diff --git a/NAMESPACE b/NAMESPACE
index 6c202a4a..2dae38fa 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -176,6 +176,9 @@ S3method(transform_to_tensor,list)
 S3method(transform_to_tensor,matrix)
 S3method(transform_vflip,default)
 S3method(transform_vflip,torch_tensor)
+S3method(vision_make_grid,"magick-image")
+S3method(vision_make_grid,default)
+S3method(vision_make_grid,torch_tensor)
 export(base_loader)
 export(batched_nms)
 export(box_area)
diff --git a/NEWS.md b/NEWS.md
index 4321e546..2a1f7c25 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -39,6 +39,7 @@
 
 ## Bug fixes and improvements
 
+* `vision_make_grid()` now accepts multiple 3D tensors with mixed uint8 and float dtype (#398).
 * `transform_random_affine()` now accepts a bare number for `shear`. It used to widen `degrees`
   instead of `shear`, which left the shear range incomplete and made the sampling fail (#390).
 * `item_transform_rotate()` no longer truncates the rotation angle to a whole number of degrees
diff --git a/R/collection-rf100-biology.R b/R/collection-rf100-biology.R
index 8c86b27e..ed88c261 100644
--- a/R/collection-rf100-biology.R
+++ b/R/collection-rf100-biology.R
@@ -104,6 +104,7 @@ rf100_biology_collection <- torch::dataset(
     size = c(81, 24, 12, 6.4, 1.8, 0.9 , 65.1,17.9,9,  0.3, 0.1, 0.05,
              1.4, 2.5, 0.8,   62, 16.8, 9,
              19, 5.3, 2.7,   69,9.0, 5.7,   192,55.6, 28,
-             45, 13, 4.5) * 1e6
+             45, 13, 4.5) * 1e6,
+    stringsAsFactors = FALSE
   )
 )
diff --git a/R/collection-rf100-damage.R b/R/collection-rf100-damage.R
index 56c197f6..8834a39e 100644
--- a/R/collection-rf100-damage.R
+++ b/R/collection-rf100-damage.R
@@ -54,6 +54,7 @@ rf100_damage_collection <- torch::dataset(
       # asbestos
       "f53af703c4ce594c3950d7e732003f2d",      "8e99904ce49e7f0e830735fb22986868",      "4fef507d057690d1a55fa043696248cc"
     ),
-    size = c(21.5, 5.7, 1.4,   8.5, 2.3, 1.5,  28, 7.8, 4) * 1e6
+    size = c(21.5, 5.7, 1.4,   8.5, 2.3, 1.5,  28, 7.8, 4) * 1e6,
+    stringsAsFactors = FALSE
   )
 )
diff --git a/R/collection-rf100-doc.R b/R/collection-rf100-doc.R
index 9e4003de..b5ce289f 100644
--- a/R/collection-rf100-doc.R
+++ b/R/collection-rf100-doc.R
@@ -133,7 +133,8 @@ rf100_document_collection <- torch::dataset(
       "ca0b96eb696da512eed629895a433bad"
     ),
 
-    size = c(rep(50, 18), 32, 9, 5, 108, 32, 23) * 1e6 # placeholder; optional
+    size = c(rep(50, 18), 32, 9, 5, 108, 32, 23) * 1e6, # placeholder; optional
+    stringsAsFactors = FALSE
   ),
 
   initialize = function(
diff --git a/R/dataset-pascal.R b/R/dataset-pascal.R
index 57a32860..a129411a 100644
--- a/R/dataset-pascal.R
+++ b/R/dataset-pascal.R
@@ -112,7 +112,8 @@ pascal_segmentation_dataset <- torch::dataset(
             "da459979d0c395079b5c75ee67908abb",
             "6c3384ef61512963050cb5d687e5bf1e",
             "6cd6e144f989b92b3379bac3b3de84fd"),
-    size = c("440 MB", "440 MB", "550 MB", "890 MB", "1.3 GB", "1.7 GB", "1.9 GB")
+    size = c("440 MB", "440 MB", "550 MB", "890 MB", "1.3 GB", "1.7 GB", "1.9 GB"),
+    stringsAsFactors = FALSE
   ),
   classes = pascal_voc_classes(),
   voc_colormap = c(
diff --git a/R/dataset-plankton.R b/R/dataset-plankton.R
index b85b7bb4..f0b088b0 100644
--- a/R/dataset-plankton.R
+++ b/R/dataset-plankton.R
@@ -56,7 +56,8 @@ whoi_small_plankton_dataset <- torch::dataset(
             "170921030aee26f9676725a0c55a4420",
             "332ebc8b822058cd98f778099927c50e",
             "f0747ae16fc7cd6946ea54c3fe1f30b4"),
-    size = c(217e6, 396e6, 383e6, 112e6)
+    size = c(217e6, 396e6, 383e6, 112e6),
+    stringsAsFactors = FALSE
   ),
 
   initialize = function(
@@ -171,7 +172,8 @@ whoi_plankton_dataset <- torch::dataset(
             "0f4d47f240cd9c30a7dd786171fa40ca",
             "db827a7de8790cdcae67b174c7b8ea5e",
             "d3181d9ffaed43d0c01f59455924edca"),
-    size = c(rep(450e6, 4), rep(490e6, 13), rep(450e6, 2))
+    size = c(rep(450e6, 4), rep(490e6, 13), rep(450e6, 2)),
+    stringsAsFactors = FALSE
   )
 )
 
