Ugh, another post hyping up FLUX with the usual “best text-to-image model” headline and a link to try it out. As someone who edits images every day, my take on FLUX is that it’s solid with structure and lighting—the outputs are clean and don’t mess up too badly, which is way more comfortable than some of the earlier models.
But “best”? I’ll just laugh at that. In text-to-image, no one model dominates everything—portraits, textures, stylization—each has its own strengths. For wedding photo post-processing, I care more about controllability: smooth skin transitions, fabric folds that don’t blur into a mess, and faces that don’t get reshaped by the model. FLUX is fine for laying down a rough direction, but for the details, I still have to layer them in manually. No model out there can replace post-processing, at least not yet.