This workflow is pretty similar to my FLUX of ALL Trades workflow but a lot more refined. It features strong img2img capabilities with much of the workflow being focused on creating highly detailed photos of people. I have a few upscalers set up that target skin detail, along with some LUTS for color grading, and some other post processing magic. The most powerful upgrade with this workflow is probably the Detail Daemon Sampler, it allows you to have far more control of the denoising process, so it can really boost detail as well as allow finer control over the change when working with img2img. I realized that in my previous workflow my way of model merging was just really dumb lmao, so I switched to the model merge simple node and have been merging FLUXMANIA with getphatFLUXReality, with FLUXMANIA being the majority model. The workflow currently allows up to 6 loras but you can add more nodes if needed. I don't have face-detailer set up in this workflow but if anyone requests it I can add it, the workflow is already kind of slow as is so it felt like an unnecessary step. Instead of "CR upscale image" or "Ultimate Upscaler" I have opted to use a basic image upscale node, then running the second refiner process at the higher resolution. I have notes written down in the workflow to try and make things a little less confusing but if you're having problems feel free to leave a comment. The skin upscalers can be found here: <a target="_blank" rel="ugc" href="https://openmodeldb.info/models/1x-SkinContrast-SuperUltraCompact">https://openmodeldb.info/models/1x-SkinContrast-SuperUltraCompact</a> <a target="_blank" rel="ugc" href="https://openmodeldb.info/models/1x-SkinContrast-High-SuperUltraCompact">https://openmodeldb.info/models/1x-SkinContrast-High-SuperUltraCompact</a> <a target="_blank" rel="ugc" href="https://openmodeldb.info/models/1x-ITF-SkinDiffDDS-v1">https://openmodeldb.info/models/1x-ITF-SkinDiffDDS-v1</a> <a target="_blank" rel="ugc" href="https://openmodeldb.info/models/1x-ITF-SkinDiffDetail-Lite-v1">https://openmodeldb.info/models/1x-ITF-SkinDiffDetail-Lite-v1</a> LUTS can be found here: <a target="_blank" rel="ugc" href="https://freshluts.com/luts/74">https://freshluts.com/luts/74</a> - Subtle Blue <a target="_blank" rel="ugc" href="https://freshluts.com/luts/66">https://freshluts.com/luts/66</a> - Red Day <a target="_blank" rel="ugc" href="https://freshluts.com/luts/34">https://freshluts.com/luts/34</a> - Flat 30 I used this <a target="_blank" rel="ugc" href="https://pixabay.com/photos/woman-girl-beauty-charm-emotion-2203100/">https://pixabay.com/photos/woman-girl-beauty-charm-emotion-2203100/</a> image to create my example photos, along with that messy paint photo. I blend a noisy or messy image over my reference image to allow more change while adhering to the features of the image. Sidenote: the workflow is pretty heavily focused on img2img but it can absolutely be used for txt2img as well, just set the img2img group to bypass and connect the empty latent to the sampler. edit: I had forgotten to mention both here and in the workflow that at the end I blend really fine noise into the final image. I'll put the image on here.

Files with SHA256: 0ed75dd0c269d0162fd721f5157658aa212d61faec4eaa734c509eb7f451774b