2-2 Pass Txt2Img
Last updated
Last updated
Here are examples showing how you can achieve the "Hires Fix" functionality. You can load these images into ComfyUI to obtain the complete workflow.
The Hires Fix simply involves creating an image at a lower resolution, enlarging it, and then sending it through img2img. Note that in ComfyUI, txt2img and img2img are the same node. Txt2Img is performed by passing an empty image to the sampler node with maximum denoising.
Here is a simple workflow in ComfyUI to achieve this with basic latent upscaling.
Here is an example of how the ESRGAN upscaler can be used for the upscaling step. Since ESRGAN operates in pixel space, the image must be converted to pixel space and then back to latent space after being upscaled.
Here is an example of a more complex 2-pass workflow. The image is first generated with the WD1.5 beta 3 illusion model, upscaled latently, and then a second pass is performed with the cardosAnime_v10 model