# AI Creator Local Tools > AI Creator is a free, local-first browser toolbox for practical image and audio workflows. Verified model inference runs on the user's device after model files load. Canonical site: https://ai-creator.top/ About and processing disclosure: https://ai-creator.top/about/ Developer profile: https://github.com/MartinDelophy Hugging Face profile: https://huggingface.co/haixin Official community: https://discord.gg/uq2uvUTBr Sitemap: https://ai-creator.top/sitemap.xml ## Tools - [AI Background Remover](https://ai-creator.top/removebackground/): Segment a foreground subject and export an image with a transparent background. - [AI Object Remover](https://ai-creator.top/photomodify/): Paint a mask over an unwanted region and reconstruct it with image inpainting. - [Photo to Line Art](https://ai-creator.top/linechange/): Extract raster contours for drawing references and creative work. - [AI Photo Relighting](https://ai-creator.top/relight/): Adjust simulated light position, radius, intensity and color. - [Photo to Anime](https://ai-creator.top/animatechange/): Apply an AnimeGAN-inspired illustration transformation to a photograph. - [Whisper Audio Transcription](https://ai-creator.top/whisper/): Convert recorded or uploaded speech into editable text in the browser. - [Timeline Studio](https://video-editor.ai-creator.top/): A separate multi-track AI video editor for voiceovers, captions and browser video production. ## Engineering resources - [Generation plugins need a media boundary](https://ai-creator.top/resources/browser-video-editor-generation-plugin-media-boundary/): Timeline Studio connects Puter.js, ComfyUI and Stable Diffusion WebUI or Forge through source-integrated manifests and isolated adapters. Provider code returns normalized media; the host validates real bytes, decodes every output and commits the full batch to My assets without automatically changing the timeline. - [From a creative brief to an editable video project](https://ai-creator.top/resources/from-creative-brief-to-editable-video-timeline-skill/): Timeline Studio's Agent Skill inspects named media, records reversible editorial decisions, operates on portable .timeline projects through revision-checked commands, builds phrase-level narration as an audio spine, routes richer composition to the browser editor, and verifies both the reopened project and decoded result video. - [NanoVSR 4× video restoration in the browser](https://ai-creator.top/resources/nanovsr-4x-video-restoration-in-browser/): Timeline Studio runs pinned T1 and five-frame T5 FP16 NanoVSR models through WebGPU. The neural working canvas scales from 320×180 to 1280×720, larger sources keep their dimensions through source-detail protection, and 12 fps video groups are composed into an H.264 MP4 with optional source audio and reversible project metadata. - [Removing watermarks from images and video in the browser](https://ai-creator.top/resources/remove-watermarks-from-images-and-video-in-browser/): Timeline Studio runs a pinned MI-GAN ONNX model through WebGPU to reconstruct tightly masked regions locally. Images produce reviewable PNG assets; videos add independent time ranges, interpolated position keyframes, sequential frame repair and FFmpeg.wasm MP4 composition with optional source audio. The article covers visual limits and the requirement to modify only authorized media. - [Edit the project file, then let the browser finish](https://ai-creator.top/resources/edit-timeline-project-file-before-browser-render/): Timeline Studio treats the portable .timeline archive as a versioned editing protocol. An Agent can inspect project, track and clip facts; preview a revision-checked, idempotent command plan; preserve archived media while writing a sibling project; and return that editable file to the browser for perceptual review and final render. - [One video-editing Skill across Agent channels](https://ai-creator.top/resources/one-video-editing-skill-across-agent-channels/): Timeline Studio packages one editable-project workflow for Codex, Claude Code, Copilot and Gemini CLI. Registered edits use a revision-checked, idempotent JSON command runner; unsupported operations fall back to verified browser control; the portable .timeline archive remains the shared source of truth. - [Mobile video timeline under two thumbs](https://ai-creator.top/resources/mobile-video-timeline-under-two-thumbs/): Timeline Studio's in-development mobile workflow keeps the multi-track project model while adding anchored pinch zoom, direct per-clip action sheets, track locks, stable sticker lanes, drag-versus-pinch cancellation and pitch-preserving audio speed for preview and offline export. - [Open media library inside a browser video editor](https://ai-creator.top/resources/open-media-library-inside-browser-video-editor/): Timeline Studio searches Wikimedia Commons or configured Pexels sources for visual media and Openverse for CC0, CC