Skip to content

feat(gallery): add Depth Anything V2 models + bump native version - #10413

Merged
mudler merged 2 commits into
masterfrom
feat/depth-anything-2-gallery
Jun 20, 2026
Merged

feat(gallery): add Depth Anything V2 models + bump native version#10413
mudler merged 2 commits into
masterfrom
feat/depth-anything-2-gallery

Conversation

@localai-bot

Copy link
Copy Markdown
Collaborator

Summary

Adds Depth Anything V2 (DA2) support to the depth-anything backend.

  • DA2 is depth-only — no camera pose, no confidence (unlike DA3). It ships in two flavours: relative (relative inverse depth) and metric (depth in metres: Hypersim indoor max_depth 20, VKITTI outdoor max_depth 80).
  • No Go backend changes — the backend is model-agnostic. This PR only bumps the pinned native version and adds gallery entries.

Changes

  1. backend/go/depth-anything-cpp/Makefile — pin DEPTHANYTHING_VERSION to e3dec57f13a52366bbc4f279ef44804915960a6b, the depth-anything.cpp commit that carries the DA2 engine + C-API routing. It is kept alive by the upstream tag da2-support so it survives a squash-merge.
  2. gallery/index.yaml — add 12 DA2 entries after the existing depth-anything-3 block:
    • Relative: depth-anything-2-base (q4_k default), -base-q8_0, -base-f16, -base-f32, -small (ViT-S f32), -large (ViT-L f32)
    • Metric (Hypersim indoor): -metric-hypersim-small/base/large (q4_k)
    • Metric (VKITTI outdoor): -metric-vkitti-small/base/large (q4_k)
    • Metric entries carry the metric-depth tag; none carry camera-pose. Descriptions recommend GenerateImage / the Depth endpoint (no pose).

Notes

  • The depth-anything.cpp side is PR llama-cli: command not found #1 (mudler/depth-anything.cpp). Once it merges, the pin here can be repointed to the master merge commit.
  • All 12 sha256 values were verified against the published SHA256SUMS.
  • Gallery YAML validated (anchors/merges resolve).

Assisted-by: Claude:claude-opus-4-8

mudler added 2 commits June 20, 2026 09:18
Add Depth Anything V2 (DA2) support to the depth-anything backend. DA2 is
depth-only (no camera pose, no confidence) and ships both relative
(relative inverse depth) and metric (depth in metres) variants. The Go
backend is model-agnostic, so no backend code changes are required — only
a native version bump and new gallery entries.

- backend/go/depth-anything-cpp/Makefile: pin DEPTHANYTHING_VERSION to the
  depth-anything.cpp commit that adds the DA2 engine + C-API routing
  (e3dec57f13a52366bbc4f279ef44804915960a6b, kept alive by the upstream tag
  da2-support so it survives a squash-merge).
- gallery/index.yaml: add 12 DA2 entries (4 base quants, small, large, plus
  Hypersim indoor and VKITTI outdoor metric models in S/B/L). Metric models
  carry the metric-depth tag; none carry camera-pose.

Assisted-by: Claude:claude-opus-4-8
PR #1 (mudler/depth-anything.cpp) merged to master as f4e17de (squash); repoint
the pin from the pre-merge commit to the canonical master commit.

Assisted-by: Claude:claude-opus-4-8
@mudler
mudler merged commit e19c43c into master Jun 20, 2026
65 of 66 checks passed
@mudler
mudler deleted the feat/depth-anything-2-gallery branch June 20, 2026 12:56
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants