One of my favourite work yet. The creator of this aesthetic is <a target="_blank" rel="ugc" href="https://x.com/The_Sycomore/">https://x.com/The_Sycomore/</a> i trained this LoRA on his images and style. Works well with Flux Krea Dev too<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/c1e5b425-2a93-4d30-92f3-120be10982be/width=525/c1e5b425-2a93-4d30-92f3-120be10982be.jpeg" /> I trained the LoRA with rank 64 to capture the most amount of details completely disregarding the size factor. I know the size can be off putting to a lot of users, so i'll be uploading a resized LoRA version soon. Im experimenting with a few different versions and will upload the best one.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/9cb31199-cba5-4b9c-a30a-ec88bb5a2d0c/width=525/9cb31199-cba5-4b9c-a30a-ec88bb5a2d0c.jpeg" />All the preview images will have same prompt+seed Value+resolution to show off the difference in LoRA rank. These are not reuploads of the same picture. If the images look very similar, that means the LoRA resize went exceptionally well.

SynChrome

Max Rank: 32; dp: 97; Average Frobenius norm retention: 97.99% | std: 0.006

Files with SHA256: 0b017db0f6f6363930414aa1f6d3640ba1352df4362d751d706e92e541016da9