Every Hiroska feature: generate video, images and audio, restyle, reframe and dub footage you have, edit on a timeline, then grade, caption and export 4K.
AI video generator — Turn a prompt into video with Kling, Seedance, Hailuo and more, then cut the result on a multi-lane timeline and export 4K MP4 with no watermark.
AI image generator — Generate images with Seedream 5, FLUX.2, Nano Banana 2 and GPT Image 2, then refine in place: reference edits, background removal and 2×–4× upscaling.
AI audio generator — Generate every sound a video needs: music beds with lyrics, natural voiceovers, one-shot sound effects and foley matched to your clip. Priced per run.
Green screen (chroma key) — Remove a green or blue screen in the browser: pick the key colour, tune similarity, edge blend and spill, and preview the exact export result live.
Auto captions — Generate captions in 30+ languages, as standard lines or karaoke word-by-word highlights. Every line is a text layer you can edit, restyle and move.
Keyframe animation — Animate any clip with keyframes on position, scale, opacity, rotation and volume. Drag keys on the canvas; the 4K export renders the same curves.
Ken Burns effect — Give any photo a cinematic zoom and pan. Frame the start and end windows on the image, and the move renders the same in preview and 4K export.
Video transitions — Cut smoothly with cross-dissolves, directional slides, zoom punches, fades and dips to black. Effects carry through, and the export matches the preview.
Picture-in-picture — Stack video lanes and float picture-in-picture overlays with drag handles, aspect-true boxes, opacity and rounded masks. Multi-layer editing in one page.
Text & titles — Add text to video: 12 fonts with true bold and italic, outline, shadow and background boxes, drag-anywhere placement, and four entrance animations.
Color grading & LUTs — Grade video in the browser with brightness, contrast, saturation and temperature, one-click film looks, and your own .cube 3D LUTs previewed accurately.
Audio editor — Multi-lane audio for video: duck music under speech, cut silences and filler words in one pass, plus fades, volume keyframes, EQ, reverb and cleanup.
Video eraser — Paint any region of a clip transparent with a soft brush, or name what to remove and let AI rebuild the background. No green screen, no key colour.
Blur, masks & spotlight — Blur a face, plate or logo with a draggable region, crop clips to feathered shapes, or aim a spotlight that dims the rest of the frame. The export matches.
Video upscaler — Upscale video and images 2× or 4× with Topaz-class AI models. Sharpen old footage, enlarge small AI renders, and keep everything editable on the timeline.
Lip sync & talking heads — AI lip sync two ways: put a new voice on footage of a speaker, or turn one photo into a talking presenter. Feed it a recording or a generated voiceover.
Motion transfer — Give a still image the motion of any video clip. The model maps the reference performance onto your subject and returns a normal clip for the timeline.
Background remover — Remove the background from any image or video in one click. Transparent PNG or alpha WebM, no green screen, ready to layer over a new scene.
Frame interpolation — Synthesize the in-between frames of any clip with RIFE: smooth choppy or low-fps footage, and stretch action into clean 2×–4× slow motion without ghosting.
Hot Mode — An opt-in mode that relaxes the default generation filters. Off until you turn it on, private to your workspace, reversible at any time.
Face swap — Swap a face into any still in your library: be in your own renders, or keep one performer across a cut. Part of Hot Mode, and permission is required.
Video restyle — Describe a change and the clip is re-rendered to match: make it snow, turn day into night, swap a jacket. Your footage stays exactly as you shot it.
Aspect ratio changer — Turn a 16:9 edit into 9:16 for Reels, Shorts and TikTok without cropping the subject out of frame. The model paints the new edges instead of chopping.
Video translator — Translate the speech in a clip into another language and re-sync the mouth to the new audio. Same performance, new language, no reshoot.
Trained styles — Train Hiroska on a dozen of your own pictures and every generation comes out in your look. Styles for a look, a product, a person or your own footage.
Voice cloning — Record thirty seconds once and narrate every video in your own voice. Three models from cheap drafts to broadcast takes, and consent recorded up front.
Connected AI — Add Hiroska as a connector in claude.ai, ChatGPT, Claude Code or Cursor. The assistant plans, generates, edits and exports videos as you, with a daily limit.