@@ -201,6 +203,7 @@ whoi_small_coralnet_dataset <- torch::dataset(
             "f4dd2d2effc1f9c02918e3ee614b85d3",
             "d66ec691a4c5c63878a9cfff164a6aaf",
             "7ea146b9b2f7b6cee99092bd44182d06"),
-    size = c(430e6, rep(380e6, 4), 192e6)
+    size = c(430e6, rep(380e6, 4), 192e6),
+    stringsAsFactors = FALSE
   )
 )
diff --git a/R/dataset-vggface2.R b/R/dataset-vggface2.R
index 0ca3fe83..02cb745c 100644
--- a/R/dataset-vggface2.R
+++ b/R/dataset-vggface2.R
@@ -49,7 +49,8 @@ vggface2_dataset <- torch::dataset(
             "bb7a323824d1004e14e00c23974facd3",
             "d08b10f12bc9889509364ef56d73c621",
             "d315386c7e8e166c4f60e27d9cc61acc"),
-    size = c(37.9e9, 62e6, 2.03e9, 3e6, 335e3)
+    size = c(37.9e9, 62e6, 2.03e9, 3e6, 335e3),
+    stringsAsFactors = FALSE
   ),
   initialize = function(
     root = tempdir(),
diff --git a/R/item-transforms-geometry.R b/R/item-transforms-geometry.R
index 3d7f9346..f6568888 100644
--- a/R/item-transforms-geometry.R
+++ b/R/item-transforms-geometry.R
@@ -34,12 +34,11 @@
 #'
 #' after <- item_transform_rotate(before, angle = angle)
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 #'
-#' grid <- vision_make_grid(
-#'   torch_stack(list(transform_resize(before_plot, c(600, 600)),
-#'                    transform_resize(after_plot, c(600, 600)))),
+#' grid <- vision_make_grid(transform_resize(before_plot, c(600, 600)),
+#'                          transform_resize(after_plot, c(600, 600)),
 #'   scale = TRUE
 #' )
 #' tensor_image_browse(grid)
@@ -241,14 +240,14 @@ item_transform_rotate.image_with_rotated_box <- function(x, angle, interpolation
 #'
 #' after <- item_transform_crop(before, top = 100, left = 200, height = 500, width = 800)
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
-#'
-#' before_plot <- transform_resize(before_plot, c(2000, 3000))
-#' after_plot <- transform_resize(after_plot, c(2000, 3000))
+#' before_plot <- before %>%
+#'   draw_bounding_boxes(colors = "blue", width = 10) %>%
+#'   transform_resize(c(2000, 3000))
+#' after_plot <- after %>%
+#'   draw_bounding_boxes(colors = "red", width = 10) %>%
+#'   transform_resize(c(2000, 3000))
 #'
-#' grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-#' tensor_image_browse(grid)
+#' tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 #' }
 #'
 #' @family item_unitary_transforms
@@ -399,10 +398,10 @@ item_transform_crop.image_with_rotated_box <- function(x, top, left, height, wid
 #'
 #' after <- item_transform_hflip(before)
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 #'
-#' grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
+#' grid <- vision_make_grid(before_plot, after_plot, scale = TRUE)
 #' tensor_image_browse(grid)
 #' }
 #'
@@ -502,11 +501,10 @@ item_transform_hflip.image_with_rotated_box <- function(x) {
 #'
 #' after <- item_transform_vflip(before)
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 #'
-#' grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-#' tensor_image_browse(grid)
+#' tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 #' }
 #'
 #' @family item_unitary_transforms
@@ -613,14 +611,13 @@ item_transform_vflip.image_with_rotated_box <- function(x) {
 #'
 #' after <- item_transform_center_crop(before, size = c(1500, 1500))
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 #'
 #' before_plot <- transform_resize(before_plot, c(1500, 1500))
 #' after_plot <- transform_resize(after_plot, c(1500, 1500))
 #'
-#' grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-#' tensor_image_browse(grid)
+#' tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 #' }
 #'
 #' @family item_unitary_transforms
@@ -917,14 +914,13 @@ item_transform_affine.dataset <- function(x, angle = 0, translate = c(0, 0),
 #'
 #' after <- item_transform_pad(before, padding = 100)
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 #'
 #' before_plot <- transform_resize(before_plot, c(2000, 3000))
 #' after_plot <- transform_resize(after_plot, c(2000, 3000))
 #'
-#' grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-#' tensor_image_browse(grid)
+#' tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 #' }
 #'
 #' @family item_unitary_transforms
@@ -1087,11 +1083,10 @@ item_transform_pad.image_with_rotated_box <- function(x, padding, fill = 0, padd
 #'
 #' after <- item_transform_perspective(before, startpoints, endpoints)
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 #'
-#' grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-#' tensor_image_browse(grid)
+#' tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 #' }
 #'
 #' @family item_unitary_transforms
diff --git a/R/item-transforms-random-geometry.R b/R/item-transforms-random-geometry.R
index 911bb9d7..b6038d80 100644
--- a/R/item-transforms-random-geometry.R
+++ b/R/item-transforms-random-geometry.R
@@ -25,11 +25,10 @@
 #'
 #' after <- item_transform_random_horizontal_flip(before)
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 #'
-#' grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-#' tensor_image_browse(grid)
+#' tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 #' }
 #'
 #' @family item_random_transforms
@@ -111,11 +110,10 @@ item_transform_random_horizontal_flip.image_with_rotated_box <- function(x, p =
 #'
 #' after <- item_transform_random_vertical_flip(before)
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 #'
-#' grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), num_rows = 1, scale = TRUE)
-#' tensor_image_browse(grid)
+#' tensor_image_browse(vision_make_grid(before_plot, after_plot, per_row = 1, scale = TRUE))
 #' }
 #'
 #' @family item_random_transforms
@@ -365,16 +363,15 @@ rescale_box_angle <- function(angle_deg, scale_w, scale_h) {
 #'
 #' after <- item_transform_random_crop(before, size = c(800, 1200))
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+#' # the crop will changes the image size, so resize is needed before vision_make_grid
+#' before_plot <- before %>%
+#'   draw_bounding_boxes(colors = "blue", width = 10) %>%
+#'   transform_resize(c(600, 600))
+#' after_plot <- after %>%
+#'   draw_bounding_boxes(colors = "red", width = 10) %>%
+#'   transform_resize(c(600, 600))
 #'
-#' # the crop changes the image size, so resize before stacking into the grid
-#' grid <- vision_make_grid(
-#'   torch_stack(list(transform_resize(before_plot, c(600, 600)),
-#'                    transform_resize(after_plot, c(600, 600)))),
-#'   scale = TRUE
-#' )
-#' tensor_image_browse(grid)
+#' tensor_image_browse(vision_make_grid(before_plot, after_plot))
 #' }
 #'
 #' @family item_random_transforms
@@ -777,11 +774,10 @@ item_transform_random_rotation.image_with_segmentation_mask <- item_transform_ra
 #'
 #' after <- item_transform_random_erasing(before)
 #'
-#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+#' before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+#' after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 #'
-#' grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-#' tensor_image_browse(grid)
+#' tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 #' }
 #'
 #' @family item_random_transforms
diff --git a/R/models-faster_rcnn.R b/R/models-faster_rcnn.R
index 5663e708..48692def 100644
--- a/R/models-faster_rcnn.R
+++ b/R/models-faster_rcnn.R
@@ -980,7 +980,7 @@ mobilenet_v3_320_fpn_backbone <- function(pretrained = TRUE) {
 #' batch_normalized <- input$unsqueeze(1)    # Add batch dimension (1, 3, H, W)
 #'
 #' # ResNet-50 FPN V2
-#' model <- model_fasterrcnn_resnet50_fpn_v2(pretrained = TRUE, , detections_per_img = 5 )
+#' model <- model_fasterrcnn_resnet50_fpn_v2(pretrained = TRUE, detections_per_img = 5 )
 #' model$eval()
 #' torch::with_no_grad({pred <- model(batch_normalized)$detections[[1]]})
 #' labels <- coco_classes(as.integer(pred$labels))
diff --git a/R/models-mask_rcnn.R b/R/models-mask_rcnn.R
index 5c5c8197..660eefc4 100644
--- a/R/models-mask_rcnn.R
+++ b/R/models-mask_rcnn.R
@@ -562,7 +562,7 @@ maskrcnn_model_v2 <- torch::nn_module(
 #' batch <- input$unsqueeze(1)
 #'
 #' # Mask R-CNN ResNet-50 FPN
-#' model <- model_maskrcnn_resnet50_fpn(pretrained = TRUE, , detections_per_img = 5)
+#' model <- model_maskrcnn_resnet50_fpn(pretrained = TRUE, detections_per_img = 5)
 #' model$eval()
 #'
 #' torch::with_no_grad({pred <- model(batch)$detections[[1]]})
diff --git a/R/models-vgg.R b/R/models-vgg.R
index 31f0bde7..75c62137 100644
--- a/R/models-vgg.R
+++ b/R/models-vgg.R
@@ -10,7 +10,7 @@ VGG <- torch::nn_module(
       torch::nn_linear(4096, 4096),
       torch::nn_relu(TRUE),
       torch::nn_dropout(),
-      torch::nn_linear(4096, num_classes),
+      torch::nn_linear(4096, num_classes)
     )
 
