By the time Flux came around, node graphs really need to be more human-readable. When I look at other people’s shared workflows, it’s often a mess of dozens of nodes connected like a spider web, and even the author themselves might not remember what each line does. Flux’s response to prompts and sampling is different from SD, so a lot of the complex stacking from the old SD era doesn’t need to be copied over—it just slows things down.
Now I have a habit when organizing workflows: merge nodes that can be merged, pull key parameters to a prominent spot, and name groups in Chinese to clearly label what each section does.
When sharing, instead of showing off a dense, flashy screenshot, it’s better to explain the logic and why each key node is set up that way. A workflow that people can understand and modify is one they’ll actually use.