Conversation
The page taught the pre-release timeline: per-clip `color`, `blur`, `chromaKey` and `filter` fields, `animateTo` for movement, shapes that placed themselves with their own `position`, `size` and `anchor`, and `output.frameTime` for a still. Ten of the twelve complete examples on it used at least one field that has moved. It now teaches one `effects` list on a clip, a track or the timeline, keyframes with 33 easings on anything that moves, and one `transform` that places every clip including text and shapes. New sections cover timers, subtitles from cues or an .srt file, Lottie animations, masks and LUTs, the `engine` entry, `output.range`, a transparent background, and the free `validate` endpoint. The transition table carries all six kinds, with wipes by shape or by your own greyscale map. Every JSON example was run through the API's own validator: 30 of them, no refusals, no warnings, nothing left in the old shape. That caught three examples I had written wrong, including an engine parameter that takes a number and a transparent background on mp4, which has no alpha.
Both ship with the compose release: a free dry run of a compose submit, and reading one effect's parameters instead of the whole reference.
…reference One page of 970 lines becomes eleven learn pages and thirteen generated reference pages, matching how Creatomate and JSON2Video both organise this: one reference page per asset kind. jobs/compose stays the overview so every existing link to it still works. The reference pages are generated from the schema the API validates with, so a field table cannot drift from what a render accepts. Effect and transition demos play the before and after clips already on the CDN. Both groups are marked expanded, so opening a reference page no longer folds the rest of the section away.
The before and after pair on that page was the engine escape-hatch demo, which renders frei0r.cartoon, captioned as a keyframed scale. That is why the result looked flat and low quality: it was a cartoon filter, not a Ken Burns push. No keyframe demo exists in the showcase set, so the page teaches from the JSON until one is rendered. A wrong clip is worse than no clip. Also removes a lead that repeated the section's own opening line.
These pages are for a quick explanation and a look at the result, not a second reference. Three of them carried no picture at all. Motion gets a real Ken Burns demo, rendered through the compiler a job uses: a still photograph scaled 1 to 1.15 with easeInOutSine, beside the untouched still. Text gets the timer, subtitles and Lottie clips, which were already rendered and sitting unused. Timeline gets a diagram of the music-bed example, because length "end" is the hardest thing on that page to picture from prose.
fonts proves the claim it makes: a display family on the headline, Arabic and Devanagari shaping through the bundled fallbacks with no font file uploaded. output gains the fit comparison, a wide photograph in a phone-shaped box, which is the framing question readers actually get wrong.
A vertical render with the title clipped at the top and the caption jammed at the bottom, beside the same two lines inside the safe area. The last learn page without a picture now has one.
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Do not merge until the compose release ships. This documents the timeline shape on
feat/compose-revampin the monorepo, which is not onmainand not deployed. Merging early would publish docs for an API that is not live.Why
The page taught the pre-release shape: per-clip
color,blur,chromaKeyandfilter,animateTofor movement, shapes placing themselves with their ownposition/size/anchor, andoutput.frameTime. Ten of its twelve complete examples used at least one field that has moved.What it teaches now
effectslist, on a clip, a track or the whole timeline, ordered, withstart/lengthand keyframed parameters. 34 effects grouped, linking to the rendered effects reference.transformthat places every clip, text and shapes included, withsizeboxes,fitmodes includingcoverandblur, andcorners..srt/.vttfile), Lottie animations, masks and LUTs, theengineentry,output.range, a transparent background, and the freevalidateendpoint.pushand wipes by shape or by your own greyscale map.Verification
Every JSON example was run through the API's own validator (the same
validateComposeDocumentthe endpoint uses): 30 examples, no refusals, no warnings, nothing in the old shape. It caught three I had written wrong, including an engine parameter that takes a number rather than a name, and a transparent background on mp4, which has no alpha.Still owed separately: the MCP page needs
validate_composeand theeffectargument onlist_job_types.