     if (init_weights)
diff --git a/R/target-transforms-detection.R b/R/target-transforms-detection.R
index 17cbaaaa..cef3391b 100644
--- a/R/target-transforms-detection.R
+++ b/R/target-transforms-detection.R
@@ -277,11 +277,7 @@ target_transform_sahi_crop.object_detection_target <- function(y, sahi_split, mi
 #' # Rotated boxes (red, drawn as polygons)
 #' after_plot <- draw_bounding_boxes(img, boxes = rotated_target$boxes, colors = "red", width = 10)
 #'
-#' grid <- vision_make_grid(
-#'   torch_stack(list(before_plot, after_plot))$to(torch_float32()),
-#'   scale = TRUE
-#' )
-#' tensor_image_browse(grid)
+#' tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 #' }
 #'
 #' @family target_transforms_detection
diff --git a/R/transforms-generics.R b/R/transforms-generics.R
index c3c39af8..911305ac 100644
--- a/R/transforms-generics.R
+++ b/R/transforms-generics.R
@@ -358,7 +358,7 @@ transform_ten_crop <- function(img, size, vertical_flip = FALSE) {
 #'   class(item) <- "image_with_bounding_box"
 #'   draw_bounding_boxes(item, colors = "red")
 #' })
-#' grid <- vision_make_grid(torch_stack(preview), scale = FALSE, num_rows = 3)
+#' grid <- vision_make_grid(torch_stack(preview), scale = FALSE, per_row = 3)
 #' tensor_image_browse(grid)
 #' }
 #'
diff --git a/R/vision_utils.R b/R/vision_utils.R
index 2e118d14..0a926992 100644
--- a/R/vision_utils.R
+++ b/R/vision_utils.R
@@ -2,37 +2,96 @@
 #' @importFrom torch torch_uint8
 NULL
 
