1a. Low-Res Inputs

7.Outputs

4.ControlNets

2.Prompt

Animate Diff Nodes

6a.Upscaler

3.LoRAS

5. Sampler

OPTIONAL. IPAdapter (Reference Image)

4. ControlNet Previews

6b. Upscaler ControlNets

1b. Upscale Inputs

STRING

PRE TEXT

image

IMAGE

samples

LATENT

pixels

clip

CONDITIONING

timestep_keyframe

CONTROL_NET

images

CONTEXT_OPTIONS

model

MODEL

CLIP

positive

negative

cnet_stack

CONDITIONING+

CONDITIONING-

image1

image2

APP TEXT

IPADAPTER

control_net

CNET_STACK

seed

prev_motion_lora

MOTION_LORA

latent_image

context_options

motion_lora

motion_model_settings

CLIP_VISION

pre_text

app_text

LOW RES_Video Combine 🎥🅥🅗🅢

height

Height

width

Width

LATENT UPSCALE_Video Combine 🎥🅥🅗🅢

MASK

ipadapter

clip_vision

attn_mask

samples1

samples2

frame_count

AnimateDiff Workflow (ComfyUI) - Vid2Vid + ControlNet + Latent Upscale + Upscale ControlNet Pass + Multi Image IPAdapter

Tyler Bernabe

Base Generation - Save

Refined Generation - Save

Group

text_g

text_l

target_width

target_height

Negative Prompt

UPSCALE_MODEL

upscale_model

tile_width

tile_height

noise_seed

Positive Prompt

Definitions

text

Initial Input Selector

ComfyUI Workflow

Anonymous

Base Prompt

Refiner Prompt

Text Prompts

Load in BASE SDXL Model

Load in REFINER SDXL Model

Empty Latent Image

VAE Decoder

Step Control

Negative Prompt (Text)

Positive Prompt (Text)

Note - Load Checkpoint BASE

Note - Load Checkpoint REFINER

Note - Text Prompts

Note - KSampler  ADVANCED General Information

Note - VAE Decoder

Note - Empty Latent Image

Note - CLIP Encode (REFINER)

Note - CLIP Encode (BASE)

steps

start_at_step

KSampler (Advanced) - REFINER

Load Checkpoint - REFINER

Load Checkpoint - BASE

end_at_step

KSampler (Advanced) - BASE

1024 x 1024: square 1:1

1216 x 832: Photo (4:3)

1254 x 836: Landscape (3:2)

1365 x 768: Widescreen (16:9)

1536 x 640: Cinematic (21:9)

768 x 1344: Vertical (9:16)

832 x 1216: Portrait (4:5)

FLOAT

Steps

COMBO

sampler_name

Sampler

scheduler

Scheduler

Seed

Upscale Image

Upscaled image

VAE Decode     

Final Image

Samplers

Models, Encoders, and LoRAs

Prompting

Gen Settings

Fundamental Negative

Refiner CFG

Base CFG

Basic Prompting Guide

Latent Sizes

Linguistic Positive (G)

Supporting Terms (L)

ascore

Positive Refiner

Negative Refiner

Image Resolution

Negative A Score

Positive A Score

Negative Base

Positive Base

Base Pass

Refiner Pass

Base Model

Refiner Model

Sytan Workflow

filename_prefix

Base CLIP

Refiner CLIP

Models

Dimensions

VAE Decoders

Samplers Settings

Hires Fix

Img2Img

Save HiRes

Note - Dimensions

BASE LoRA

REFINER LoRA

latent1

latent2_opt

latent3_opt

latent4_opt

Save Refiner

Preview Img2Img

SDXL start step

KSampler (Advanced) - REFINER HF

Preview Base

SDXL hires fix start step

SDXL total steps

SDXL model change step

conditioning

Optional second pass to increase resolution.

Generate 4 different noisy images by only sampling the first 4 steps

Composite the noisy images together

Width Offset

Height Offset

text1

text2

separator

Sub Image Dimensions

Feather

SDXL\sd_xl_offset_example-lora_1.0.safetensors

Width Upscale Multiplier

Centered Formula

conditioning_to

conditioning_from

samples_to

samples_from

feather

Prompt

Neg Prompt

SDXL-REFINER

GLOBAL PARAMETERS

SDXL-BASE

Ultimate SD Upscaler

Comfy.ICU

Share and Run ComfyUI workflows

Starter Templates

Explore