Generates an image in SDXL and upscale it to 2x the original resolution using Tile Controlnet, Tiled Diffusion and colour matching. You can provide your own image (e.g. from Flux) for upscaling and disable the image generation process. NodesInstall the following custom nodes using ComfyUI Manager:<ul><li>rgthree's ComfyUI Nodes</li><li>KJNodes for ComfyUI</li><li>ComfyUI Essentials</li><li>ComfyUI-Advanced-ControlNet</li><li>Tiled Diffusion &amp; VAE for ComfyUI</li></ul>Models RequiredCheckpoint: <ul><li>Dreamshaper Turbo <a target="_blank" rel="ugc" href="https://civarchive.com/models/112902?modelVersionId=351306">https://civarchive.com/models/112902?modelVersionId=351306</a></li><li>You can use your favourite checkpoint but be sure the change the settings in the KSampler</li></ul>Tile Controlnet:<ul><li>xinsir/ControlNet++: All-in-one ControlNet (ProMax model) - higher detail</li><li>xinsir/ControlNet Tile SDXL (better colours) </li></ul> This workflow implements the ideas from this video by Latent Vision:<div data-youtube-video><iframe width="640" height="480" allowfullscreen="true" autoplay="false" disablekbcontrols="false" enableiframeapi="false" endtime="0" ivloadpolicy="0" loop="false" modestbranding="false" origin playlist src="https://www.youtube.com/embed/PZstjnT1cpI" start="0"></iframe></div>

Tiled Diffusion Tile Controlnet Upscale

Re-introduced the Tiled KSampler as the Random Strict Tiling Strategy seems to have solved the artifacts issue. Introduced a line art / depth controlnet to allow the output image to better follow the structure of the image to be upscaled.

Files with SHA256: d57f8802bcf690c1dd69153bd8696bc8b57eacb4075f391baf7d3cf86bb28df7