+.min_max_scale <- function(x) {
+  min <- x$min()$item()
+  max <- x$max()$item()
+  x$clamp_(min = min, max = max)
+  x$add_(-min)$div_(max - min + 1e-5)
+  x
+}
+
 #' A simplified version of torchvision.utils.make_grid
 #'
-#' Arranges a batch B of (image) tensors in a grid, with optional padding between
-#'   images. Expects a 4d mini-batch tensor of shape (B x C x H x W).
+#' Arranges images in a grid, with optional padding between images.
+#'
+#' For `torch_tensor` input, accepts either:
+#' - one or more 3D tensors of shape (C x H x W) passed as separate arguments, or
+#' - one or more 4D batch tensor of shape (B x C x H x W).
+#'
+#' For `magick-image` input, arranges frames using `magick::image_montage()`.
 #'
-#' @param tensor tensor of shape (B x C x H x W) to arrange in grid.
-#' @param scale whether to normalize (min-max-scale) the input tensor.
-#' @param num_rows number of rows making up the grid (default 8).
-#' @param padding amount of padding between batch images (default 2).
-#' @param pad_value pixel value to use for padding.
+#' @param tensor A 4D `torch_tensor` of shape (B x C x H x W), a 3D `torch_tensor`
+#'   of shape (C x H x W), or a `magick-image` object.
+#' @param ... Additional 3D `torch_tensor` objects (when first argument is a 3D
+#'   tensor), or additional `magick-image` objects (when first argument is a
+#'   `magick-image`).
+#' @param scale whether to normalize (min-max-scale) the input tensor. Only
+#'   applied for `torch_tensor` input.
+#' @param per_row maximum number of images per row (i.e., number of columns);
+#'   remaining images wrap to the next row. Default 8.
+#' @param num_rows Deprecated. Use `per_row` instead.
+#' @param padding amount of padding between images in pixels (default 2).
+#' @param pad_value pixel value (0–1) to use for padding background.
+#'
+#' @return a 3D `torch_tensor` of shape
+#'   \eqn{\approx(C , n\_rows \times H , per\_row \times W)} and of dtype `torch_float()`.
 #'
-#' @return  a 3d torch_tensor of shape \eqn{\approx(C , num\_rows \times H , num\_cols \times W)} of all images arranged in a grid.
 #' @family image display
 #' @export
-vision_make_grid <- function(tensor,
-                             scale = TRUE,
-                             num_rows = 8,
-                             padding = 2,
-                             pad_value = 0) {
+vision_make_grid <- function(tensor, ..., scale = TRUE, per_row = 8, padding = 2, pad_value = 0, num_rows=NULL) {
+  if (!is.null(num_rows)) {
+    deprecated("'num_rows' is deprecated, use 'per_row' instead.")
+    per_row <- num_rows
+  }
+  dots <- list(...)
+  if (length(dots) > 0) {
+    primary_class <- class(tensor)[1]
+    non_matching <- Filter(function(x) !inherits(x, primary_class), dots)
+    if (length(non_matching) > 0)
+      cli_abort(c(
+        "All arguments in {.arg ...} must be {.cls {primary_class}} objects.",
+        "x" = "Got {.cls {class(non_matching[[1]])[1]}}."
+      ))
+  }
+  UseMethod("vision_make_grid")
+}
+
+#' @rdname vision_make_grid
+#' @export
+vision_make_grid.default <- function(tensor, ..., scale = TRUE, per_row = 8, padding = 2, pad_value = 0, num_rows=NULL) {
+  cli_abort("The provided {.var tensor} class {.cls {class(tensor)}} is not supported by {.fn vision_make_grid}")
+}
+
+#' @rdname vision_make_grid
+#' @export
+vision_make_grid.torch_tensor <- function(tensor, ..., scale = TRUE, per_row = 8, padding = 2, pad_value = 0, num_rows=NULL) {
+  extra_tensors <- list(...)
+
+  if (!tensor$ndim %in% c(3L, 4L))
+    value_error("tensor must be 3D (C x H x W) or 4D (B x C x H x W)")
+
+  if (length(extra_tensors) > 0) {
+    all_ndims <- c(tensor$ndim, vapply(extra_tensors, function(x) x$ndim, integer(1)))
+    if (length(unique(all_ndims)) > 1)
+      value_error("All tensors must have the same number of dimensions (all 3D or all 4D)")
+  }
 
