Image Studio
ComfyUI URL (your pod)
Connect
Remember on this device
Paste the pod's ComfyUI link. It changes each time you deploy a new pod.
Forget
Model
realvisxlV50_bakedvae.safetensors
LoRA (optional)
(none)
LoRA strength
Prompt
Negative prompt
bad anatomy, deformed, disfigured, mutated, extra limbs, extra arms, extra legs, missing limbs, missing arms, missing legs, fused limbs, extra fingers, missing fingers, mutated hands, malformed hands, cloned face, two heads, multiple faces, duplicate, long neck, low quality, worst quality, blurry, watermark, text
Reference images (optional)
Add image(s)
Clear all
Use the image(s) for…
Blend / amalgam of all images
Same face (InstantID)
Copy pose (ControlNet)
Transform one image (img2img)
Reference strength:
0.6
Blends features of every image into a new generation.
Face strength:
0.8
Reproduces the face from the first image in a new scene. Prompt describes the scene/pose.
Pose strength:
0.9
Copies the body pose from the first image; the prompt defines who/what.
Change amount (denoise):
0.5
Uses the first image. Lower = keeps it faithful.
Describe details to keep in the prompt.
Auto-fix faces & hands
(FaceDetailer — recommended)
Upscale 2× (extra detail, slower)
Generate
Advanced
Steps
CFG
Aspect (keep ~1MP to avoid duplicate faces/limbs)
Portrait 832×1216 (best for people)
Portrait 896×1152
Square 1024×1024
Landscape 1152×896
Landscape 1216×832
Sampler
dpmpp_2m
Scheduler
karras
Seed (blank=random)
# images
1
2
3
4
Prompt history
Clear
Saved images
Clear all