Thank you for LaMa — the inpainting quality is excellent and it runs well on-device.
I'm preparing to ship an Android app that runs big-lama locally for photo restoration, and I want to be sure I'm respecting your terms. I can't determine what governs the pretrained weights:
The repository LICENSE is Apache 2.0 (Copyright 2021 Samsung Research), but reads as covering the code.
Several projects, including mine, reference a LICENSE-MODELS.md for the weights, but I can't find that file in this repository — the raw URL returns 404 and it isn't in the file listing. Did it exist previously?
The smartywu/big-lama mirror on Hugging Face is tagged apache-2.0, but that appears to be the uploader's tag rather than yours.
big-lama was trained on Places2, and I'm unclear whether that imposes any restriction downstream.
Could you clarify whether the published big-lama weights may be used commercially, and if so under which licence? A one-line answer here would settle it for everyone hitting the same question.
Thanks for your time.
Thank you for LaMa — the inpainting quality is excellent and it runs well on-device.
I'm preparing to ship an Android app that runs big-lama locally for photo restoration, and I want to be sure I'm respecting your terms. I can't determine what governs the pretrained weights:
The repository LICENSE is Apache 2.0 (Copyright 2021 Samsung Research), but reads as covering the code.
Several projects, including mine, reference a LICENSE-MODELS.md for the weights, but I can't find that file in this repository — the raw URL returns 404 and it isn't in the file listing. Did it exist previously?
The smartywu/big-lama mirror on Hugging Face is tagged apache-2.0, but that appears to be the uploader's tag rather than yours.
big-lama was trained on Places2, and I'm unclear whether that imposes any restriction downstream.
Could you clarify whether the published big-lama weights may be used commercially, and if so under which licence? A one-line answer here would settle it for everyone hitting the same question.
Thanks for your time.