Building compositions in CSuite.
Every control in the video and audio composition editors, with a figure for each, followed by recipes that take a piece from idea to exported file. The short version lives on the main guide; this page is for when you want to know exactly what a control does.
Get around
The editor
Open Compositions from the rail, the first entry of the second group. A composition is where single clips, stills, audio and titles become one finished piece: a video on a multi-track timeline, or an audio piece with its own speakers. Each composition is a file in your project folder beside the media it uses, saved about half a second after every change.
- 1 · Compositions list
- Your project folder pill, the + that creates one, and every composition in the folder with a frame or a type glyph. Escape, or pressing Compositions in the rail again, closes the open one.
- 2 · Header
- The title (edit it in place), Undo and Redo, the aspect-ratio pill, Export, and the toggle for the side panel. A Models pill beside it sets the chat, image, video, voice, music and sound-effect models the editor generates with.
- 3 · Export
- Opens the render settings; needs an active license.
- 4 · Preview
- The canvas at the composition’s aspect ratio, playing the mix of every lane. Drag a title or image here to move it; pull a corner to resize.
- 5 · Transport
- Play, Back to start, the time, a scrubber, the duration, and the timeline toggle (T).
- 6 · Timeline
- One lane per kind, Text, Visuals and Audio, top lane painting on top. Clips are bars in the kind colours with a thumbnail at the leading edge.
- 7 · Side panel
- Two tabs: Chat, where you describe the piece or an edit, and Layer, the properties of the selected clip. C toggles the panel; it is open by default.
Every change, yours or the chat’s, lands in one undo history (two hundred steps): ⌘Z takes back the last step whoever made it, ⇧⌘Z redoes. The side panel and timeline visibility are remembered across compositions.
Creating one
Delete on a list row removes the composition file only: “Files you already exported stay in your project folder.”
Aspect ratio
The pill in the header shows the current ratio and opens nine presets: Square (1:1), Landscape (21:9, 16:9, 3:2, 4:3) and Portrait (9:16, 2:3, 4:5, 3:4), each with its canvas size. The short edge is always 1080 pixels, so 21:9 is 2560×1080 and 9:16 is 1080×1920.
On an empty canvas the switch is immediate. With clips on it, a dialog explains that every clip is rescaled to keep its on-screen proportions (nothing stretches or leaves the frame), that one ⌘Z undoes the whole switch, and offers an Enhance with AI box: ticked, the chat opens with a suggested prompt to rebalance the layout for the new shape, which you review before sending.
Build the video
Drafting with AI
The assistant also picks a short visual theme (palette, lighting, mood) and appends it to every image and video prompt, so the shots read as one shoot. It is stored in the composition and needs no attention. The chat never adds video clips of its own; it fills an empty video layer you added, and otherwise works with images, which you can turn into video per layer.
The timeline
- Lanes: Text, Visuals and Audio, in the order they paint, top lane on top. The icon on the Audio lane’s header is its mute button. Clips that overlap in time within one lane stack into rows, row 0 on top; the timeline shows up to five rows before it scrolls.
- Move by dragging a bar’s body; drag it up or down to change its row, or into another lane of the same kind. ← and → on a focused bar nudge it a tenth of a second, a full second with Shift.
- Trim from either edge; a clip keeps at least a fifth of a second. Video clips trim from the end only, since footage starts at its first frame, and the left handle says so. A left trim on audio keeps the end where it is.
- Snapping to every other clip’s edges, the playhead and the start, with a guide line, plus the point where a clip’s own media runs out while you trim. There is no grid.
- Zoom with the −, + and Fit buttons, or
⌘/Ctrl+ scroll and a pinch, around the pointer. - Playhead: click the ruler to seek, drag its cap to scrub. Selecting a different clip seeks to its start.
- Delete or Backspace removes the selected clip; Esc deselects.
On the canvas
Visual clips active at the playhead are live on the preview. Click one to select it (the topmost wins where they overlap), drag its body to reposition with centre-line and edge snapping, and pull a corner handle to resize anchored at the opposite corner, down to five percent of the canvas. Each gesture is one undo step. Double-click a clip here or on the timeline to open the Layer tab on it. A clip that fills the whole canvas shows a rounded outline with its handles inside the frame.
Adding clips
- Add offers Text, Image, Video and Audio. Text inserts at once; the media buttons open a menu that leads with an Empty … layer (audio splits into music, voice and SFX, since the role picks the model later) above the files in your project folder, with thumbnails and lengths. Inserts land at the playhead on the first matching lane, creating it if needed; an empty layer opens the Layer tab so you can generate into it.
- Drop from your desktop onto a lane (a highlight and a snapped insert line show where) or onto the preview (an image or video lands centred at the drop point, audio at the playhead). Files already in the project are reused; others are copied in. The first file of a drop is used; an unsupported type says so.
