Pretrained model weights (v1)

Pretrained weights for the Face2HDRI two-stage pipeline.

Asset	Stage	Model	Size
stage1_sphere_query.pt	Stage 1	spherical-query transformer (frozen DINOv3-S + ViT-Ti), peak fine-tuned	134 MB
stage2_flow_prior.pt	Stage 2	flow-matching detail prior	213 MB

Together these are everything needed for end-to-end portrait → HDRI inference.
The frozen DINOv3-S backbone is fetched automatically by timm on first load.

Usage

uv run python -m src.scripts.sample_hdri_prior \
    --stage1 stage1_sphere_query.pt \
    --stage1-kwargs '{"use_dino": true}' \
    --prior  stage2_flow_prior.pt \
    --guidance-scale 2.0

SHA-256

6f15e8da34513a8c47e6d504dba6f7cc078c8af361b9b0d2aa1bb4429bf9fb6b  stage1_sphere_query.pt
b5abbe0a84a1bdd240d872aea3ddd14270c6e8636b149a729591fcaa961ff8dd  stage2_flow_prior.pt