-second clip.
Current supported character pages show 12 tokens for a 5-second Scene Video.
Other beta video interfaces have shown 5-second and 10-second options. Availability can differ by workflow.
Live Action is an interactive feature where supported characters can perform selectable actions.
It is the current beta version visible on supported character pages, with XP, levels and action unlocking.
Candy AI says users earn XP by chatting naturally with their AI companion. Spam messages do not earn XP.
Yes. Candy AI's image-generation interface includes image-to-video functionality on supported content.
Yes. Video generation is token-based in current workflows.
Premium currently includes 100 tokens per month.
No. Premium unlocks video access, but generation uses tokens.
Final Verdict
Candy AI's video tools make the most sense when viewed as extensions of the companion experience.
Scene Video
Turns a conversation into motion.
Image-to-video
Animates a still frame you already like.
Live Action
Gives you direct character actions and progression through XP.
None of these tools should be treated as unlimited professional video generation. They are short-form, character-focused features designed to make an AI companion feel more active and visual.
The main limitation is cost control. With 100 included monthly tokens and a current Scene Video example of 12 tokens for five seconds, careless regeneration can empty the allowance quickly.
Write the scene in chat, fix it as an image, then spend video tokens only when the idea is ready to move.
Make your companion move
Try Candy AI Video
Scene Video, image-to-video and Live Action all sit inside the same companion platform. Premium includes 100 tokens per month. Adults 18+ only.
Affiliate link. We may earn a commission at no extra cost to you. Token costs checked 10 September 2026 and can change.
More Candy AI Guides
Official Sources Used for Verification
- https://candy.ai/generate-image
- https://candy.ai/ai-girlfriend/olivia-carter/live-actions
- https://candy.ai/ai-girlfriend/nia-caldwell
- https://candy.ai/subscriptions
- https://candy.ai/blocked-content-policy