Video, Voice & Music

Studio video is a deterministic composition workflow. It assembles clips, still images, SVG/vector assets, music, voiceover, captions, and titles into finished MP4 outputs.

Current verified path: The deployed Itskatzu proof project has successfully rendered a vertical TikTok/Reel MP4 from timeline assets. The verified artifact is 1080 x 1920, includes a video stream and an audio stream, uses a Piper voiceover generated from an approved campaign script rather than the raw project prompt, and is stored as an Immich-backed final video asset.

Video Modes

ModeUse it forRequired inputs
SlideshowPromotional video from images, SVGs, and/or video clips.At least one image, vector, or video asset.
Title cardSimple prompt-driven video with optional music or voice.No visual attachments. Optional music/voice assets only.
Variation / editComposed edit from existing video footage, plus optional stills, overlays, music, and voice.At least one video clip.

Video Format Presets

Choose the intended channel before generating. Studio will include the size and aspect ratio in the compiled video instructions.

PresetSizeAspectUse
Reels, TikTok & Shorts1080 x 19209:16Instagram Reels, TikTok, YouTube Shorts.
Square Feed Video1080 x 10801:1Instagram and Facebook square feed.
Horizontal Feed Video1920 x 108016:9Facebook and X/Twitter feed video.
YouTube Video HD1920 x 108016:9Standard YouTube video.
YouTube Video 4K3840 x 216016:94K UHD video.
YouTube Banner2560 x 144016:9YouTube channel art or banner video asset.
YouTube Thumbnail1280 x 72016:9Video thumbnail creative.
Presentation Video1920 x 108016:9Deck or presentation video.
Mobile Video Ad1080 x 19209:16Vertical mobile ad.
LinkedIn Landscape Video1920 x 108016:9LinkedIn landscape video.
LinkedIn Portrait Video1080 x 13504:5LinkedIn portrait feed creative.

Smart Resize Expectations

When you choose a new video size, Studio should recompose the visual material for the new canvas. It should not simply stretch the input. A good resize keeps subjects, logo, text, captions, and CTA areas inside safe zones.

Auto Captions

Auto captions are available in video prompt controls. When on, Studio adds readable captions from approved creative copy, voiceover script, or campaign copy.

Timeline Render Readiness

Before Studio allows a composed timeline render, the timeline readiness panel checks for required features. If a video prompt asks for a full campaign video, expect these checks:

FeatureHow to satisfy it
Visual sceneAdd a video, image, or vector/SVG visual asset to the timeline.
Logo overlayAdd a logo asset to the timeline. Logo assets are rendered as overlays, not as separate late stills.
Text or CTA overlayAdd a text layer, CTA layer, or approved copy overlay.
CaptionsTurn on auto captions or add a caption layer.
Music bedAdd a music asset generated by Audiocraft or uploaded as a reusable audio asset.
VoiceoverAdd a Piper voiceover asset.

When every required feature shows OK, Render timeline creates a final MP4. The current production path writes the generated file as a project asset and syncs the final video to Immich.

Voiceover With Piper

Piper is the integrated text-to-speech engine. Voice mode creates standalone voiceover audio from script text. For best results, write the actual narration script in approved campaign language; do not rely on the general generation prompt as the spoken script.

  1. Open a project.
  2. Select Voice.
  3. Write or paste the script text.
  4. Choose an available Piper voice.
  5. Generate the voiceover.
  6. Attach the voice asset to a video project if the video prompt asks for narration.
Important: If a video prompt asks for voiceover, the current validation expects a voice asset to be attached. Generate the voiceover first, then attach it to the video job.

Music With Audiocraft

Audiocraft is the integrated music generation surface. Music mode creates a reusable background music asset.

  1. Open a project.
  2. Select Music.
  3. Choose mood: uplifting, cinematic, or ambient.
  4. Choose duration: 15, 30, or 45 seconds.
  5. Write a short music direction such as calm premium tropical launch music, instrumental, warm and restrained.
  6. Generate and preview the music asset.
  7. Attach it to a video project as a music bed.

Common Video Blocks And Fixes

Message or problemFix
Slideshow mode needs at least one image, SVG, or video asset.Attach a visual asset or switch to title-card mode.
Title-card mode uses the prompt plus optional music or voice assets only.Remove image/video assets or use slideshow/variation mode.
Video edit mode needs at least one existing video clip.Attach an MP4 or switch to slideshow/title-card mode.
Video asks for voiceover but no voice asset is attached.Generate or upload a voiceover asset first, then attach it.
Silent export when audio was expected.Confirm the prompt asks for music/voice and that the correct audio assets are attached.
Render timeline is disabled.Open the readiness panel and add the missing visual, logo, text/CTA, caption, music, or voice asset.
Logo appears as a separate still instead of an overlay.Use the current deployed build and add the logo from the Assets tab. Logo assets now compile as overlays over the visual scene.

Current MLT Editor Capability

The current video workflow is a validated MLT-backed editor and render flow. The project workspace supports multi-asset timelines, split/duplicate/trim actions, transitions, transform and crop controls, opacity and transform keyframes, caption settings, audio gain/fades/ducking, and export profile controls for supported video formats. Render timeline creates an MP4 from that editor state.

Use this editor when a generated video needs further user-controlled cleanup. The supported edit state is saved with the project so a generated video can be reopened, adjusted, and rendered again into the selected platform format.