Transparency
Data & AI
What stays on your device, what Lumen sends for generation, and how temporary files move through the service.
Last updated August 24, 2026
The Lumen path
- 01You choose a shot
No project image is uploaded merely by opening Lumen.
- 02Scouter creates a durable job
The backend authenticates the request, reserves a generation credit, and stores a temporary source object.
- 03The provider relights the frame
The image and instruction are sent to Alibaba Cloud Model Studio using the Qwen or Wan route assigned to that job. Provider keys never ship inside the app.
- 04The result returns to your project
The app downloads the generated version, writes it locally, and confirms receipt.
- 05Temporary images are removed
Source uploads are deleted at terminal completion. Results are deleted after receipt or the 24-hour fallback deadline.
What the model receives
The model receives the selected image and the lighting instruction. It may also receive technical context such as camera, format, lens, focal length, aspect ratio, look, focus, exposure, white balance, and image dimensions. It does not receive your full project library or precise location metadata.
Provider processing
Lumen currently uses Alibaba Cloud Model Studio in the China (Beijing) service region. Alibaba Cloud states that Model Studio customer content is not used to develop or improve its models without separate consent. For an asynchronous Wan generation, the provider keeps task status and the temporary result URL available for up to 24 hours so Scouter can recover interrupted processing. Scouter downloads the result promptly and keeps its own source and result copies only under the temporary retention rules described here.
What AI output means
Lumen images are visual planning references, not photometric simulations. They can invent, omit, or alter details and may not predict real fixture output, exposure, safety, weather, access, or production feasibility. A qualified crew must validate the actual lighting plan.
Provider changes
Scouter's backend keeps each generation bound to its assigned provider and model so existing jobs can finish safely during a provider transition. If a different provider or materially different processing terms are introduced, this page and the privacy policy will be updated before the change applies to users.