Built like a pro tool. Feels like play.
Every frontier image and video model behind one compact prompt box. Compare them side by side, edit with plain English, animate any still, shoot a whole scene's coverage from one sentence, cut it all into a finished film in Director, and see the exact price on the button before you click. Native on Mac and Linux.
Three tabs. The whole frontier.
Plus the new Director tab, which cuts what they make into a finished film.
Images from the top of the leaderboard
Twenty-three state-of-the-art image models behind one compact picker: Nano Banana 2, Seedream 5.0, GPT Image 2, Qwen Image 3, the Ideogram family, FLUX.2, Recraft with true SVG output. Choose a flagship or a turbo drafter, and the studio adapts its controls to whatever you pick.
- Compare mode: run one prompt across 3 models side by side
- ✦ Enhance: one click rewrites your prompt for the selected model
- Live cost on the button: see the price before you hit Generate
- Reference images: attach up to 10 refs to lock a character, product, or style
- Task chips: filter the model list by what you're making: product accuracy, readable text, character continuity, cheap drafts
- Brand kits: one chip applies a client's anchor images, palette and house rules to every prompt
- Aspect ratios up to 21:9, 1K/2K/4K tiers, batches up to 4, seeds & negative prompts
Cinema-grade video, start frame to final cut
Now with Seedance 2.5, ByteDance's newest flagship, up to 30 seconds in a single take with native audio. Thirty video models in all: Kling 3.0 Pro, Standard & native-4K (the current text-to-video arena leader), FLUX 3, Veo 3.1, Wan 3.0, MiniMax H3 Max, HappyHorse 1.1 and more, with duration, resolution and bitrate controls per model.
- Image-to-video: drop any render in as a start frame; the clip keeps your image's exact ratio
- Start and end frames on Seedance, Kling 3.0, Kling O3, Wan 3.0 and Wan 2.7
- Extend: auto-grabs a clip's last frame to keep the shot going
- Kling multi-shot: on Kling 3.0 and O3, write up to six timed scenes and get them back as one continuous take
- Honest prompt limits: a counter shows each model's own character cap and stops an over-long prompt before it's sent
- Audio On/Off: silent clips for your own sound design; Kling O3's price drops with audio off
- Director panel: genre, palette, lighting, camera move and lens as one-click direction for a shot, with saveable "looks"
- Social presets: Instagram, Facebook, TikTok and X formats in one tap
Edit with instructions: images and video
"Make the bottle in the red box emerald green." Draw numbered boxes right on your image, reference them in plain English, and the edit lands precisely where you pointed. No layers, no selections, and no file leaves the app.
- A canvas, not a thumbnail: the source fills the workspace with scroll to zoom and drag to pan, and one button swaps original against edit at the same zoom and position
- Mask a region: paint over the part that may change, and every pixel outside your stroke stays your original's, with a soft edge. Images only
- Mark up: drag numbered boxes on any source and type what changes in each; it writes the instruction for you
- Sixteen image editors, family-matched: Nano Banana 2, GPT Image 2, Seedream 5.0, Qwen Image 3, FLUX.2, Ideogram and more; a render opens in the editor from the family that made it
- Videos edit the same way: open any clip, describe the change: Kling O3 Edit, Wan 2.7 Edit, Gemini Omni Flash Edit and more, priced per second of your actual clip
- New Shot: a different camera angle of the same clip's subjects and scene
- Drag-drop or paste straight from your clipboard
Cast your own references into the shot
Gemini Omni References doesn't animate your image. It stars it. Attach up to four references (a product, a face, a location) and direct them by number. The model composes a brand-new moving shot around exactly the things you gave it, sound included. Seedance 2.5 References goes further, blending images, clips and audio into one take.
- Role binding: "the bottle from image 1 on the cliff from image 2, slow orbit"
- It speaks your lines: put dialogue in quotes and your referenced character says it, lips synced
- Native audio: ambience, foley and voice generated with the picture
- 3–10 second takes · about $0.13 per second, previewed on the button
This clip was made with it: one prompt, reference images, nothing else.
Storyboards: the film before the film New in 2.6
In Director's The story section, pick the characters and location you've saved, write a brief, and Plan shots writes the shot list: framing, angle, action, who speaks and what carries into the next shot. Then every shot gets a Scene Still Frame drawn from the people actually in it, so the same faces, wardrobe, set and grade hold all the way down the board.
- ≈ 2¢ to plan, ≈ 12¢ a still. A six-shot board lands under a dollar, before any video is bought
- Your Details win: what you wrote on a character beats the photo on hair, clothing and accessories. The face always comes from the photo
- Contact sheet or production board: description, framing, location and duration under every frame; save a PNG or the whole board as a PDF
- Straight to video: every approved frame opens its own shot, or send the whole board to Seedance as one multi-shot draft
- Dialogue and lip-sync: planned lines, a voice per character, and one button to make a still talk
Director: from shots to a finished cut
The 🎬 Director tab plans a piece as up to sixteen shots, all on one model so the look holds. Storyboard them first or go straight to motion: generate takes, approve a winner per shot, narrate it in one voice, lay a soundtrack from your Audio gallery under it, set the mix, choose the transitions, and export one finished film to your gallery.
- Shared world: reference images, a saved character or a brand kit ride into every shot; an anchor frame opens every shot, or each shot chains from the last
- One narrator, a line per shot, with
(pause)cues and a start time inside the shot - Eight transitions plus fades in and out, and exports with them render on your own computer, free
- Sequences save as 🎬 cards in the Gallery and reopen right where you left them
Parallel queue
Fire multiple generations at once, across all tabs, with live queue status.