- Every new visual clip fills the frame (fit: cover). Switch to Contain in the Layer tab to letterbox.
The Layer tab
The tab opens on the selected clip (“No layer selected” until then) with a kind icon, the clip’s name and lane in its tooltip, and a delete button. Content and timing are open; Position, Motion and Generate with AI are collapsed below. Every field commits on blur or Enter as one undo step.
- Text: the text, Font (the families installed on your machine, with a Suggested group of cross-platform picks on top), Size, Weight, Color and Align.
- Image and video: File (rename it in place; double-click the label to reveal it in Finder) and Fit, Cover or Contain. Video adds Volume (zero mutes) and Loop: when a clip outruns its source, Loop repeats the footage and off slows it to fill (“it plays at 67% speed to fill the clip”).
- Audio: File, Volume, Loop (hidden for voice), and Role (Music, Voice or SFX) on files you brought in yourself; generated audio shows its role in the header.
- Timing: Start, Length, Fade in and Fade out in seconds. Once a video or audio clip’s length drifts from where its source ends, a reset button puts it back.
Position and motion
- Position: X, Y, Width and Height as percentages of the canvas, plus Rotate in degrees. The same values the canvas handles edit; type them for exact placement.
- Motion: a From and To for Move X, Move Y, Scale and Rotate, played over the clip’s length with an Easing: Linear, Ease in, Ease out, Ease in and out, Ease out with overshoot, Strong ease out, Strong ease in and out. Reset clears it. A slow scale from 1.0 to 1.1 is the classic slideshow move.
Generate, Make video, Extend
Every clip except text has a Generate with AI section. What it offers depends on the layer:
- An empty layer gets a prompt and Generate, asking for that layer’s own length where the model takes one (“this layer’s (6s) is what gets asked for”). An empty video layer is a fresh text-to-video run at the composition’s ratio and theme.
- A generated image or audio clip gets Regenerate, pre-filled with the prompt that made it, replacing the media in place.
- An image clip also gets Make video: the still becomes the start frame, the prompt describes the motion, your video model renders, and the clip is swapped for the video in place with its timing and framing kept. One
⌘Zputs the image back; the file stays in your project either way. - A video clip with footage gets whichever of Extend (use video) and Extend (use last frame) your video model supports, with “Describe what happens next…”. The extension lands as a new clip right after the source on the same track, selected, one undo step. A Runware draft clip adds Render in 1080p, which replaces the draft in place.
Progress and a Stop button show inline, with a percentage for video. One conversion, extension or render runs at a time (“Another layer is already rendering.”). If you change the layer while its generation is still running, the result is saved to your project instead of swapped in, with a toast saying where it went. Generated audio grows to a longer result only where no length could be requested; video shrinks to a shorter render; nothing else on the timeline moves.
Export
Export in the header opens the settings: Format MP4 (H.264), WebM (VP9) or MOV (H.264); Resolution Native, 480p, 720p, 1080p or 2160p, aspect preserved; Framerate Native, 24, 30 or 60. A line sums it up (“Capture: 1920×1080 @ 30 fps · 15s (450 frames)”). The render walks Loading, Warming up, Capturing, Encoding and Saving with a percentage and Cancel export, entirely on your machine. A dialog names the file (“it’s in the Video workspace”) with Show in Finder; an OS notification fires if you are elsewhere. Exporting needs an active license. An empty layer still on the timeline renders as its dashed outline, so fill or delete it first.
Audio compositions
The audio editor
Pick Audio when you create a composition and the editor swaps the canvas for a Script panel over a timeline with one lane per speaker plus Music and SFX lanes, and the side panel gains a third tab: Chat, Cast and Clip. A new one starts with “Speaker 1” and their lane. The header keeps the title, undo and redo, Export and the panel toggle; there is no aspect ratio.
- 1 · Export
- Mixes every audible lane down to one audio file.
- 2 · Script
- The lines in order, each a voice clip on its speaker’s lane: who says it, the status, Generate, delete, and the text.
- 3 · Timeline
- A lane per speaker in their colour, plus Music and SFX once they hold a clip. Bars show the waveform; the lane icon mutes.
- 4 · Cast
- Each speaker’s name, colour, voice model and voice, with a preview.
The script
- + Line (in the script header or the timeline’s Add row) asks “Who says it?”, listing the cast and New speaker. The line goes after the selected one, later clips shift to make room with a short gap, and the text field is focused (“What does Maya say?”). Change the speaker from the name on the line.
- Length: an unvoiced line is sized from its word count (about two and a half words a second, one second minimum) and rewording re-estimates it. When the real audio lands the line takes its true length and everything after it moves to fit, so lines never overlap.
