demobook

ComfyUI: Tiled upscaling with Ultimate SD Upscale

ComfyUITry it →Watch full video →AI Search ·

Demo summary

A detailed demo of the Ultimate SD Upscale node, showing how tiled upscaling generates significantly higher facial detail than standard upscaling.

Step-by-step

  1. Delete the K Sampler and Latent Image nodes from your workflow.
  2. Add the Ultimate SD Upscale node and connect the Model, Positive Prompt, Negative Prompt, and VAE inputs.
  3. Add a Load Image node and connect it to the Image input of the Ultimate SD Upscale node.
  4. Add an Upscale Model Loader node, select a model like 4x-UltraSharp, and connect it to the Upscale_Model input.
  5. Set the Upscale by value (e.g., 2) and adjust the Denoise strength.
  6. Set the Tile Width and Tile Height to match the dimensions of your original input image.
  7. Add a Preview Image node to the output of the Ultimate SD Upscale node and click Queue Prompt.

Options

  • Chain multiple Ultimate SD Upscale nodes together to upscale in stages (e.g., 2x then 2x again).
  • Adjust Mask Blur and Tile Padding to improve how tiles blend together.
  • Use Ctrl+Shift+V to paste a node while maintaining existing input connections.

Watch out for

  • When chaining upscalers, you must update the Tile Width and Tile Height in the second node to match the dimensions of the image being passed to it (e.g., 1024x1024 if the first node already doubled a 512x512 image).

Tips

  • Set the Denoise strength to a relatively low value like 0.15 or 0.2 to retain the original image structure while adding detail.
  • If you see obvious lines between tiles in the final image, try adjusting the Mask Blur and Tile Padding settings.
  • Tiled upscaling is significantly better for generating fine details like hair and facial features compared to a standard model upscale.

Highlights

“you can literally just customize the workflow to whatever you want... and I'll show you the insane quality that this can generate”

All demos from “Master AI image generation - ComfyUI FULL TUTORIAL”

  1. 7:390:44Basic text-to-image generation in ComfyUIThe user demonstrates a basic text-to-image workflow using the RealVisXL model to generate an image of a castle in a forest.ComfyUIText to Image
  2. 8:418:40Building a ComfyUI workflow from scratchA step-by-step walkthrough of adding nodes (Checkpoint, CLIP Text Encode, KSampler, VAE Decode) and connecting them to build a custom image generation pipeline.ComfyUIAI Image Generator
  3. 19:571:09Generating a medieval warrior with batch processingDemonstrates generating multiple images of a medieval warrior by increasing the batch size in the Empty Latent Image node.ComfyUIAI Image Generator
  4. 25:330:32Importing workflows from image metadataShows how to drag an existing AI-generated image into the ComfyUI canvas to automatically reconstruct the entire node workflow used to create it.ComfyUIAI Image Generator
  5. 30:073:41Upscaling images using modelsDemonstrates upscaling a low-resolution 512x512 image using the 4X Ultra Sharp model to add clarity and detail.ComfyUIAI Image Upscaler
  6. 36:021:32Using Preview Chooser for selective upscalingShows how to use the Image Chooser node to generate a batch of images and selectively progress only one specific image to the upscaler.ComfyUIAI Image Upscaler
  7. 41:391:29Image-to-image generation with denoisingDemonstrates the image-to-image process by loading an existing photo and adjusting the denoising strength to create variations.ComfyUIImage to Image
  8. 43:496:52Tiled upscaling with Ultimate SD UpscaleCurrentA detailed demo of the Ultimate SD Upscale node, showing how tiled upscaling generates significantly higher facial detail than standard upscaling.ComfyUIAI Image Upscaler
  9. 58:094:05Controlling character poses with ControlNetDemonstrates using the OpenPose pre-processor and ControlNet Union model to force a generated princess character into a specific pose from a reference image.ComfyUIAI Image Generator
  10. 1:02:141:23Landscape composition control with Canny ControlNetShows how to use the Canny edge pre-processor to maintain the shape of mountains while changing the lighting to a sunset.ComfyUIAI Inpainting
  11. 1:09:444:55Face swapping with InstantIDDemonstrates the full setup and execution of InstantID to swap Will Smith's face onto a generated policeman character in both realistic and watercolor styles.ComfyUIAI Face Swap Generator
  12. Watch “Master AI image generation - ComfyUI FULL TUTORIAL” →

AI Image Upscaler

  1. 30:073:41Upscaling images using modelsDemonstrates upscaling a low-resolution 512x512 image using the 4X Ultra Sharp model to add clarity and detail.AI Search
  2. 36:021:32Using Preview Chooser for selective upscalingShows how to use the Image Chooser node to generate a batch of images and selectively progress only one specific image to the upscaler.AI Search
  3. 43:496:52Tiled upscaling with Ultimate SD UpscaleCurrentA detailed demo of the Ultimate SD Upscale node, showing how tiled upscaling generates significantly higher facial detail than standard upscaling.AI Search
  4. 2:432:38Generate and upscale images with Flux KontextThe creator demonstrates setting up a basic Flux Kontext workflow in ComfyUI, generating a character image, and then using a KSampler-based upscaler to increase resolution.pixaroma
  5. 9:521:53Upscale an existing image to 4K with PiDThe video demonstrates uploading a landscape photo into a ComfyUI workflow and using the PiD (Pixel Diffusion) model to upscale it to 4096 resolution in under 10 seconds.AI Search
  6. 13:200:38Compare original vs upscaled image using RG3 nodesThe user adds an 'image compare' node by RG3 to the ComfyUI canvas to show a side-by-side slider comparison of the blurry original image versus the sharp PiD upscaled result.AI Search
  7. 14:274:08Generate and upscale images using Zee Image and PiDA demonstration of a combined workflow where Zee Image Turbo generates a 1K image of a red panda which is then automatically passed into the PiD upscaler for a 4K final output.AI Search
  8. 1:491:28Simple non-AI image resizing in ComfyUIThe creator demonstrates a basic 'Upscale Image By' node that performs a standard resolution increase without AI, similar to Photoshop resizing, and uses an RG3 image comparer node to view the results.pixaroma
  9. 3:193:04Upscaling with Neural Network modelsThis clip shows how to use the 'Upscale Image (using Model)' node in ComfyUI with various PTH models like UltraSharp and 8x upscalers to sharpen and reconstruct details.pixaroma
  10. 6:233:22Tiled Diffusion upscaling with Zimage TurboThe creator walks through a complex workflow using Tiled Diffusion and the Zimage Turbo model to upscale images while managing VRAM by processing the image in smaller tiles.pixaroma
  11. 9:451:24Upscaling with Nunchaku Flux ManiaA demonstration of using a modified Flux.1 Dev model (Flux Mania) via Nunchaku nodes to perform a 2x upscale that enhances skin textures and fine details.pixaroma
  12. 11:103:10High-fidelity upscaling with SeedVR2The video shows the setup and execution of the SeedVR2 node pack, demonstrating how to select specific models and adjust resolution for sharp, consistent upscaling results.pixaroma
demobook

Show, don’t tell.

Categories

Video & AnimationImage Generation & EditingSales TechEmail Marketing

About

SubmitAboutPrivacyTermsContact

© demobook.co