BY or public-domain-marked music, preserves provider and license metadata, previews remote assets and resolves selected files into Blob-backed timeline objects. Users must still verify each source page and its license. - [Fixing audio source offsets after a timeline split](https://ai-creator.top/resources/fix-audio-source-offset-after-timeline-split/): Timeline Studio v0.4.1 documents how a split clip must preserve its accumulated sourceStart so playback, seeking, waveform slices, offline mixing and compatibility export all address the same source-media range. - [Browser Talking Portraits: Encoder Test and Model Costs](https://ai-creator.top/resources/create-talking-digital-human-in-browser/): A source and cost assessment of the experimental JoyVASA/LivePortrait route: four-second duration cap, about 739 MB of preview models, sparse neural keyframes, nearest-frame reuse and a silent intermediate WebM. A new non-neural test-card video verifies the encoder. The historical neural run was not repeated; a bounded one-second portrait preview is proposed before a larger render. - [Browser auto-edit: why a short shot can disappear](https://ai-creator.top/resources/browser-auto-edit-scene-detection-caption-rough-cut/): Inspect four actual test videos: a short shot can be missed before or after sampling. Compare candidate scores, timing failures and reproducible source evidence. - [Local-first browser video editor architecture](https://ai-creator.top/resources/local-first-browser-ai-video-editor-architecture/): Architecture decisions for Workers, model caching, temporal analysis, timeline state and resilient export. - [Porting LivePortrait 5D GridSample to WebGPU](https://ai-creator.top/resources/liveportrait-5d-gridsample-webgpu-port/): An earlier 5D GridSample port record with fixed-input numerical errors, historical per-frame timing and explicit limits on current perceptual and audiovisual evidence. - [FFmpeg media pipelines for reliable web playback](https://ai-creator.top/resources/ffmpeg-web-player-media-pipeline/): An engineering field note about demuxing, decoding, compatibility fallbacks, browser playback and export boundaries. - [A transaction boundary for a video-editing Agent](https://ai-creator.top/resources/video-editing-agent-transaction-boundary-deepseek-harness/): A source-reviewed DeepSeek Harness integration with narrow tools, revision-safe writes, dry runs and a browser-owned media finish. ## Evidence-backed workflow guides - [Remove an image background locally](https://ai-creator.top/resources/remove-image-background-locally/): A source/output cutout pair, alpha-edge observations and a repeatable acceptance check over contrasting backgrounds. - [Image matting versus background removal](https://ai-creator.top/resources/image-matting-vs-background-removal/): A difficult portrait boundary, diagnostic background and documented limits for an interactive selection workflow. - [Whisper transcription workflow](https://ai-creator.top/resources/whisper-transcription-workflow/): A playable 3.5-second test recording, the observed browser transcript and a human verification method. - [Remove unwanted objects from photos](https://ai-creator.top/resources/remove-unwanted-objects-from-photos/): A marked source, reconstructed output and visual inspection method for texture, lines, blur and brightness. - [Photo relighting guide](https://ai-creator.top/resources/photo-relighting-guide/): A same-size source/result comparison that separates creative lighting changes from claims about recovered illumination. - [Photo to line art guide](https://ai-creator.top/resources/photo-to-line-art-guide/): An aligned split artifact for reviewing retained contours, omitted color and unwanted scene detail. - [Photo to anime responsibly](https://ai-creator.top/resources/photo-to-anime-responsibly/): A source/result pair documenting recognizable attributes, redrawn details and consent boundaries. - [Verify browser AI privacy](https://ai-creator.top/resources/browser-ai-privacy-explained/): A repeatable DevTools method for separating model downloads, media uploads, browser storage and optional remote services. ## Important limitations - AI outputs are estimates and must be reviewed before publication or important use. - Local-first does not mean offline on first visit: model files must be downloaded and initialized. - Users are responsible for source-file rights, privacy and compliance with applicable rules. - No account is required for the tools listed on ai-creator.top. ## Machine-readable access - [Agent access and processing boundaries](https://ai-creator.top/agents.md) - [Workflow directory (read-only)](https://ai-creator.top/api/workflows.json) - [Full published page content](https://ai-creator.top/llms-full.txt) - [Skill discovery index](https://ai-creator.top/.well-known/agent-skills/index.json) Published pages support Accept: text/markdown and expose a direct Markdown alternate.