Avatar AI (Photo AI) is an AI photo generation tool created by Dutch indie developer Pieter Levels (@levelsio), focused on generating realistic human photos and videos through AI. The tool originated from the AI avatar trend that went viral in 2022, and currently offers over 120 unique avatar styles, covering a variety of creative looks from futuristic designs to artistic interpretations.
[Uses] Users can use Avatar AI to generate their own AI avatars, personal portraits, and videos for scenarios such as social media avatars, gaming profiles, and Discord servers; they can also generate model product photos for commercial use, for example, Shopify store owners showcasing clothing effects; they can also create fully synthetic AI influencers. According to the official introduction, the tool has cumulatively generated over 30,926,097 photos.
[Main Capabilities]
- AI model training: Upload 5 to 15 selfies taken in different locations, scenes, and times, and click "Create a new AI model" to start training, which takes about 5 minutes.
- Photo generation: After training the model, users can generate unlimited photos resembling themselves. After the user describes a scene (such as a 1920s Paris cafe or a moonlit beach in Bali), a realistic photo can be generated in about 20 seconds.
- AI video: Convert generated AI photos into videos, and can also add lines to make the AI model speak.
- Motion capture video: Upload a motion video (such as dancing, walking, gestures, or speaking), then add an AI model photo, and the AI will transfer the motion onto the photo, generating a video of the AI character performing the same action.
- Outfit change feature: Screenshot or save a desired outfit (such as from Shein or Zara), paste it into Photo AI and click "Try on clothes," and the model can wear that outfit; the new version supports patterns and prints.
- AI influencer creation: Design a synthetic person from scratch, choosing attributes such as source, body type, hairstyle, and eye color; after generation, users can take photos, generate videos, or even make them speak.
- Multi-person support: Couple photos can be created by selecting the "Couple" type and ensuring both people are present in each uploaded photo; two photos can also be merged into one AI-generated image, preserving the identities and styles of the people.
- Video editing: Provides an AI-first video editor that can generate and regenerate videos in the browser, supporting multi-track timeline editing, dragging clips, stacking video and audio tracks, muting, disabling, speeding up clips, and exporting the finished video.
[How to Use] Users need to first register an account, upload selfies to train an AI model, and then generate photos by entering text descriptions of scenes or selecting photo packs (Packs). After generation, they can [the text ends here]Perform "Upscale" to improve clarity. It supports operation on mobile phones or computers. To cancel a subscription, you can log in and click the logo in the upper left corner, then go to the "Billing" page and select "Cancel Subscription"; if you cannot log in, you can use the Stripe billing portal to enter your registered email to manage the subscription; iOS app subscriptions are managed through Apple. Regarding pricing, the official statement mentions that the annual plan is more than 50% cheaper than the monthly plan, equivalent to getting more than 6 months free, and Photo AI has no coupon codes. Refund policy: self-service refunds are only available if no model has been created, the account has generated fewer than 20 photos in total, and the user did not register through someone else's referral link; users who have already created a model or generated photos or videos cannot get a refund.
[Use cases]
- Personal social media avatars and profile photos
- Couple photos, holiday-themed photos (such as Valentine's Day packs)
- Gaming and Discord avatars
- E-commerce product model photos (such as Shopify stores)
- AI influencer and virtual avatar creation
- Short video and motion capture content production
[Limitations]
- Training an AI model requires uploading 5-15 selfies, and couple models require every photo to include both people.
- Refunds have strict conditions, and users who have already used the generation feature are not supported for refunds.
- Some features such as 10-30 second long videos and background music are not yet available, and the official source says they will be added soon.
- Video and photo generation depends on GPU resources and may incur higher costs, so refunds are not supported for users who have already used it.
- Regarding specific pricing amounts, available platforms (other than iOS and web), and other information, the materials do not currently provide it.