-  min_max_scale <- function(x) {
-    min <- x$min()$item()
-    max <- x$max()$item()
-    x$clamp_(min = min, max = max)
-    x$add_(-min)$div_(max - min + 1e-5)
-    x
+  to_float_unit <- function(t) {
+    if (t$dtype == torch::torch_uint8()) t$to(dtype = torch::torch_float32())$div(255) else t
   }
-  if(scale) tensor <- min_max_scale(tensor)
+  float_tensors <- lapply(c(list(tensor), extra_tensors), to_float_unit)
+  all_tensors <- if (scale) lapply(float_tensors, .min_max_scale) else float_tensors
+
+
+  if (tensor$ndim == 3) {
+    tensor <- torch::torch_stack(all_tensors)
+  } else {
+    tensor <- if (length(all_tensors) > 1) torch::torch_cat(all_tensors, dim = 1) else all_tensors[[1]]
+  }
+
 
   nmaps <- tensor$size(1)
-  xmaps <- min(num_rows, nmaps)
+  xmaps <- min(per_row, nmaps)
   ymaps <- ceiling(nmaps / xmaps)
   height <- floor(tensor$size(3) + padding)
   width <- floor(tensor$size(4) + padding)
@@ -62,6 +121,29 @@ vision_make_grid <- function(tensor,
   grid
 }
 
+#' @rdname vision_make_grid
+#' @export
+`vision_make_grid.magick-image` <- function(tensor, ..., scale = TRUE, per_row = 8, padding = 2, pad_value = 0, num_rows=NULL) {
+  rlang::check_installed("magick")
+
+  imgs <- tensor
+  extra_imgs <- list(...)
+  if (length(extra_imgs) > 0) {
+    imgs <- do.call(c, c(list(imgs), extra_imgs))
+  }
+
+  # transform_to_tensor already normalises to float32 [0,1]; apply per-frame
+  # scaling here so behaviour matches the 3D-tensor path (per-image, not global).
+  frame_tensors <- lapply(seq_along(imgs), function(i) {
+    t <- transform_to_tensor(imgs[i])
+    if (scale) .min_max_scale(t) else t
+  })
+
+  batch <- torch::torch_stack(frame_tensors)
+  vision_make_grid.torch_tensor(batch, scale = FALSE, per_row = per_row,
+                                padding = padding, pad_value = pad_value)
+}
+
 
 #' Draws bounding boxes on image.
 #'
diff --git a/man/item_transform_center_crop.Rd b/man/item_transform_center_crop.Rd
index 80e6bb08..3cacb80f 100644
--- a/man/item_transform_center_crop.Rd
+++ b/man/item_transform_center_crop.Rd
@@ -39,14 +39,13 @@ class(before) <- c("image_with_bounding_box", "list")
 
 after <- item_transform_center_crop(before, size = c(1500, 1500))
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 
 before_plot <- transform_resize(before_plot, c(1500, 1500))
 after_plot <- transform_resize(after_plot, c(1500, 1500))
 
-grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-tensor_image_browse(grid)
+tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 }
 
 }
diff --git a/man/item_transform_crop.Rd b/man/item_transform_crop.Rd
index 33fce70c..c0b0b410 100644
--- a/man/item_transform_crop.Rd
+++ b/man/item_transform_crop.Rd
@@ -45,14 +45,14 @@ class(before) <- c("image_with_bounding_box", "list")
 
 after <- item_transform_crop(before, top = 100, left = 200, height = 500, width = 800)
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+before_plot <- before \%>\%
+  draw_bounding_boxes(colors = "blue", width = 10) \%>\%
+  transform_resize(c(2000, 3000))
+after_plot <- after \%>\%
+  draw_bounding_boxes(colors = "red", width = 10) \%>\%
+  transform_resize(c(2000, 3000))
 
-before_plot <- transform_resize(before_plot, c(2000, 3000))
-after_plot <- transform_resize(after_plot, c(2000, 3000))
-
-grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-tensor_image_browse(grid)
+tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 }
 
 }
diff --git a/man/item_transform_hflip.Rd b/man/item_transform_hflip.Rd
index 65b2b11f..878775cd 100644
--- a/man/item_transform_hflip.Rd
+++ b/man/item_transform_hflip.Rd
@@ -32,10 +32,10 @@ class(before) <- c("image_with_bounding_box", "list")
 
 after <- item_transform_hflip(before)
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 
-grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
+grid <- vision_make_grid(before_plot, after_plot, scale = TRUE)
 tensor_image_browse(grid)
 }
 
diff --git a/man/item_transform_pad.Rd b/man/item_transform_pad.Rd
index 074927fe..00f8a7aa 100644
--- a/man/item_transform_pad.Rd
+++ b/man/item_transform_pad.Rd
@@ -59,14 +59,13 @@ class(before) <- c("image_with_bounding_box", "list")
 
 after <- item_transform_pad(before, padding = 100)
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 
 before_plot <- transform_resize(before_plot, c(2000, 3000))
 after_plot <- transform_resize(after_plot, c(2000, 3000))
 
-grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-tensor_image_browse(grid)
+tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 }
 
 }
diff --git a/man/item_transform_perspective.Rd b/man/item_transform_perspective.Rd
index d41dfc7d..30f3327d 100644
--- a/man/item_transform_perspective.Rd
+++ b/man/item_transform_perspective.Rd
@@ -62,11 +62,10 @@ endpoints <- list(c(100, 50), c(2700, 0), c(2800, 1800), c(80, 1900))
 
 after <- item_transform_perspective(before, startpoints, endpoints)
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 
-grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-tensor_image_browse(grid)
+tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 }
 
