diff --git a/tools/mtmd/mtmd-image.cpp b/tools/mtmd/mtmd-image.cpp
index e795b47e06..fac55ef8e3 100644
--- a/tools/mtmd/mtmd-image.cpp
+++ b/tools/mtmd/mtmd-image.cpp
@@ -1606,12 +1606,19 @@ mtmd_image_preproc_out mtmd_image_preprocessor_granite::preprocess(const clip_im
 
     const clip_image_size orig_size = img.get_size();
     const int             tile_size = hparams.image_size;
+    GGML_ASSERT(tile_size > 0);
 
     // llava-next always encodes an overview plus a grid of tiles, even for small images
     const clip_image_size refined_size = select_best_resolution(orig_size, hparams.image_res_candidates);
     const int             grid_x       = refined_size.width  / tile_size;
     const int             grid_y       = refined_size.height / tile_size;
 
+    // the stacked image is tile_size * (1 + grid_x*grid_y) tall, it must fit in an int
+    const int64_t stacked_h = (int64_t) tile_size * (1 + (int64_t) grid_x * grid_y);
+    if (grid_x < 0 || grid_y < 0 || stacked_h > std::numeric_limits<int>::max()) {
+        throw std::runtime_error("invalid image grid, check image_grid_pinpoints in the mmproj file");
+    }
+
     clip_image_u8 overview;
     img_tool::resize(img, overview, {tile_size, tile_size}, hparams.image_resize_algo_ov,
                         hparams.image_pad_ov, hparams.image_pad_color_ov);