AI Companion Photos, Videos and Voice
Understand DreamBunny's photo, video and voice features, how to describe a scene, and what to check about tokens and generated results.
Create your companionPlan a small media session
- Check which features your membership includes and the token balance available to you.
- Read the token requirement for the specific action. A still image, a voice message and different video lengths are not interchangeable.
- Generate one result first. For a photo, keep the character, setting and lighting instructions simple; for voice, try a short reply before a longer interaction.
- Review the result and remaining balance before generating again. Do not assume regenerating a result is free.
Describe the scene you want
A photo request works best with a clear scene: your character, an environment and a small number of visual details. For example, ask for a portrait in a sunlit cafe rather than combining several locations and poses. Character appearance provides a starting identity, but generated results can vary.
Choose photos, videos or voice deliberately
A still image, a moving scene and an audio message are different kinds of output. Use the relevant controls in the app and describe the requested action clearly. Availability depends on the feature and your account's access. Do not assume a photo request automatically includes video or voice.
Check tokens and review the result
Media actions can consume tokens, and costs differ by feature. Review the displayed cost before generating and compare token packs on the pricing page when you need a refill. AI media is fictional and can include visual or audio errors. Keep requests within the content rules and never present generated media as evidence of a real person's actions.
How to think about token costs
Membership access and token balance answer different questions: whether you can use a feature, and how many paid creative actions you can request. For planning, multiply the displayed cost per action by the number you want to try, then add room for any further paid attempts. Check video duration and whether a new source image is needed. A voice message is not the same product as a live call. The app's current action requirement and checkout terms take precedence; this guide does not promise an allowance, a fixed rate or a refund.
Evaluate the result before repeating
Look for identity drift, extra fingers, inconsistent backgrounds and unwanted text in pictures. Listen for pronunciation, language and pacing in audio. Change one instruction at a time. The public language selector translates the interface; it does not certify a particular accent or Cantonese voice. Do not upload someone else's image or voice without the necessary permission.