AfterDark [Z-Image Turbo]

AfterDark for Z-Image Turbo

This LoRA adds some lomo film photography styling to your images; film grain, realistic lens bokeh, realistic textures, depth of field, and lighting. Note that it often does introduce soft focus (particularly on portraits), which was intentional, but you may need to adjust the strength and your prompt to compensate if this is not the style you're after. It's been trained on a mixture of photographic content and has no specific feature in mind. Not sure Z-Image Turbo needs or uses a trigger word, but you can try: 4ft3rd4rk

This is an experimental LoRA that will update over time. I created a series of LoRAs for Flux from similar datasets, but for Z-Image Turbo, I will likely just stick with one LoRA for now.

Suggested model strength: 0.25-0.8

Notes for Next Update: I plan to adjust the training data a bit because while I wanted a soft focus lomo style, this lost too much detail for me. You can compensate by lowering the strength, but then you lose some of the lighting and other lomo style imperfections. When it came to Flux, it helped correct the plastic looking skin textures, but Z-Image Turbo is already a bit better with that to begin with. I noticed "Flux face" has been exchanged for "Z hair" in that hair is often plastic looking with Z-Image Turbo. This LoRA does make hair look more natural which is nice. Z-Image Turbo is also already aware of things like vignetting so you can get a decent vintage photography style without a LoRA at all, but it will struggle with film grain, lighting, light leaks, and other imperfections.