 }
diff --git a/man/item_transform_random_crop.Rd b/man/item_transform_random_crop.Rd
index 98581d04..0fb0e283 100644
--- a/man/item_transform_random_crop.Rd
+++ b/man/item_transform_random_crop.Rd
@@ -67,16 +67,15 @@ class(before) <- c("image_with_bounding_box", "list")
 
 after <- item_transform_random_crop(before, size = c(800, 1200))
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+# the crop will changes the image size, so resize is needed before vision_make_grid
+before_plot <- before \%>\%
+  draw_bounding_boxes(colors = "blue", width = 10) \%>\%
+  transform_resize(c(600, 600))
+after_plot <- after \%>\%
+  draw_bounding_boxes(colors = "red", width = 10) \%>\%
+  transform_resize(c(600, 600))
 
-# the crop changes the image size, so resize before stacking into the grid
-grid <- vision_make_grid(
-  torch_stack(list(transform_resize(before_plot, c(600, 600)),
-                   transform_resize(after_plot, c(600, 600)))),
-  scale = TRUE
-)
-tensor_image_browse(grid)
+tensor_image_browse(vision_make_grid(before_plot, after_plot))
 }
 
 }
diff --git a/man/item_transform_random_erasing.Rd b/man/item_transform_random_erasing.Rd
index 83d3ceef..df11658f 100644
--- a/man/item_transform_random_erasing.Rd
+++ b/man/item_transform_random_erasing.Rd
@@ -61,11 +61,10 @@ class(before) <- c("image_with_bounding_box", "list")
 
 after <- item_transform_random_erasing(before)
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 
-grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-tensor_image_browse(grid)
+tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 }
 
 }
diff --git a/man/item_transform_random_horizontal_flip.Rd b/man/item_transform_random_horizontal_flip.Rd
index 6661f878..8d0d4e74 100644
--- a/man/item_transform_random_horizontal_flip.Rd
+++ b/man/item_transform_random_horizontal_flip.Rd
@@ -35,11 +35,10 @@ class(before) <- c("image_with_bounding_box", "list")
 
 after <- item_transform_random_horizontal_flip(before)
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 
-grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-tensor_image_browse(grid)
+tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 }
 
 }
diff --git a/man/item_transform_random_vertical_flip.Rd b/man/item_transform_random_vertical_flip.Rd
index 1c9e8e02..4ad77961 100644
--- a/man/item_transform_random_vertical_flip.Rd
+++ b/man/item_transform_random_vertical_flip.Rd
@@ -35,11 +35,10 @@ class(before) <- c("image_with_bounding_box", "list")
 
 after <- item_transform_random_vertical_flip(before)
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 
-grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), num_rows = 1, scale = TRUE)
-tensor_image_browse(grid)
+tensor_image_browse(vision_make_grid(before_plot, after_plot, per_row = 1, scale = TRUE))
 }
 
 }
diff --git a/man/item_transform_rotate.Rd b/man/item_transform_rotate.Rd
index e6a49bc5..a114f692 100644
--- a/man/item_transform_rotate.Rd
+++ b/man/item_transform_rotate.Rd
@@ -59,12 +59,11 @@ class(before) <- c("image_with_bounding_box", "list")
 
 after <- item_transform_rotate(before, angle = angle)
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 
-grid <- vision_make_grid(
-  torch_stack(list(transform_resize(before_plot, c(600, 600)),
-                   transform_resize(after_plot, c(600, 600)))),
+grid <- vision_make_grid(transform_resize(before_plot, c(600, 600)),
+                         transform_resize(after_plot, c(600, 600)),
   scale = TRUE
 )
 tensor_image_browse(grid)
diff --git a/man/item_transform_vflip.Rd b/man/item_transform_vflip.Rd
index fafc26f7..0089d489 100644
--- a/man/item_transform_vflip.Rd
+++ b/man/item_transform_vflip.Rd
@@ -32,11 +32,10 @@ class(before) <- c("image_with_bounding_box", "list")
 
 after <- item_transform_vflip(before)
 
-before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)$to(torch_float())$div(255)
-after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)$to(torch_float())$div(255)
+before_plot <- draw_bounding_boxes(before, colors = "blue", width = 10)
+after_plot <- draw_bounding_boxes(after, colors = "red", width = 10)
 
-grid <- vision_make_grid(torch_stack(list(before_plot, after_plot)), scale = TRUE)
-tensor_image_browse(grid)
+tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 }
 
 }
diff --git a/man/model_fasterrcnn.Rd b/man/model_fasterrcnn.Rd
index 40574ea2..389a2513 100644
--- a/man/model_fasterrcnn.Rd
+++ b/man/model_fasterrcnn.Rd
@@ -119,7 +119,7 @@ input <- image  \%>\%
 batch_normalized <- input$unsqueeze(1)    # Add batch dimension (1, 3, H, W)
 
