Light that moves with you
Adjust direction and elevation, then fine-tune the softness, intensity and fill of your key light.
Change the mood of any photo with directional lighting, soft studio effects and precise highlight and shadow control — right in your browser.
Choose a photograph, shape its light and compare the result.
Drag a photo here or choose one to start editing.
High-resolution local export. Your original is never overwritten.
Adjust direction and elevation, then fine-tune the softness, intensity and fill of your key light.
Use optional on-device person segmentation or paint a subject mask to focus lighting on selected parts of the image.
Images are decoded, adjusted and exported in your browser. Optional AI downloads model software, not your image.
No. It uses image-space luminance gradients to approximate directional lighting; it cannot recover true depth, cast accurate object shadows, or reveal detail hidden in the original exposure. Optional AI detects portrait subjects only.
No photo is uploaded by this tool. When AI portrait selection is enabled, the first Apply downloads MediaPipe software and its model from third-party CDNs. Those providers may receive standard network metadata, and MediaPipe may collect runtime metrics according to its own notice. The photograph itself is processed locally.
Yes. Entire-photo lighting and manual masks work with most image types. The optional AI selection is trained to identify people rather than general objects.
Preview is resized for interactive performance. Exports are rendered from the original source up to the documented maximum. Lighting is approximate and fine details may vary slightly with resolution.
Open a supported photograph, choose a lighting look and adjust direction, tone or the local spotlight. Target the whole image, a painted region or an optional AI person mask. Apply, compare and export; reusable JSON contains settings rather than photos or masks.
Lighting uses deterministic image-space approximations, not true depth, 3D reconstruction or generative scene synthesis. It cannot recover occluded textures or create physically accurate new shadows. Optional person segmentation downloads MediaPipe assets and has external request/privacy boundaries. File/preview/export limits and device memory apply. JPEG uses a white matte and most source metadata is discarded.
References: MediaPipe Image Segmenter
Adjust image-space lighting direction, tone and softness, then compare the whole-image effect, a painted region or an optional person mask.
Reusable JSON stores settings rather than the photo or mask. Exported JPEG uses a white matte and most source metadata is discarded; keep the original separately.
Authored example: On a fictional portrait, compare a whole-image light adjustment with a localized spotlight and inspect the background for unwanted changes. This is an authored review exercise, not a physically accurate lighting demonstration.
Limits and model handling: Lighting uses deterministic image-space approximations, not inferred depth, 3D reconstruction or generative scene synthesis. It cannot recover occluded texture or create physically correct new shadows. Optional person segmentation downloads MediaPipe assets and makes external requests.
No. It applies image-space lighting approximations and cannot infer true geometry or create physically accurate new shadows.
Yes. The optional MediaPipe person segmentation loads external runtime/model assets; the manual and whole-image lighting controls are separate.