Elucidator｜AI Visual Assistant

Elucidator uses on-device AI to describe live camera views in natural text and speech, even offline. See scenes, objects, positions and readable text on screen or hear them aloud. Use compatible webcams, RTSP cameras and AI profiles. Choose Apple Foundation Models, downloaded local models or external AI services for image analysis. ---------- ■ Analyze What Your Camera Sees Analyze at regular intervals, manually or with Motion detection. Results appear over the camera view. Stream text as it is generated, and preview text size and display settings. Haptics signal new results. External displays show “Text only” or “Image & text”. ---------- ■ Hear Descriptions Aloud Turn speech on or off and adjust the voice and rate. Highlight spoken text to follow along. ---------- ■ Set Up AI Profiles Choose from eight built-in AI profiles to focus on different aspects of the same scene. Save and edit prompts, triggers, intervals and image counts in reusable AI profiles. Compare up to three images for changes. ---------- ■ Use Your Own Knowledge with RAG Add reference images, names and notes to your Knowledge library and link them to an AI profile. On-device image search finds matches and passes their details to the AI. RAG (retrieval-augmented generation) adds your information to image descriptions. ---------- ■ Choose Your Camera and Analysis Area Analyze the full view or adjust the visible area with features unlocked. Select “Circle to analyze” in an AI profile and outline an area with your finger. Switch front/rear cameras, use macro on supported devices, mirror the front camera, pinch to zoom, tap to focus or lock focus and exposure. Unlock features to use UVC USB webcams on compatible iPads or RTSP IP cameras. Save and select connections. You can also select a video from Photos or Files for silent, looping analysis. ---------- ■ Review Recent Results View the last 50 results with images, times and prompts. History clears when the app closes. The app does not take photos or record videos. ---------- ■ Connect with Other Apps and Tools The GET API returns results, images and app/device status as JSON over your network. The Control API starts or stops analysis and changes prompts while the app is in the foreground. curl notifications can send text, images, prompts, completion time or device status to an HTTP/HTTPS endpoint. Use external image AI through OpenAI-compatible APIs, including OpenAI API, Gemini and Claude. Save connections and models to switch between them and on-device AI. Tested with vLLM; compatibility varies by service and model. External integrations are off by default and require unlocking. Only selected data is sent outside your device when enabled. ---------- ■ On-Device AI Image Analysis A VLM (vision-language model) understands images and generates descriptions. Depending on your device and OS, Foundation Models uses Apple Intelligence’s AFM 3 Core or AFM 3 Core Advanced. Add local models such as Gemma 4, Qwen3.5 and Qwen3-VL. On-device AI analyzes offline without sending images or descriptions to external AI servers. Model downloads, external AI, curl, ads, purchase checks and support use the network. ---------- ■ Useful For Describe scenes and objects, read text, observe clothing, rooms or plants, or hear visual information. Use webcams, RTSP cameras, GET API or curl notifications. Explore generative AI, VLMs and Apple Foundation Models. ---------- AI results may be inaccurate. For medical, safety or navigation decisions, verify the actual situation rather than relying on AI alone. Requires iOS 18 or later. Apple Foundation Models analysis requires a compatible OS and device with access to those models. Models and features vary by device and OS. Terms of Use (Apple Standard EULA): https://www.apple.com/legal/internet-services/itunes/dev/stdeula/