diff --git a/img/cover.png b/img/cover.png
deleted file mode 100644
index 415250c..0000000
Binary files a/img/cover.png and /dev/null differ
diff --git a/img/cover3.png b/img/cover3.png
deleted file mode 100644
index ec303e5..0000000
Binary files a/img/cover3.png and /dev/null differ
diff --git a/img/cover4.png b/img/cover4.png
new file mode 100644
index 0000000..6159bf3
Binary files /dev/null and b/img/cover4.png differ
diff --git a/readme.md b/readme.md
index 5b2078d..b51f95d 100644
--- a/readme.md
+++ b/readme.md
@@ -1,10 +1,10 @@
 # Jsonformer: A Bulletproof Way to Generate Structured JSON from Language Models.
 
-## Problem: Getting models to output structed JSON is hard
+### Problem: Getting models to output structed JSON is hard
 
-## Solution: Only generate the content tokens and fill in the fixed tokens
+### Solution: Only generate the content tokens and fill in the fixed tokens
 
-![cover](img/cover3.png)
+![cover](img/cover4.png)
 
 Generating structured JSON from language models is a challenging task. The
 generated JSON must be syntactically correct, and it must conform to a schema