Decouple deterministic smoke from GPU quota

Release Docker Simam3D 0.4.8

Release Docker Simam3D 0.4.7

Sync Simam3D 0.4.6 application

perf: reduce hosted fusion export cap

perf: bound DA3 resolution and fusion output

fix: optimize bounded direct Gaussian export

fix: bound direct Gaussian export size

update: target Gaussian-capable DA3 Nested checkpoint

release: Simam3D 0.4 direct DA3 Gaussians

feat: export DA3 direct Gaussian parameters

update: use refreshed DA3-LARGE-1.1 checkpoint

feat: preserve full fusion evidence in manifests

fix: use modern matplotlib colormap API

Harden optional Spaces GPU decorator fallback

Record software versions in failure reports

Serialize heavyweight inference requests

Test DA3 package layout compatibility

Support DA3 package layout variants

Add downloadable failure reports

Persist structured failure manifests

Expose DA3 API preflight diagnostics

Harden Gradio multi-view upload path handling

Release inactive backend before GPU model switching

Improve hosted error visibility and startup version banner

Add relative-depth reprojection consistency metrics

Add fused-point projection coverage diagnostics

Record exact DA3 fallback exception in manifest

Centralize and test DA3 output contract normalization

Validate DA3 prediction contract before fusion

Add explicit provenance claims contract to manifests

Fingerprint exported artifacts in schema v2 manifest

Add explicit CPU export pipeline smoke test

Add depth-normal 3D light direction estimator

Add deterministic fused-point turntable preview

Add uncertainty mapping and next-view reveal planning

Add What is Behind the Image lighting reveal prototype

Make DA3 optional and add diagnostics

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub