How To Use

1. Upload an Image – Start by selecting or dragging an image into the workflow.
2. Choose an Upscaling Model – The workflow utilizes an advanced image-to-image upscaling model. You can select a model from the available options.
3. Select the Scaling Factor – Choose how much you want to upscale your image (e.g., 2x, 4x).
4. Set Creativity (Optional) – Adjust how much freedom you want to give the model in the upscaling process. Low creativity stays close to the original image, while high creativity allows for more change. When using a prompt, high creativity increases the influence of the prompt.
5. Set Structure (Optional) - Adjust how closely you want the model to follow the original structure of the image when upscaling. Low structure allows major changes, while high structure strictly maintains the original composition and layout.
6. Select a Generation Model - Select the model that will be used to generate each tile of the image in the upscaling process. This must be an SD 1.5 or SDXL model.
7. Run the Workflow – Click the Invoke button to process the image.‍
8. Save the Final Image – Once processing is complete, the upscaled image can be saved and used in your projects.

How It Works

1. Image Input – The original image is fed into the workflow.
2. Scaling Processing – The workflow applies an image-to-image model designed for high-quality upscaling.
3. Tile-Based Diffusion – For large images, the process is split into tiles to prevent quality loss and allow efficient memory usage.
4. Denoising & Refinement – A multi-diffusion denoising process is applied to ensure smooth gradients and remove unwanted noise.
5. ControlNet Enhancement – ControlNet is used to guide structural details and maintain the integrity of the image during upscaling.
6. Final Refinement with Unsharp Masking – The workflow sharpens details to improve the crispness of the final image.‍
7. Output & Storage – The processed high-resolution image is generated and saved for download or further editing.