Nodes/CRT-Nodes/Image Upscale Sampler (CRT)
ComfyUI Node

Image Upscale Sampler (CRT)

A ComfyUI node in CRT/Sampling with 35 inputs and 2 outputs.

By PGCRT·Created 2 years ago·Updated 7 days ago· 129
Image Upscale Sampler (CRT)
  • model
  • positive
  • negative
  • image
  • latent
  • vae
  • image
  • latent
sampler_nameeuler
schedulersimple
cfg1.0
steps4
denoise0.33
seed0
seed_shift0
edit_model_flux2kleinfalse
enable_upscale_modelfalse
upscale_model_name
tile_size_megapixels1.0
tile_grid4x4
tile_padding5.0
mask_blur10.0
stage1_sigma_factor1.005
stage2_sigma_factor0.995
stage1_sigma_start0.300
stage1_sigma_end0.500
stage2_sigma_start0.000
stage2_sigma_end0.700
details_amount_stage10.50
details_amount_stage20.15
enable_noise_injectiondisable
injection_point0.75
injection_seed_offset1
injection_strength0.25
normalize_injected_noiseenable
vae_decode_tiledtrue
color_match_strength1.00
CategoryCRT/Sampling

Inputs (35)

NameTypeDefaultDescription
modelMODEL
positiveCONDITIONING
sampler_nameCOMBOeuler44 options: euler, euler_cfg_pp, euler_ancestral, euler_ancestral_cfg_pp, heun, heunpp2, +38
schedulerCOMBOsimple9 options: simple, sgm_uniform, karras, exponential, ddim_uniform, beta, +3
cfgFLOAT1.00–100Classifier Free Guidance scale. Higher values follow the prompt more closely.
stepsINT41–10000
denoiseFLOAT0.330–1Amount of denoising to apply. 1.0 = full denoising (txt2img), 0.5-0.8 typical for img2img.
seedINT00–18446744073709550000
seed_shiftINT0-100000–100000Offset added to the main seed for variation
edit_model_flux2kleinBOOLEANfalseEnable per-tile ReferenceLatent-style conditioning for edit models
enable_upscale_modelBOOLEANfalseEnable AI model upscaling before sampling
upscale_model_nameCOMBOSelect upscale model from models/upscale_models folder
tile_size_megapixelsFLOAT1.00.1–16Target megapixels per tile used to derive upscale factor
tile_gridCOMBO4x4Grid size for tiling (e.g., 4x4 = 16 tiles)
tile_paddingFLOAT5.00–25Padding around each tile as percentage of tile size
mask_blurFLOAT10.00–50Blur radius for tile blending as percentage of tile size
stage1_sigma_factorFLOAT1.0050–100
stage2_sigma_factorFLOAT0.9950–100
stage1_sigma_startFLOAT0.3000–1
stage1_sigma_endFLOAT0.5000–1
stage2_sigma_startFLOAT0.0000–1
stage2_sigma_endFLOAT0.7000–1
details_amount_stage1FLOAT0.50-5–5
details_amount_stage2FLOAT0.15-5–5
enable_noise_injectionCOMBOdisableEnable noise injection during sampling
injection_pointFLOAT0.750–1Percentage of steps after which to inject noise
injection_seed_offsetINT1-100–100Offset added to main seed for injection noise
injection_strengthFLOAT0.25-20–20Strength of injected noise
normalize_injected_noiseCOMBOenableNormalize injected noise to match latent statistics
vae_decode_tiledBOOLEANtrueUse tiled VAE decode. Tile size and overlap are derived from tile size megapixels and tile padding.
color_match_strengthFLOAT1.000–2Strength of color matching between original and enhanced result. 0.0 = disabled, 1.0 = full matching
negativeoptCONDITIONING
imageoptIMAGE
latentoptLATENT
vaeoptVAE

Outputs (2)

NameTypeDescription
imageIMAGE
latentLATENT