octimot
Direct export of story as timeline via Davinci Resolve API
Describe the bug Currently, StoryToolkitAI only allows exporting stories to Resolve via EDL or FCPXML. There is no direct way to connect to DaVinci Resolve and create out of a Storytoolkitai story a new Davinci timeline from chosen Davinci media pool clips (or already composed clips like multicamclip).
Expected functionality would be:
- After linking StoryToolkitAI with Resolve (API integration), the user should be able to select clips inside StoryToolkitAI → send them directly into new Resolve’s timeline (or even existing).
- Right now, this workflow doesn’t exist (or is not documented), which forces an indirect export/import step.
To Reproduce Steps to reproduce the behavior:
- Link StoryToolkitAI with DaVinci Resolve with the API.
- Only “export EDL / FCPXML” option is available. But that needs source_file_path set, what is additional unnecessary step in terms of Davinci integration.
Expected behavior After linking with Resolve, I expect to be able to:
- Choose clips from the Resolve media pool inside StoryToolkitAI.
- Generate a Resolve timeline directly (without exporting/importing XML/EDL). This would allow building a story structure seamlessly inside Resolve.
Log file N/A (no crash or error, but missing feature).
System (please complete the following information):
- OS: MacOS 15.4.1
- Python versions installed: 3.11.13
- StoryToolkitAI Version: 0.25.1 (installation from source)
Additional context It’s unclear whether this is a missing feature or a bug in the Resolve API integration. If it’s not yet implemented, could you clarify whether direct Resolve timeline creation from StoryToolkitAI is planned?
Little update:
I see that my problem arises from importing and converting a srt file to storytoolkit ai. The audio_file_path field is not set and thats of course not being propaged to edl. This is where export edl function starts to complain. By setting it manually in the .transcription.json file helps to prevents the error.
As I understand it is also possible to set timeline_name and that one is being used instead of the source_file_path in the story file (that itself takes audio_file_path from transcription.json)
It would be nice to have the freedom to decide which clip I'm referencing to without going through the transcribtion processing with whisper inside of storytoolkitai. But even when, I want at the end decide which clip/multicamclip I'm actually referecing to.
p.s. if my understanding is correct of the programflow of storytoolkitait also thats a little bit different story, in case the other option is not working doing it indeed manually, but here are also big trouble right now
Ok I see my confusion is big about how storytoolkitai is working :) Some steps are really not obvious and the program sometimes not guide me if I'm wrong.
From my conversation with copilot:
How StoryToolkitAI Handles This StoryToolkitAI already supports this workflow when you use the Resolve integration for transcription/ingest:
When you transcribe a timeline directly from Resolve using the Resolve API, the tool fetches the current timeline's name and writes it to the timeline_name field in your .transcription.json. This is handled in the Resolve integration code, such as MotsResolve.render_timeline and related methods. The field is then used for EDL/XML export if you enable "Use Timelines".
I see that it is somehow already done with integrations maybe like semi-automatically but sometimes something is not set and only god knows where and why :)
Hey!
If I read you correctly, I think you're referring to two different flows:
- StoryToolkitAI to Resolve timeline -> not supported
- Resolve timeline to StoryToolkitAI (for transcription) -> supported as you mentioned in your last post
The integration from StoryToolkitAI to Resolve timeline (1.) was not possible for Resolve Studio 18 and 19 due to the limitations of the Resolve API. In other words, Resolve doesn't allow you to edit the timeline items. If I'm not mistaken, I think it's the same for Resolve Studio 20, although I haven't checked the details (API options can seen here: https://gist.github.com/X-Raym/2f2bf453fc481b9cca624d7ca0e19de8)
If option 1. would be possible, I'd be happy to implement.
Cheers
