Fix for better continuity

funpack.py

@@ -698,7 +698,19 @@ class FunPackStoryWriter:
             for seq_idx in range(prompt_count):
                 # Add **only** the fresh sequence instruction each time
                 messages.append({"role": "system", "content": f"""This sequence's ID is {seq_idx + 1} out of total {prompt_count} sequences requested."""})
-                messages.append({"role": "user", "content": sequence_system_prompt})
+                messages.append({"role": "user", "content": seq_prompt = f"""
+                    {sequence_system_prompt}
+
+                    Current sequence number: {seq_idx + 1}
+                    Total sequences requested: {prompt_count}
+
+                    Original user prompt: {user_prompt}
+
+                    Previous sequences (for continuity):
+                    {chr(10).join(outputs[:seq_idx]) if seq_idx > 0 else "This is the first sequence."}
+
+                    Generate the next sequence now.
+                    """})
 
                 llm_tokens = llm_tokenizer.apply_chat_template(
                     messages, add_generation_prompt=True, return_tensors="pt", tokenize=True
@@ -1447,5 +1459,6 @@ NODE_DISPLAY_NAME_MAPPINGS = {
 
 
 
+