Slideshow
2–100 images over one optional audio track → MP4 — locally rendered, no AI model, zero credits.
Overview
The Slideshow node is the N-image companion of Still to Video: instead of chaining N single-image nodes into Combine Videos, one node takes an ordered set of images and renders the whole sequence — with per-slide motion, transitions, and a timing model anchored to the audio. It runs on the platform’s local media engine (FFmpeg) — no provider, no GPU — and costs 0 credits.
For a single image, use Still to Video — same output, no list needed.
Feeding it images
The images input takes an ordered set:
- A List node — wire its image column and set the edge to Bundle (“all items at once”); every row becomes a slide, in row order.
- Direct connections — wire 2–100 image nodes straight in; wire order is slide order.
The transition input takes the Transition parameter node — its pick sets the transition type (mapped onto the local blend vocabulary; unmappable cinematic picks fall back to a hard cut). Nothing wired = cut.
Timing model
| Case | Behavior |
|---|---|
| Audio wired (primary) | Total duration is the audio’s duration — never cropped, the video never ends before it. Slides split it equally. |
Per-slide overrides (API/SDK/MCP imageDurations) |
Pinned rows keep their value; the remainder splits equally across auto rows. |
| All rows pinned, sum ≠ audio | Everything scales proportionally to fit the audio, and the applied factor is disclosed on the node and in the job output. |
| No audio | Total = N × Per Image (default 3s). The output is silent — no audio track at all. |
Transitions consume time from the outgoing slide, never the incoming one — so the total duration stays exact.
Configuration
| Field | Type | Default | Description |
|---|---|---|---|
| Per Image | Number (0.5–60s) | 3 | Seconds per slide — used only when no audio is wired |
| Transition duration | Number (0–5s) | 0.5 | Blend length; the transition type comes from the wired Transition parameter node (unwired = cut) |
| Motion | Select | none | none, zoom-in, zoom-out, ken-burns, alternate (flips zoom in/out per slide — avoids a uniform mechanical push) |
| Intensity | Number (1–10) | 3 | Motion strength. Ignored when Motion is none |
| Resolution | Select | 1080p | 720p, 1080p, 4K |
| Aspect Ratio | Select | 16:9 | 16:9, 9:16, 1:1, 4:3 |
| FPS | Select | 30 | 24 or 30 |
| Fit | Select | cover | cover crops each still to fill; contain letterboxes with the pad color |
| Pad Color | Color | #000000 | Bar color when Fit is contain (disabled otherwise) |
Credits
0 credits — always. No provider cost, no formula, no per-second component.
Inputs & Outputs
Inputs: Images (2–100, ordered — required) · Audio (optional — sets the length) · Transition (optional — the Transition parameter node) Outputs: Video (MP4, H.264; AAC audio when an audio track is wired, silent otherwise)
Limits
- Fewer than 2 images → a clear error pointing at Still to Video.
- More than 100 → an error rather than an hours-long render; trim the set upstream.
Best Practices
alternateat intensity 2–4 is the fastest way to make a photo reel feel edited rather than mechanical- Keep transition duration well under your shortest slide — it blends inside the outgoing slide’s time
- With
contain, the still moves inside the letterbox; the bars stay static - 4K with motion is the slow path (every slide renders through the motion pipeline);
nonestays fast at any size
Common Use Cases
- Narrated photo essays: a List of stills + a voiceover track
- Music-backed reels: product shots or event photos over a generated track
- Storyboard animatics: Generate Image outputs in order, dissolves between beats
Tips
- Chain: Generate Image ×N → List (or direct wires) → Slideshow → Combine Videos to intercut with real footage
- Per-slide durations are available through the API/SDK/MCP (
imageDurations,null= auto); the List node cannot author them yet - The design treats slide width as duration on the node’s strip — what you see is the real timing