StableDiffusion 3.5 Large — IP-Adapter + UltraReal LoRA
Prompt
Prompt
Text Scale
↺
0
4
Run
Advanced Settings
▼
Image Prompt (Required)
Drop File Here
- or -
Click to Upload
Img 1 Scale
↺
0
16
Image Prompt 2 (Optional)
Drop File Here
- or -
Click to Upload
Img 2 Scale
↺
0
16
Image Prompt 3 (Optional)
Drop File Here
- or -
Click to Upload
Img 3 Scale
↺
0
16
Image Prompt 4 (Optional)
Drop File Here
- or -
Click to Upload
Img 4 Scale
↺
0
16
Image Prompt 5 (Optional)
Drop File Here
- or -
Click to Upload
Img 5 Scale
↺
0
16
SigLIP encoder (InstantX was trained on google/siglip-so400m-patch14-384)
Multi-image combine (concat = keep each image's tokens; mean = blend moods)
Overall Image Scale
↺
0
2
UltraReal LoRA scale (0 = off; inert until the kohya weights are converted for SD3)
↺
0
1.5
Negative prompt 1
Negative prompt 2
Negative prompt 3
Width
↺
256
4096
Height
↺
256
4096
Guidance scale
↺
0
30
Number of inference steps
↺
1
500
Examples