Problem: I want to reuse SD 1.5 checkpoints and loras with SDXL. Solution: This shows how to load and use img2img on sdxl output with a 1.5 checkpoint and lora. Uses comfyui, based on https://github.com/SytanSD/Sytan-SDXL-ComfyUI . Before(sdxl+refiner): After (sdxl+ aniverse sd15 checkpoint + trip lora): This lets you create SDXL lora based on sd 1.5 checkpoint + lora as the output images are 1024x1024. Install Comfy UI and Comfy UI manager https://github.

This tutorial is also on civit at: https://civitai.com/models/58873/conceptmod-tutorial-fire-updated-term-train-any-lora-with-just-text-no-data-required Conceptmod (https://github.com/ntc-ai/conceptmod) is a technique that can change the internal latent space of trained stable diffusion models. It is based on https://erasing.baulab.info/ but has been heavily modified. Conceptmod can be used to remove, replace or emphasize concepts. This tutorial shows you how to use runpod (a gpu rental service) with conceptmod to train on your own text representations. Use the runpod at https://runpod.