<h1 id="heading-2624">Introduction</h1>This is an inpainting model of the excellent <a target="_blank" rel="ugc" href="https://civarchive.com/models/112902/dreamshaper-xl?modelVersionId=354657">Dreamshaper XL model</a> by @Lykon similar to <a target="_blank" rel="ugc" href="https://civarchive.com/models/403361/juggernaut-xl-inpainting">the Juggernaut XL inpainting model</a> I just published. People seem to really like both the Dreamshaper XL and lightning models in general because of their speed, so I figured at least some people might like an inpainting model as well.The whole purpose of inpainting models is to provide at least somewhat better results when inpainting or outpainting an image. When e.g. outpainting an image, with the normal checkpoint models you tend to get a very visible seam between the original part of the image and the newly extended part -- this model helps eliminate that seam! Also, using a specific version of an inpainting model instead of the generic SDXL-one tends to get more thematically consistent results.Here is an example of a rather visible seam after outpainting:<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/b4127d89-fd86-42ce-85b6-ae4b293ca932/width=525/b4127d89-fd86-42ce-85b6-ae4b293ca932.jpeg" />The original model on the left, the inpainting model on the right. (Yes, I cherrypicked one of the worst examples just to demonstrate the point)Disclaimer: you definitely can get good results even without, but it's easier with an inpainting model. No need for any offensive comments about how I'm bad and should feel bad.<h2 id="heading-2625">How to use:</h2>Basically, the same general guidelines apply as for the original model, ie. DPM++ SDE Karras as the sampler and 3-8 steps, denoising strength 0.8 and CFG scale 2 for generally good results. I am still experimenting with lightning models and inpainting/outpainting as to how to get the best results within the least amount of time.<h3 id="heading-2626">Other stuff:</h3>Created in Automatic1111 by merging <a target="_blank" rel="ugc" href="https://huggingface.co/wangqyqq/sd_xl_base_1.0_inpainting_0.1.safetensors/tree/main">the base SDXL inpainting model (by wangqyqq)</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/112902/dreamshaper-xl?modelVersionId=354657">Dreamshaper XL Lightning DPM++ SDE</a>, <a target="_blank" rel="ugc" href="https://huggingface.co/ByteDance/SDXL-Lightning/blob/main/sdxl_lightning_8step.safetensors">SDXL Lightning 8-step base</a> and the fp16 fix VAE. As a complete newcomer to this whole AI image generation scene, I am not at all certain this is the correct way of making lightning inpaint models, but it seems to work fine in my own experiments.

Dreamshaper XL Lightning inpainting

Initial release based on Dreamshaper XL Lightning released on Feb 21, 2024.

Files with SHA256: 1a49cd4473dad2121261bfa875e9fd18dc5fff707b995c0561f8110b7b2c644f