update

README.md

@@ -104,11 +104,11 @@ python demo.py --model_path /path/to/checkpoint.pt \
 
 # University scene
 python demo.py --model_path /path/to/checkpoint.pt \
-    --image_folder example/university4 --mask_sky
+    --image_folder example/university --mask_sky
 
 # Loop scene (loop closure trajectory)
 python demo.py --model_path /path/to/checkpoint.pt \
-    --image_folder example/loop --mask_sky
+    --image_folder example/loop
 ```
 
 ### Streaming Inference from Images
@@ -197,6 +197,17 @@ If you run into out-of-memory issues, try one (or both) of the following:
 - **`--offload_to_cpu`** — offload per-frame predictions to CPU during inference (on by default; use `--no-offload_to_cpu` only if you have memory to spare).
 - **`--num_scale_frames 2`** — reduce the number of bidirectional scale frames from the default 8 down to 2, which shrinks the activation peak of the initial scale phase.
 
+### Faster Inference
+
+Lower the number of iterative refinement steps in the camera head to trade a small amount of pose accuracy for wall-clock speed:
+
+```bash
+python demo.py --model_path /path/to/checkpoint.pt \
+    --image_folder /path/to/images/ --camera_num_iterations 1
+```
+
+`--camera_num_iterations` defaults to `4`; setting it to `1` skips three refinement passes in the camera head (and shrinks its KV cache by 4×).
+
 # 📜 License
 
 This project is released under the Apache License 2.0. See [LICENSE](LICENSE.txt) file for details.

demo.py

@@ -122,6 +122,7 @@ def load_model(args, device):
         kv_cache_cross_frame_special=True,
         kv_cache_include_scale_frames=True,
         use_sdpa=args.use_sdpa,
+        camera_num_iterations=args.camera_num_iterations,
     )
 
     if args.model_path:
@@ -260,6 +261,9 @@ def main():
         help="Streaming only. Every N-th frame after scale frames is kept as a keyframe. 1 = every frame.",
     )
     parser.add_argument("--kv_cache_sliding_window", type=int, default=64)
+    parser.add_argument("--camera_num_iterations", type=int, default=4,
+                        help="Camera head iterative-refinement steps. Default 4; set 1 for faster inference "
+                             "(skips 3 refinement passes at a small accuracy cost).")
     parser.add_argument("--use_sdpa", action="store_true", default=False,
                         help="Use SDPA backend (no flashinfer needed). Default: FlashInfer")
     parser.add_argument(