Negative / Unconditioned Prompts (#661)

Co-Authored-By: rabidcopy <8052832+rabidcopy@users.noreply.github.com>

Co-authored-by: rabidcopy <8052832+rabidcopy@users.noreply.github.com>

docs/other/CONTRIBUTORS.md

@@ -2,15 +2,16 @@
 title: Contributors
 ---
 
-The list of all the amazing people who have contributed to the various features that you get to experience in this fork.
+The list of all the amazing people who have contributed to the various features that you get to
+experience in this fork.
 
 We thank them for all of their time and hard work.
 
-## __Original Author:__
+## **Original Author:**
 
 - [Lincoln D. Stein](mailto:lincoln.stein@gmail.com)
 
-## __Contributions by:__
+## **Contributions by:**
 
 - [Sean McLellan](https://github.com/Oceanswave)
 - [Kevin Gibbons](https://github.com/bakkot)
@@ -52,8 +53,9 @@ We thank them for all of their time and hard work.
 - [Doggettx](https://github.com/doggettx)
 - [Matthias Wild](https://github.com/mauwii)
 - [Kyle Schouviller](https://github.com/kyle0654)
+- [rabidcopy](https://github.com/rabidcopy)
 
-## __Original CompVis Authors:__
+## **Original CompVis Authors:**
 
 - [Robin Rombach](https://github.com/rromb)
 - [Patrick von Platen](https://github.com/patrickvonplaten)
@@ -65,4 +67,5 @@ We thank them for all of their time and hard work.
 
 ---
 
-_If you have contributed and don't see your name on the list of contributors, please let one of the collaborators know about the omission, or feel free to make a pull request._
+_If you have contributed and don't see your name on the list of contributors, please let one of the
+collaborators know about the omission, or feel free to make a pull request._

ldm/dream/conditioning.py

@@ -13,7 +13,20 @@ import re
 import torch
 
 def get_uc_and_c(prompt, model, log_tokens=False, skip_normalize=False):
-    uc = model.get_learned_conditioning([''])
+    # Extract Unconditioned Words From Prompt
+    unconditioned_words = ''
+    unconditional_regex = r'\[(.*?)\]'
+    unconditionals = re.findall(unconditional_regex, prompt)
+
+    if len(unconditionals) > 0:
+        unconditioned_words = ' '.join(unconditionals)
+
+        # Remove Unconditioned Words From Prompt
+        unconditional_regex_compile = re.compile(unconditional_regex)
+        clean_prompt = unconditional_regex_compile.sub(' ', prompt)
+        prompt = re.sub(' +', ' ', clean_prompt)
+
+    uc = model.get_learned_conditioning([unconditioned_words])
 
     # get weighted sub-prompts
     weighted_subprompts = split_weighted_subprompts(
@@ -34,6 +47,7 @@ def get_uc_and_c(prompt, model, log_tokens=False, skip_normalize=False):
     else:   # just standard 1 prompt
         log_tokenization(prompt, model, log_tokens)
         c = model.get_learned_conditioning([prompt])
+        uc = model.get_learned_conditioning([unconditioned_words])
     return (uc, c)
 
 def split_weighted_subprompts(text, skip_normalize=False)->list: