What's new

New

Tool: HEIC to JPG / PNG

Many sites won't accept the HEIC photos iPhones take. In Tools, pick many at once, convert them to JPG or PNG and download — free. Photos are converted on your device only; nothing is uploaded or stored.

Improved

Connected AIs can now manage your library, fix faces and remove logos

ChatGPT and Claude connected to MagicLumen can now do almost everything the app can. Already connected? No need to reconnect — start a new chat to see the new abilities.

• Library: create characters (from text or photos), auto-draw reference sheets, save expressions, assign voices; create scenes and props; find characters and scenes in scripts or decks
• Tools: face fix, logo removal
• Voice cloning: make your own voice from a recording (turn it on in the app first)
• Projects: rename, delete, write scripts, look up generation history
• Before using a real person's photo or voice, editing someone else's video, or deleting anything, the AI asks you first

Improved

Library and My characters, now in one place

Library in the left menu now has three tabs: characters, locations and props. Characters is the former My characters, with reference sheets, voices and expressions; locations and props can be uploaded or generated from a sentence, and open large when you click them.

New

Import into Library from files

Upload a character bible, script, pitch deck or concept art (PDF, Word, PowerPoint, Excel, images or plain text) and AI pulls out the characters, locations and props. Tick what you want, tweak the descriptions and create them in one go: characters get a full reference set and a voice, and any character photos in your files are used as reference. Free imports every day.

New

My characters

New in the left menu: My characters. Upload photos of yourself or an actor, or describe a look in words, and AI draws a set of reference sheets (avatar, full body, expressions, turnaround). Pull them from the character library on any project's canvas so generated images and videos keep the same face.

• When uploading photos of real people, you confirm you have the right to use their likeness
• Expressions made with Emotion adjust can be saved to the character and placed on the canvas with it
• Characters you saved before in Library or from the canvas are already here

Improved

12 voices for voiceovers, and characters can have their own

Audio nodes on the canvas let you pick a voice and speed, with a preview for each. There are 6 Chinese voices, from gentle and bright female voices to a young man and a mature uncle, plus English, Japanese and Korean. Pick a voice for a character in My characters, and audio nodes connected to that character's image read in its voice automatically.

New

Effects library

14 ready-made video effects you can preview on hover: bullet time, crash zoom, through the clouds, FPV dive, day to night, into light… Pick one, add an image and a line about the scene, and get a short clip with camera moves.

• Use it right from Tools, no canvas needed
• Video nodes on the canvas have an Effects button too; the effect you pick is applied to that generation

New

Tool: Face fix

Faces in AI full-body shots are often blurry. Face fix sharpens the eyes, lashes, lips and skin texture without touching the body or background, in a few seconds. Free photos every day, and you can fix image nodes right on the canvas.

Improved

Step-by-step guides for every tool

The first time you open Smart edit, Remove logo or Face fix, a guide walks you through it once; click “How it works” to see it again. Tool pages now show before-and-after examples too.

New

Smart edit

Add a Smart edit node to the canvas and connect video nodes to it. In the editor you can order clips, trim them, add captions and music, and export one finished video.

• AI captions: writes a line for each shot from what's in it, so the captions tell one story
• Shrink to 15, 30 or 60 seconds in one click; each clip is trimmed and keeps its middle
• Automatically ordered left to right as on the canvas
• Cut or fade between clips, and use connected audio as music

New

Two new image models: Qwen Image and FLUX.2 Klein

Pick them in any image node on the canvas. Qwen Image is fast, good for trying out compositions; FLUX.2 Klein adds more detail for final images.

Improved

Image points now follow the actual size

Square and smaller images now cost fewer points; large landscape and portrait images cost the same as before. The canvas shows how many points a generation will use before you send it.

New

New tool: Remove Logo

Upload your own video, draw boxes around the logos or text you want gone, and download the cleaned video a few minutes later. You get free uses every day, then it costs points. Only use it on videos you own or are licensed to use.

New

Now in Traditional Chinese, Simplified Chinese, English and Japanese

Switch languages from the website header or the app's left menu. Scripts and Luman's replies follow the language you choose.

New

Desktop app for Mac

It uses the same account and projects as the web app. New features update automatically when you open the app, so you never have to download it again.

Improved

The website remembers you're signed in

After signing in to the app, the website shows your account in the top-right corner. Click Open the app to go straight back to your work.

Fix

Cleaner watermark removal on landscape videos

Landscape videos sometimes kept a faint trace of the watermark in the bottom-right corner. It's now removed completely.

New

Free points when you sign up

You get points the first time you sign in, ready to use on images and videos.

New

MagicLumen launches

Go from a single sentence to a script, characters, storyboards and video, all on one canvas.

• Canvas: connect image, video, audio and text nodes freely, and use each result as the reference for the next step
• Script setup: blend two genres into one story and get an outline plus the first episode's script
• Character library: place a full set of reference sheets for official or your own characters on the canvas, so each character looks the same in every shot
• Luman: your creative assistant on the canvas. It can brainstorm, rewrite prompts, or do the work for you
• Generated videos have the corner watermark removed automatically and are saved to your asset library