Enter a topic or a press release and choose the speaker and scene. The system writes the script, records the voiceover, creates pictures from three camera angles (wide, medium and 45°), generates a lip-synced talking video for each shot, and finally joins them into one finished presenter video. You can check each step along the way and only move on when you're happy.
In the Your talk board, type what you want to talk about into What do you want to say?. It can be a one-line topic, or you can paste in a whole press release, for example:
Introduce our company's newly launched AI digital human product: one sentence turns into a presenter video. It's built for marketing and operations teams at small and mid-sized businesses.
In the Speaker board, first choose the speaker's Gender (Male or Female; Male by default), then a Style:
Realistic (default) Anime (JP) 3D Cartoon Chinese Style Other (custom)
Choosing “Other (custom)” adds a text box where you write the style you want, such as “cyberpunk” or “watercolor picture book”.
The Voice list switches automatically to match the gender, and it follows the interface language, so in English you get native English voices. Each voice is named after its tone (for example “Compelling Lady”, “Calm Woman”); pick the one that best fits your content.
Under Stage, enter the mood you want in Stage and mood, such as “cool tech blue”, “warm tones” or “minimal white”. The speaker desk, big screen and other layout are arranged automatically, so you don't need to describe them.
When everything is set, click Take the stage →. The button changes to “Drafting & building session…”, and after a moment the generation workspace opens.
Once in the workspace, the system writes the broadcast script (“Script Generation”) and creates the digital human (“Character Design”) at the same time. All 7 steps are listed on the left, and the top bar shows “x / 7 steps” with a progress bar.
When a step finishes, its status changes to “◆ Awaiting Confirmation”. Take a look:
After the character is confirmed, AI seats the speaker in the studio and produces a blocking image (“Scene Blocking”). The three camera angles are generated only after you confirm it.
Multi-Camera Shots generates three pictures from the blocking image — wide, medium and 45° — each labeled with its angle in the bottom-left corner.
As soon as the script is confirmed, the voiceover is recorded shot by shot (“Voiceover Audio”), in parallel with the picture steps above. Each segment is labeled “Shot N voiceover” and can be played right away:
Once the camera shots and the voiceover are confirmed, “Shot Videos” generates a lip-synced talking video for each shot, showing each one as it completes (“Clip 1”, “Clip 2” …). This step needs no confirmation, but it takes the longest, so please be patient.
When all shots are done, “Video Assembly” automatically joins them into one finished video. The title changes to “Spotlumen · Complete”, and you can play it right on the page.
Click ⬇ Download Package in the top bar to download a zip file. It contains one folder per step with the final video and every asset generated along the way (script, character images, camera shots, each voiceover segment and shot video). The final video is in the folder starting with V-02.
Everything you make is saved under My Works in the top bar. Each entry shows its title, status (Complete / Generating / Failed …), how many steps are done, whether the package is ready, and when it was created. Click one to go back to its workspace.
Completed works have a 🔗 share icon on the right. It opens the “Share this video” dialog: click Create share link, then click “Copy” next to the link and send it.