 # ResNet-50 FPN V2
-model <- model_fasterrcnn_resnet50_fpn_v2(pretrained = TRUE, , detections_per_img = 5 )
+model <- model_fasterrcnn_resnet50_fpn_v2(pretrained = TRUE, detections_per_img = 5 )
 model$eval()
 torch::with_no_grad({pred <- model(batch_normalized)$detections[[1]]})
 labels <- coco_classes(as.integer(pred$labels))
diff --git a/man/model_maskrcnn.Rd b/man/model_maskrcnn.Rd
index e314ad35..a9448a38 100644
--- a/man/model_maskrcnn.Rd
+++ b/man/model_maskrcnn.Rd
@@ -109,7 +109,7 @@ input <- img \%>\%
 batch <- input$unsqueeze(1)
 
 # Mask R-CNN ResNet-50 FPN
-model <- model_maskrcnn_resnet50_fpn(pretrained = TRUE, , detections_per_img = 5)
+model <- model_maskrcnn_resnet50_fpn(pretrained = TRUE, detections_per_img = 5)
 model$eval()
 
 torch::with_no_grad({pred <- model(batch)$detections[[1]]})
diff --git a/man/target_transform_rotate.Rd b/man/target_transform_rotate.Rd
index fd061bdb..7dbce8ae 100644
--- a/man/target_transform_rotate.Rd
+++ b/man/target_transform_rotate.Rd
@@ -63,11 +63,7 @@ rotated_target <- target_transform_rotate(target, angle = 4)
 # Rotated boxes (red, drawn as polygons)
 after_plot <- draw_bounding_boxes(img, boxes = rotated_target$boxes, colors = "red", width = 10)
 
-grid <- vision_make_grid(
-  torch_stack(list(before_plot, after_plot))$to(torch_float32()),
-  scale = TRUE
-)
-tensor_image_browse(grid)
+tensor_image_browse(vision_make_grid(before_plot, after_plot, scale = TRUE))
 }
 
 }
diff --git a/man/transform_sahi_crop.Rd b/man/transform_sahi_crop.Rd
index 538a34e2..b01dc173 100644
--- a/man/transform_sahi_crop.Rd
+++ b/man/transform_sahi_crop.Rd
@@ -100,7 +100,7 @@ preview <- lapply(1:dim(crops)[1], function(i) {
   class(item) <- "image_with_bounding_box"
   draw_bounding_boxes(item, colors = "red")
 })
-grid <- vision_make_grid(torch_stack(preview), scale = FALSE, num_rows = 3)
+grid <- vision_make_grid(torch_stack(preview), scale = FALSE, per_row = 3)
 tensor_image_browse(grid)
 }
 
diff --git a/man/vision_make_grid.Rd b/man/vision_make_grid.Rd
index e69d7ba7..8fa91cf8 100644
--- a/man/vision_make_grid.Rd
+++ b/man/vision_make_grid.Rd
@@ -2,33 +2,86 @@
 % Please edit documentation in R/vision_utils.R
 \name{vision_make_grid}
 \alias{vision_make_grid}
+\alias{vision_make_grid.default}
+\alias{vision_make_grid.torch_tensor}
+\alias{vision_make_grid.magick-image}
 \title{A simplified version of torchvision.utils.make_grid}
 \usage{
 vision_make_grid(
   tensor,
+  ...,
   scale = TRUE,
-  num_rows = 8,
+  per_row = 8,
   padding = 2,
-  pad_value = 0
+  pad_value = 0,
+  num_rows = NULL
+)
+
+\method{vision_make_grid}{default}(
+  tensor,
+  ...,
+  scale = TRUE,
+  per_row = 8,
+  padding = 2,
+  pad_value = 0,
+  num_rows = NULL
+)
+
+\method{vision_make_grid}{torch_tensor}(
+  tensor,
+  ...,
+  scale = TRUE,
+  per_row = 8,
+  padding = 2,
+  pad_value = 0,
+  num_rows = NULL
+)
+
+\method{vision_make_grid}{`magick-image`}(
+  tensor,
+  ...,
+  scale = TRUE,
+  per_row = 8,
+  padding = 2,
+  pad_value = 0,
+  num_rows = NULL
 )
 }
 \arguments{
-\item{tensor}{tensor of shape (B x C x H x W) to arrange in grid.}
+\item{tensor}{A 4D \code{torch_tensor} of shape (B x C x H x W), a 3D \code{torch_tensor}
+of shape (C x H x W), or a \code{magick-image} object.}
 
-\item{scale}{whether to normalize (min-max-scale) the input tensor.}
+\item{...}{Additional 3D \code{torch_tensor} objects (when first argument is a 3D
+tensor), or additional \code{magick-image} objects (when first argument is a
+\code{magick-image}).}
 
-\item{num_rows}{number of rows making up the grid (default 8).}
+\item{scale}{whether to normalize (min-max-scale) the input tensor. Only
+applied for \code{torch_tensor} input.}
 
-\item{padding}{amount of padding between batch images (default 2).}
+\item{per_row}{maximum number of images per row (i.e., number of columns);
+remaining images wrap to the next row. Default 8.}
 
