diff --git a/pyproject.toml b/pyproject.toml
index ba645f8..0345e69 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "m2stitch"
-version = "0.1.1"
+version = "0.2.0"
 description = "M2Stitch"
 authors = ["Yohsuke Fukai <ysk@yfukai.net>"]
 license = "MIT"
diff --git a/src/m2stitch/_stage_model.py b/src/m2stitch/_stage_model.py
index ec44d1f..238f1a9 100644
--- a/src/m2stitch/_stage_model.py
+++ b/src/m2stitch/_stage_model.py
@@ -124,7 +124,9 @@ def compute_image_overlap(
                     return tuple(map(float, model["x"]))
                 elif model["fun"] < best_model["fun"]:
                     best_model = model
-    assert not best_model is None
+    assert (
+        not best_model is None
+    ), "Overlap model estimation failed, try raising the value of overlap_prob_uniform_threshold"  # noqa :
     #    best_model_params : Tuple[Float,Float,Float] = tuple(best_model["x"])
     assert len(best_model["x"]) == 3
     return tuple(map(float, best_model["x"]))
diff --git a/src/m2stitch/stitching.py b/src/m2stitch/stitching.py
index be26dde..1e7d347 100644
--- a/src/m2stitch/stitching.py
+++ b/src/m2stitch/stitching.py
@@ -27,6 +27,7 @@ def stitch_images(
     rows: Sequence[Any],
     cols: Sequence[Any],
     pou: Float = 3,
+    overlap_prob_uniform_threshold: Float = 80,
     full_output: bool = False,
 ) -> Tuple[pd.DataFrame, dict]:
     """Compute image positions for stitching.
@@ -45,6 +46,11 @@ def stitch_images(
     pou : Float, default 3
         the "percent overlap uncertainty" parameter
 
+    overlap_prob_uniform_threshold : Float, default 80
+        the upper threshold for "uniform probability".
+        raise this value to ease assumption
+        that the displacement is following non-uniform distribution.
+
     full_output : bool, default False
         if True, returns the full computation result in the pd.DataFrame
 
@@ -105,9 +111,13 @@ def get_index(row, col):
             for j, key in enumerate(["ncc", "x", "y"]):
                 grid.loc[i2, f"{direction}_{key}_first"] = max_peak[j]
 
-    prob_uniform_n, mu_n, sigma_n = compute_image_overlap(grid, "top", W, H)
+    prob_uniform_n, mu_n, sigma_n = compute_image_overlap(
+        grid, "top", W, H, prob_uniform_threshold=overlap_prob_uniform_threshold
+    )
     overlap_n = 100 - mu_n
-    prob_uniform_w, mu_w, sigma_w = compute_image_overlap(grid, "left", W, H)
+    prob_uniform_w, mu_w, sigma_w = compute_image_overlap(
+        grid, "left", W, H, prob_uniform_threshold=overlap_prob_uniform_threshold
+    )
     overlap_w = 100 - mu_w
 
     overlap_n = np.clip(overlap_n, pou, 100 - pou)