Recommended strength: 0,8Trigger word: liljcolAuxiliary words: 1girl, solo, looking at viewer, realistic, brown hair, short/medium/long hairLoRA of British-American actress <a target="_blank" rel="ugc" href="https://en.wikipedia.org/wiki/Lily_Collins">Lily Collins</a> of Emily in Paris fame. Dataset consists of 30 face images and 20 body images, all from the period 2015-2018. Decided to make one as she's pretty and the ones out right now don't quite capture her likeness.Likeness is there in this model. Poses and focal length seem rather baked into the model (Lily will not be foreshortened when using a tag like GoPro). Preview images were generated at rather low res to avoid artefacts like double-heads (an issue of SD 1,5, not the LoRA). You may experiment with upscaling or hires fixes in your usage. Considering training for SDXL next time around.Finally, many thanks to <a target="_blank" rel="ugc" href="https://civarchive.com/articles/5010/celebrity-lora-training-guide-consolidated-sd-15">this article</a> from <a target="_blank" rel="ugc" href="https://civarchive.com/user/__2_">__2_</a> which I used. Please share your creations and feedback in the form of a comment! That way, I can improve on future LoRAs.

Files with SHA256: 6e7328d466bd26090c8dc6d843ed3c32dd08d0fdc8e3f41d23ac8470a2e8f890