- Generate per line with its arrow, or Generate all for every written line that is new, stopped or changed (hidden when nothing needs it). Typing costs nothing; a line is billed only when generated. A single line plays back once as soon as it lands.
- Statuses: Generating, Stopped (cancelled or failed), Changed (“The text changed since this line was generated.”), or the audio’s length when it is current. Regenerating a current line asks first; the others go straight away.
- Chat drafts for you: “write a 2-minute interview about home coffee” adds the speakers and the lines, and “give Speaker 2 a warmer intro” edits them. Each request is one undo step.
The cast
The Cast tab holds a card per speaker: a colour (eight swatches, or any colour), the name (renaming also renames their lane), a Model and that model’s voice and settings with the same preview button as the audio workspace. Speed and output format are not offered here; the line’s length is the script’s business. A speaker with no model uses the voice model picked under Models with its default voice, and one piece can mix voices from different providers. Add speaker adds a lane under the last one; removing a speaker with lines confirms (“Their lane and 4 clips on it are removed too. One ⌘Z brings them back.”).
The mix
- Lanes: one per speaker in their colour, then Music and SFX, which stay out of the timeline until they hold a clip. The lane icon is the mute button; clip volume lives in the Clip tab.
- Add offers Line, Music and SFX; the media menus start with an empty layer to generate into, then your project’s audio. Music added by the chat comes in quiet and looping so it sits under the voices. Only audio files can be dropped here; a file dropped on a speaker lane becomes a voice clip.
- Clip tab: the selected clip’s timing, volume, fades and file, the audio counterpart of the Layer tab. Script lines are generated from the script, not from here.
- Playback plays the whole mix with mutes applied, from the transport or by scrubbing.
Exporting the mix
Export mixes every audible lane into one file: MP3 at 192 kbps, M4A (AAC) or WAV (16-bit), as long as the composition. It renders on your machine with the same progress strip and Cancel export, lands in your project folder and shows up in the Audio workspace, where the audio tools can trim or convert it further. With nothing voiced yet the button says “There’s no audio to export yet.”
Recipes
A slideshow with motion
From one still to a sequence
A vertical cut of a landscape piece
⌘Z undoes the switch if you change your mind.A two-voice episode
A narrated explainer
Reference
Keyboard shortcuts
⌘ is Ctrl on Windows and Linux. Keys are ignored while you are typing. The main guide lists every workspace’s shortcuts together.
Editor
⌘Z / ⇧⌘Z- Undo and redo, yours and the chat’s alike
C- Show or hide the side panel
T- Show or hide the timeline
Esc- Deselect; with nothing selected, close the composition
Clips
← / →- Nudge the selected clip a tenth of a second (a second with Shift)
Delete / Backspace- Remove the selected clip
Double-click- Open the Layer tab on a clip
⌘ + scroll- Zoom the timeline (or pinch)
Chat
Enter / Shift+Enter- Send a message / add a line
Troubleshooting
- The + is disabled. “Select a project folder first”: pick a folder from the pill.
- “Pick a Chat model from the Models menu in the toolbar to enable AI editing.” The assistant needs a tool-calling chat model. “Auto couldn’t pick a model” means none of your providers offers one; set one by name.
- A clip says Stopped. Its generation failed or was cancelled; the toast said why. Open the Layer tab and Regenerate, or fix the model under Models.
- “No video model configured — pick one under Models.” Make video, Extend and empty video layers need a video model.
- “The selected video model can’t start from an image.” Make video and Extend (use last frame) need an image-to-video model; “can’t take a source clip” means Extend (use video) needs one that extends footage.
- “Another layer is already rendering.” (or converting, extending) One at a time; wait or Stop it.
- “This layer changed while generating — the new footage was saved to your project.” You edited the layer mid-run; the result is in the Video workspace, add it by hand.
- The chat replied “Skipped n changes”. The document could not take those edits (an extra audio lane, a video clip of its own). The rest applied.
- A video clip will not trim from the left. By design: “Video clips trim from the end”. Trim the source in the Video workspace first if you need a later in-point.
- A video clip plays slowly or repeats. It is longer than its source; toggle Loop in the Layer tab, or reset its length.
- A dashed outline in the export. An empty layer was left on the timeline; generate into it or delete it.
- Export is refused. Exporting needs an active license; see Account & license.
- “There’s no audio to export yet.” Generate at least one line, or add a music or effect clip.
- “Drop audio files here, not image.” The audio editor and audio lanes take audio only; the video editor routes a drop to a lane of the right kind.
- Lines overlap after a regeneration. They should not: a landed line ripples the ones after it. If a take is much longer than expected, shorten the text and regenerate; a Changed badge marks lines whose text moved on.
- I want to change the type. The type is fixed at creation; make a new composition of the other type and reuse the files from your project folder.