-\item{pad_value}{pixel value to use for padding.}
+\item{padding}{amount of padding between images in pixels (default 2).}
+
+\item{pad_value}{pixel value (0–1) to use for padding background.}
+
+\item{num_rows}{Deprecated. Use \code{per_row} instead.}
 }
 \value{
-a 3d torch_tensor of shape \eqn{\approx(C , num\_rows \times H , num\_cols \times W)} of all images arranged in a grid.
+a 3D \code{torch_tensor} of shape
+\eqn{\approx(C , n\_rows \times H , per\_row \times W)} and of dtype \code{torch_float()}.
 }
 \description{
-Arranges a batch B of (image) tensors in a grid, with optional padding between
-images. Expects a 4d mini-batch tensor of shape (B x C x H x W).
+Arranges images in a grid, with optional padding between images.
+}
+\details{
+For \code{torch_tensor} input, accepts either:
+\itemize{
+\item one or more 3D tensors of shape (C x H x W) passed as separate arguments, or
+\item one or more 4D batch tensor of shape (B x C x H x W).
+}
+
+For \code{magick-image} input, arranges frames using \code{magick::image_montage()}.
 }
 \seealso{
 Other image display:
diff --git a/tests/testthat/test-vision-utils.R b/tests/testthat/test-vision-utils.R
index ea6d2654..29d41e4f 100644
--- a/tests/testthat/test-vision-utils.R
+++ b/tests/testthat/test-vision-utils.R
@@ -1,15 +1,75 @@
 context("vision-utils")
 
-test_that("vision_make_grid", {
-
+test_that("vision_make_grid works with 4D batch tensor", {
   images <- torch::torch_randn(c(4, 3, 16, 16))
+  grid <- vision_make_grid(images, per_row = 2, padding = 0)
+  expect_tensor_shape(grid, c(3, 32, 32))
+  expect_equal_to_r(grid$max() - grid$min(), 1, tolerance = 1e-4)
+})
+
+test_that("vision_make_grid works with multiple 3D tensors in ...", {
+  imgs <- lapply(1:4, function(i) torch::torch_randn(c(3, 16, 16)))
+  grid <- vision_make_grid(imgs[[1]], imgs[[2]], imgs[[3]], imgs[[4]], per_row = 2, padding = 0)
+  expect_tensor_shape(grid, c(3, 32, 32))
+  expect_equal_to_r(grid$max() - grid$min(), 1, tolerance = 1e-4)
+})
+
+test_that("vision_make_grid works with multiple 4D tensors in ...", {
+  batch1 <- torch::torch_randn(c(2, 3, 16, 16))
+  batch2 <- torch::torch_randn(c(2, 3, 16, 16))
+  grid <- vision_make_grid(batch1, batch2, per_row = 2, padding = 0)
+  expect_tensor_shape(grid, c(3, 32, 32))
+})
+
+test_that("vision_make_grid normalizes single uint8 4D batch to float [0,1]", {
+  images <- torch::torch_randint(0L, 256L, size = c(4, 3, 16, 16))$to(torch::torch_uint8())
+  grid <- vision_make_grid(images, per_row = 2, padding = 0, scale = FALSE)
+  expect_tensor_shape(grid, c(3, 32, 32))
+  expect_tensor_dtype(grid, torch::torch_float32())
+  expect_gte(grid$min()$item(), 0)
+  expect_lte(grid$max()$item(), 1)
+})
 
-  grid <- vision_make_grid(images, num_rows = 2, padding = 0)
+test_that("vision_make_grid normalizes mixed uint8/float inputs to float [0,1]", {
+  img_float <- torch::torch_rand(c(3, 16, 16))
+  img_uint8 <- torch::torch_randint(0L, 256L, size = c(3, 16, 16))$to(torch::torch_uint8())
+  grid <- vision_make_grid(img_float, img_uint8, per_row = 2, padding = 0, scale = FALSE)
+  expect_tensor_shape(grid, c(3, 16, 32))
+  expect_tensor_dtype(grid, torch::torch_float32())
+  expect_gte(grid$min()$item(), 0)
+  expect_lte(grid$max()$item(), 1)
+})
+
+test_that("vision_make_grid errors when ... contains non-matching types", {
+  img <- torch::torch_randn(c(3, 16, 16))
+  expect_error(vision_make_grid(img, "not_a_tensor"), "<torch_tensor>")
+})
+
+test_that("vision_make_grid errors on mixed 3D/4D tensors in ...", {
+  t3d <- torch::torch_randn(c(3, 16, 16))
+  t4d <- torch::torch_randn(c(2, 3, 16, 16))
+  expect_error(vision_make_grid(t3d, t4d), "same number of dimensions")
+})
 
+test_that("vision_make_grid works with magick-image", {
+  skip_if_not_installed("magick")
+  imgs <- magick::image_read(rep(system.file("img", "Rlogo.png", package = "png"), 4))
+  h <- magick::image_info(imgs[1])$height
+  w <- magick::image_info(imgs[1])$width
+  grid <- vision_make_grid(imgs, per_row = 2, padding = 0)
+  expect_tensor_shape(grid, c(3L, 2L * h, 2L * w))
+})
 
-  expect_equal(grid$size(), c(3, 32, 32))
-  expect_equal(as.numeric(grid$max() - grid$min()), 1, tolerance = 1e-4)
+test_that("vision_make_grid errors on unsupported type", {
+  expect_error(vision_make_grid(list(1, 2, 3)), "is not supported by")
+})
 
+test_that("vision_make_grid emits deprecation warning for num_rows", {
+  images <- torch::torch_randn(c(4, 3, 16, 16))
+  expect_warning(
+    vision_make_grid(images, num_rows = 2),
+    class = "deprecated"
+  )
 })
 
 test_that("draw_bounding_boxes works", {