1338.7338 [darktable_ai] dt_ai_env_init start.
  1338.7351 [darktable_ai] discovered: mask sam2.1 hiera small (mask-object-sam21-small, backend=onnx)
  1338.7362 [darktable_ai] discovered: upscale realplksr (upscale-realplksr, backend=onnx)
  1338.7365 [darktable_ai] discovered: mask segnext vitb-sax2 hq (mask-object-segnext-b2hq, backend=onnx)
  1338.7368 [darktable_ai] discovered: denoise nind (denoise-nind, backend=onnx)
  1338.7371 [darktable_ai] discovered: raw denoise nind (rawdenoise-nind, backend=onnx)
  1338.7371 [darktable_ai] attempting to enable Apple CoreML...
  1338.7422 [darktable_ai] Apple CoreML enabled successfully.
  1338.7422 [darktable_ai] provider auto -> CoreML for mask-object-segnext-b2hq
  1338.7422 [darktable_ai] loading: /Users/stevethomson/.local/share/darktable/models/mask-object-segnext-b2hq/encoder.onnx
  1338.7422 [darktable_ai] CoreML format: NeuralNetwork
  1338.7422 [darktable_ai] attempting to enable Apple CoreML (V2)...
  1338.7428 [darktable_ai] Apple CoreML (V2) enabled successfully.
  1340.4127 [darktable_ai] loading: /Users/stevethomson/.local/share/darktable/models/mask-object-segnext-b2hq/decoder.onnx
  1340.4127 [darktable_ai] using CPU only (no hardware acceleration)
  1340.5040 [segmentation] tensor routing: dec[0] <- enc[0]
  1340.5041 [segmentation] model loaded: mask-object-segnext-b2hq [SegNext] (enc_outputs=1, num_masks=1, dec_dims=1024x1024, prev_mask_dim=1024)
  1340.5891 [object mask] rendering 1536x1025 (scale=0.255) for encoding...
  1341.3884 [segmentation] encoding image 1536x1025 (scale=0.667)...
  1365.6570 [segmentation] image encoded successfully (24.245s)
  1365.6844 [segmentation] disk cache: saved imgid 345 (1536x1025)
  1365.6845 [segmentation] warming up decoder...
  1367.9119 [segmentation] decoder warmup done in 2.227s (real embeddings)
  1376.2243 [segmentation] has_prev_mask=0 (no previous mask), n_points=1
  1378.0067 [segmentation] mask computed (1.782s)
  1378.0120 [segmentation] resized mask (1024x1024 -> 1536x1025, scale=0.6667)
  1378.0210 [segmentation] has_prev_mask=1, prev_mask range=[-10.703, 2.959], n_points=2
  1378.9187 [segmentation] mask computed (0.898s)
  1378.9238 [segmentation] resized mask (1024x1024 -> 1536x1025, scale=0.6667)
  1379.2208 [object mask] CRF refinement: 1536x1025 (0.30s)
  1386.4112 [segmentation] has_prev_mask=1, prev_mask range=[-2.537, 1.540], n_points=2
  1387.4671 [segmentation] mask computed (1.056s)
  1387.4738 [segmentation] resized mask (1024x1024 -> 1536x1025, scale=0.6667)
  1387.4810 [segmentation] has_prev_mask=1, prev_mask range=[-7.978, 3.013], n_points=3
  1388.4028 [segmentation] mask computed (0.922s)
  1388.4080 [segmentation] resized mask (1024x1024 -> 1536x1025, scale=0.6667)
  1388.7202 [object mask] CRF refinement: 1536x1025 (0.31s)
  1392.0952 [segmentation] has_prev_mask=1, prev_mask range=[-4.665, 2.838], n_points=3
  1393.0027 [segmentation] mask computed (0.907s)
  1393.0076 [segmentation] resized mask (1024x1024 -> 1536x1025, scale=0.6667)
  1393.0151 [segmentation] has_prev_mask=1, prev_mask range=[-4.653, 2.762], n_points=4
  1393.9470 [segmentation] mask computed (0.932s)
  1393.9520 [segmentation] resized mask (1024x1024 -> 1536x1025, scale=0.6667)
  1394.2169 [object mask] CRF refinement: 1536x1025 (0.26s)