Crash-proof jobs
In-flight renders survive an app restart and reconnect automatically.
Finishing tools
Topaz upscaling, background removal, reframing, slow motion and clean plates on any render, straight from the gallery, with the price quoted first.
Cost receipts
Every item records its estimated cost; the top bar tracks monthly spend.
Gallery tools
Search, filter, favorites, version history, delivery packs with a manifest, and a portable Back up / Restore.
Prompt history
Your last 50 prompts, one click away. Plus a Surprise-me generator.
Render chime
A soft two-note chime and a native notification when a render lands.
Media vault
Auto-saves to ~/Pictures/Asset Lab. Group renders into projects and the folders mirror your gallery.
The studio that makes you better at this.
The creative-workflow suite: a guided prompt builder, reusable characters, an AI render doctor, and a gallery that never forgets a single render.
✦ Prompt Lab
A guided builder that walks you through what strong prompts are made of (subject, setting, camera, light, mood, finish, motion), with AI drafting and polish for a fraction of a cent. Send the result straight to Image or Video.
Characters that persist
Save any render as a character, build a multi-angle character sheet, and cast the same face into new scenes, outfits and worlds with one click, across image and video models alike.
✦ Analyze & fix
Tell the AI what you don't like about a render. It looks at the actual image beside its prompt, diagnoses the miss, and hands back a surgical edit and an improved prompt. One click to either fix.
A gallery with no ceiling
No render cap: every generation stays reachable forever, with its seed and settings on record, and now its whole family of edits, extensions and upscales as a Versions strip.
Library, blueprints & brand kits
Saved prompts, reusable blueprints, your characters, and brand kits: up to eight anchor images plus a palette and house rules that ride into every Image, Video and Director prompt.
Scores that watch the picture
Score a clip with no prompt and Asset Lab reads the footage first (era, setting, energy), then writes a period-accurate brief before the music is composed.
One prompt.
A whole scene's coverage.
Describe a moment once and Asset Lab shoots it six ways: Wide, Medium, Close-up, Over-shoulder, Low angle and Aerial. Every angle is anchored to the same hero frame, so the wardrobe, the vehicle, the weather and the light hold from shot to shot. You get a scene, not six unrelated pictures.
A real Shot Set, start to finish: one sentence in, these six frames out. Nothing was re-prompted, retouched or hand-picked.
Locked to one scene
Each angle references the hero frame, so your subject and location stay put.
Priced before you shoot
The total for the whole set sits on the button. Pick fewer angles, pay less.
Arrives as a set
The angles land grouped in the gallery as one project, and on disk the same way.
Every frame animates
Send any angle straight to a video model or, in Resolve, on to the timeline.
Storyboard a scene, build a product's whole ad set, or block out coverage before you ever roll a camera. Included in both editions at no extra cost.
Built for people who ship visuals.
If you need a steady stream of original images and video, and you'd rather pay the models at cost than rent another subscription, Asset Lab was built for you.
AI artists & Fal power users
You already have the key, so skip the API console. Every frontier model in one prompt box, side-by-side compare runs, and a local vault that never loses a render.
Indie developers & startup founders
Hero images, App Store shots, and landing-page art shipped tonight, with no designer on payroll. The price sits on the button, and most renders cost cents.
Indie game developers
Character concepts, environments, and item references at volume. Lock a look with seeds, iterate fast on a turbo model, then upscale the keepers 2×.
YouTubers & short-form creators
Thumbnails that pop and AI b-roll from the same prompt box, then turn any still into motion with Kling, Veo, or Wan using one click: "Use as start frame."
Freelance designers & illustrators
Moodboards and concept rounds in minutes, not mornings. Explore wide, refine with a flagship, edit with plain-English instructions. No subscription eating your margin.
Self-published authors
Cover concepts, character portraits, and launch promos for every book, with typography-strong models like GPT Image 2 that can actually spell your title.
Cut in DaVinci Resolve? The same studio ships as AssetLab Resolve, docked beside your timeline.
About the studio
What happens to my renders?
They land in your in-app gallery and auto-save to ~/Pictures/Asset Lab as real files, organised into projects. Even when a model host's temporary links expire, your local copies stay.
Can I compare AI models side by side?
Yes. Compare mode runs one prompt across up to three models at once, so you pick the best result for the shot instead of guessing from reputation. Every render records its model, seed, settings and cost for an exact re-roll later.
Can it edit video with a text instruction?
Yes. Open any clip from the gallery and describe the change. Kling O3 Edit and Edit Pro, Wan 2.7 Edit, Gemini Omni Flash Edit, Grok Imagine Edit and HappyHorse Video Edit are priced per second of the source clip, and the exact total sits on the button before you commit. Rates are on the rate card.
Can it make a multi-shot video, not just single clips?
Yes, two ways. Director (new in 2.4) builds a finished cut from up to sixteen separately generated shots (storyboarded first if you like) with narration or character voices, a music bed, a mixer, transitions and fades. On the Kling 3.0 and O3 models, Multi-shot mode renders up to six timed scenes as one continuous take.
Does it make voiceover and music too?
Yes. The Audio module is included in both editions: 272 voices you can audition free, voice cloning, six music models, sound effects and music scored to a cut.
Look around before you buy.
The free demo is the real app: every tab, every model, every live price. It just won't generate until you unlock it.