X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?p=flatland.git;a=blobdiff_plain;f=README.md;h=c1ff0a278434c5f5a2e28b0c42b77884e51c0d79;hp=47633c05504d53425408b3cfc5caa13e1487b337;hb=db40d75f91d5d332b475a260df8c034103159286;hpb=b8076332f83b24a86e727b4e4399b5965a983e7a

diff --git a/README.md b/README.md
index 47633c0..c1ff0a2 100644
--- a/README.md
+++ b/README.md
@@ -3,6 +3,37 @@
 This is an implementation of a simple 2d physics simulator usable from
 the pytorch framework.
 
+The main function is
+
+```
+torch.ByteTensor generate_sequence(bool pulling,
+                                   int nb_sequences,
+                                   int nb_images,
+                                   int image_height, int image_width,
+                                   int nb_shapes,
+                                   bool random_shape_size, bool random_colors)
+```
+with
+
+ * `pulling` indicating if one of the shape should be pulled upward,
+ * `nb_sequences` the number of sequences to generate
+ * `nb_images` the number of frames per sequence
+ * `image_height` an `image_width` the individual image size
+ * `nb_shapes` how many rectangular shapes to put in the simulation
+ * `random_shape_size` should they be of different size
+ * `random_colors` should they be of different colors
+
+The returned ByteTensor has five dimensions:
+
+ * Sequence index
+ * Image index
+ * Channel (3, as for RGB)
+ * Pixel row
+ * Pixel col
+
+Warning: For improper parameters values (e.g. too many shapes, too
+small images), the simulation may be stuck.
+
 # Installation and test #
 
 ```