DD32 commited on
Commit
d721093
·
verified ·
1 Parent(s): 0492427

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +34 -1
README.md CHANGED
@@ -8,4 +8,37 @@ tags:
8
  - SDXL
9
  - illustrious
10
  - Pony
11
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8
  - SDXL
9
  - illustrious
10
  - Pony
11
+ ---
12
+
13
+ mg_7lambda_negative
14
+ Short description: a LoRA for suppressing unwanted visual traits (negative bias). Tuned and tested with SDXL (ComfyUI).
15
+
16
+ Note: perceived strength depends on the base model and prompt. Start low and increase gradually.
17
+
18
+ Files
19
+
20
+ How to Use (ComfyUI)
21
+
22
+ Download the file and place it in ComfyUI/models/loras.
23
+ Classic graph: add a Load LoRA node, select the file, set strength_model and strength_clip.
24
+ MagicNodes CombiNode: enable a slot (use_lora_X = True) and pick this LoRA in the dropdown.
25
+ Recommended Settings (starting points)
26
+
27
+ strength_model: -0.40–1.00
28
+
29
+ strength_clip: 0.20–0.40
30
+
31
+ If the effect is too strong, lower both by 0.05–0.10 per step.
32
+
33
+ For more controlled results: try slightly lower CLIP (e.g., 0.35) and slightly higher model (e.g., 0.55).
34
+
35
+ Prompt Tips
36
+
37
+ Put the negative prompt: (super-wrong:1)
38
+
39
+
40
+ Known Limitations
41
+
42
+ Primarily tuned for SDXL; other bases may need different strengths.
43
+
44
+ Very high values can “wash out” details.