Build a complete video or image workflow
Choose a recipe, start from its complete project example, then adapt one concern at a time. Examples show the project inside the REST payload envelope where labelled. Use an API key and free validation before submitting. Rendering spends credits; validation and template dry-runs do not.
Add captions to a clip
Start with the subtitle examples and the complete inspirational video.
- Set the video source to your reachable media URL and bound its source/output timing.
- Add caption text and start/end times under the root
subtitleproperty. Use optional word timing when you need precise alignment; words are automatically timed from the caption text when explicit words are absent. - Choose a caption box position, typography and readable background. Rounded backgrounds belong to the documented subtitle background style fields.
- Validate, render and inspect the first/last caption and scene boundaries. Zvid's renderer consumes supplied captions; do not assume a render automatically transcribes your source audio.
Caption and word timestamps both use absolute seconds on the project timeline. Follow the exact timing rules in Caption and Word.
Create a branded still image
Use the complete request in Rendering images.
- Choose
type: "image", an output size and a supported still-image format. - Replace copy, colors and image sources. Fit variable text deliberately and check its longest likely value.
- Remove unsupported video elements and timeline fields; a video payload cannot be converted by changing
typealone. - Validate, submit to the image endpoint, then retrieve the completed image URL.
Use PNG or WebP when transparency is required. The layout guide explains how anchors and position presets affect placement.
Adapt a product or promotional video
Start from the complete Zvid ad or pizza reel. Keep the displayed fixture and rendered preview paired while studying them; your own edits will create a different result.
- Preserve the timeline and layer structure while replacing copy and source assets.
- Check source aspect ratios, text lengths and available audio/video durations.
- Review the timing after trims, speed changes and transition overlaps.
- Validate the whole composition, render it, and inspect the output before sending it to a customer.
For an assistant, the Creator workflow can discover an appropriate library example and prepare a draft/quote for review.
Turn a list into a slideshow
Use Scenes for a complete scene project, then Dynamic content for iterate and conditions.
Each array item in a scene's iterate creates one scene inside one output. Use condition to show or hide scenes and elements. A bulk request creates multiple outputs. For a reusable stored video template, give each scene a positive explicit duration. Validate both an empty/short dataset and the largest intended dataset so resolved scene counts and duration stay within account limits.
Combine narration and background music
Use the audio item fragments in Audio elements, adding them to the project's or scene's audios array.
Keep narration and music as separate audio items. Set volume, source trims and timeline bounds explicitly; matchDuration follows the containing timeline and loops short source segments to fill it. Narration should supply a bounded duration when it determines scene length. Background music should not accidentally determine the length of an Auto-mode project. Check timing and listen to the rendered output; passing structural validation does not prove an audible mix is balanced.
Render one result per customer or product
Follow Template basics, then Bulk rendering.
Save a template, preview representative variable sets for free, validate costs and limits, then submit the bulk request. Store the batch ID and each item result. Handle partial acceptance and failures per item; do not resubmit successful items as part of an indiscriminate retry. The REST hard cap is 500 items with possible lower account limits; MCP and installed automation integrations can impose lower limits.
Use signed webhooks or bounded polling to deliver only completed media. Keep template design changes separate from variable-data changes so failures can be traced to the correct input.