(((Peeをする)))

I created this first workflow as a semi-beginner, I was inspired by things I saw online, particularly in tutorials on Civitai. I added descriptions of how it works in French and English. The goal is to do image to image in SDXL then upscale before processing in the samplet in order to improve the final result.

On a rainy day in a city back alley, a woman encounters a lost kitten. She is contemplating what to do with the kitten...ㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡㅡIf a high-quality image has been generated, you can set kSampler control_after_generate to "fixed" to produce the same results.

>Select your checkpoint (If you want you can also use lora model for different styles),>Load your image, >Adjust your controlnet strength if needed and boom click generate> チェックポイントを選択してください (必要に応じて、別のスタイルに lora モデルを使用することもできます)> 画像をロードします> 必要に応じてコントロールネットの強度を調整し、「生成」をクリックします。

Multiple ControlNets

Flux Image to Image Upscale