WorkflowListedExternal

HR Beautify Krita AI custom workflow (SDXL): refine after upscale and go as high as 24K

TaskMultiCapabilitiesImage to image · UpscalingFamilyStable DiffusionEstimated VRAMEstimate unavailableBgrade

Runtime profile

Access: External workflow
Status: Listed
Source: Civitai
Updated: Jul 19, 2026

Source description

HR Beautify (‘HR’ stands for Hires, or High resolution) is an advanced Krita AI custom workflow to refine upscaled images. It doesn’t include the usual upscale with AI model component. Instead, it concentrates on best refinement of raw-upscaled hires images possible, optionally beautifying them with a style transfer. Thereby the workflow extends the plugin's functionality which it lacks in the hires department.

The workflow includes two ControlNets, for the most powerful and flexible img2img guidance, and two IP Adapters, to enhance images with style transfer and composition or reference guidance. The models for these components, as well as for Clip Vision, can all be customized. Also included in the workflow are two post processing components: enhancing image’s HDR range and advanced image sharpening.

The tiling component that enables hires img2img refining is the widely used Tiled Diffusion (TD), together with Tiled VAE Encode & Decode. TD is compatible with all ControlNet models and can assist in processing images of up to 24K resolution, optionally adding creative detail in the output along the way.

AI-generated commentary

AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.

Purpose

Use HR Beautify in Krita to refine an already upscaled hires image into a refined hires image with tiled high-resolution img2img processing, optional style transfer, composition or reference guidance, creative detail, HDR enhancement, and advanced sharpening.

Flow

Tiled Diffusion works with Tiled VAE Encode and Decode for hires img2img refinement, while two ControlNets provide img2img guidance and two IP Adapters support style, composition, or reference guidance.
The workflow includes post-processing for HDR range enhancement and advanced image sharpening.

Inputs and outputs

Input: a raw-upscaled hires image; the usual AI upscaling component is not included.
Output: a refined hires image that may include style transfer, creative detail, HDR range enhancement, and advanced sharpening.

Setup

This Krita AI workflow uses SDXL 1.0 and is published by Minsky91.

Dependencies

Model files named for the workflow are CLIP-ViT-H-14-laion2B-s32B-b79K.safetensors, diffusion_pytorch_model_promax.safetensors, flux1-redux-dev.safetensors, and xinsirtile-sdxl-1.0.safetensors.
Named node components include ACN_AdvancedControlNetApply_v2, CLIPTextEncode, CLIPVisionLoader, ControlNetApplyAdvanced, ControlNetLoader, ETN_KritaCanvas, ETN_KritaImageLayer, ETN_KritaOutput, ETN_KritaStyle, ETN_Parameter, Fast Groups Bypasser (rgthree), HintImageEnchance, and
Additional named components include IPAdapterEmbeds, IPAdapterEncoder, IPAdapterModelLoader, Seed (rgthree), SetUnionControlNetType, TiledDiffusion, workflow>Clip text (concatenated), workflow>ControlNet 1 mask handler, workflow>HDR intensity group, workflow>IPAdapter 2 mask

Configuration

You can customize the models used for the two ControlNets, two IP Adapters, and Clip Vision.
Tiled Diffusion is compatible with all ControlNet models and can assist with processing images up to 24K resolution, optionally adding creative detail.

Cautions

Before use, verify that the named model files and node components are available in your Krita AI setup.
Suggestion · not verified
No license is specified.

Does this need to be edited?

Sources

1 source

CivitaiPrimaryVersion 1819814

civitai.com/models/1608099

Verified Jul 19

Source excerpts

1 excerpt

NoteJul 19civitai.com/models/1608099

Source context: 342 downloads · Type Workflows · Base model SDXL 1.0