monster/modelbeast
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

benchmarks: Qwen-Image-Layered working on MLX (m1ultra) — 4-layer RGBA decomposition verified, PR302 fork vendored

Browse Source 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
m3ultra 2026-07-19 20:40:02 +10:00
parent 79bfd713c1
commit 64afe54474
1 changed files with 20 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
BENCHMARKS.md
View File

@ -329,3 +329,23 @@ quality gap closed for albedo. Not yet carried: metallic/roughness maps
(vertex colors are albedo-only) — MR needs the UV path (Option B) or a
COLOR_1 convention. MESHGOD finish-farm verts→texture bake = the
remaining Option-A step for game-budget assets.
## 2026-07-19 — Qwen-Image-Layered on MLX (m1ultra) — WORKING
New capability: image → N editable RGBA layers, fully local. Model:
Qwen/Qwen-Image-Layered (20B, BF16, Apache 2.0, 54GB in m1 HF cache).
Runtime: mflux PR#302 fork (unmerged upstream; vendored at
monster/Qwen-Image-Layered-MRP-MLX, pin tag pr302-head-a255e4f, isolated
venv ~/qwen-layered-staging/mflux-layered/.venv — production mflux venv
untouched). CLI: `mflux-generate-qwen-layered`.
Smoke (anatomy test image, 4 layers, 640, q8-on-the-fly): **1180.8s wall
first-run** (dominated by 20B load + live quantization), 27.1GB peak.
Output verified: 4 true RGBA PNGs — bg layer, atmosphere layer, clean
figure cutout (74.5% transparent, crisp to the fingers), detail layer.
The figure-cutout layer doubles as a high-end matting result — relevant
to the MESHGOD prep frontline for hero assets.
Optimization in progress: `mflux-save --quantize 8` pre-baked model →
skips load+quant cost; timed gen-only rerun to follow. Next: 1024-res,
q4/q6 quality ladder, m3ultra cross-bench.