Subject Plus — v1 (Vibes Update) 🎨📷
Subject Plus v1 is a Z-Image LoRA trained for moody, aesthetically consistent photographic realism.
This version intentionally shifts the base look away from neutral, stock-style realism and toward a more cinematic, photographer-driven aesthetic. The model is slightly overfit by design to stabilize this look across different subjects and environments.
The goal is not maximum flexibility, but consistent visual character. ⚙️
Workflow: Subject Plus Vibes V1
👉 https://discord.gg/4zbGm5j6jW
Recommended Settings ⚙️
LoRA strength: ~0.7 (higher values may degrade image) 🔧
Prompt length: 500–1300 characters ✍️
Trigger: sbjpls
Minimal adjustment required across portraits, products, and wide shots 🎛️
Primary Use Cases ✅
Shoulder-up studio and environmental portraits 🏙️
Editorial and lifestyle photography 📸
Product and commercial images with controlled lighting 🛍️
Landscapes and architectural scenes with atmosphere 🌄
Technical Characteristics 🧪
DSLR-style sharpness without excessive micro-contrast 🔹
Natural lighting response with preserved shadow detail 🌗
Clear subject separation and depth cues 🎯
Stable color behavior across indoor and outdoor scenes 🌤️
This LoRA acts as a realism + aesthetic bias layer, not a full style override.
Training Notes 📝
Reduces generic “stock photo” output ❌
Slightly overfit to lock in mood, contrast, and tonal behavior 🔒
Optimized for consistent results rather than wide stylistic range 📏
Poor prompting and weak base workflows will be more visible with this model ⚠️.
Intended Audience 👤
Users prioritizing photographic realism with aesthetic intent 🎨
Creators producing professional or editorial-style outputs 🏆
Users comfortable working with constrained, opinionated models ⚡
Not designed for neutral realism or broad aesthetic experimentation ❌
Description
FAQ
Comments (16)
What's the recommended strength?
0.85 works for me most of the time.
At 0.85, it produces artifacts and perspective distortions, and Laura is not intended to complement any person's trained laura.
@Lw24AI Try using this solution https://www.reddit.com/r/StableDiffusion/comments/1pi8ixb/multiple_lora_solution_in_zimage_also_other_models/
Whenever I use it, my photos get washen out and are worse than not using the LORA. I tried different strengths. Any special requirements for this LORA?
Have you tried the workflow? Its at the discord link in the description
@UltraMuse the discord link says invite invalid
@AntiqueHuman try this one
https://discord.gg/4zbGm5j6jW
@UltraMuse It should not be required to register on Discord and get verified through registering at another website. You should present the workflow directly, here on Civitai. I did not see your workflow, it was too complicated to download it.
@tastefulaiDA Agreed, Discord is also a security and privacy risk. Asking people to sign up for it is asking people to take unacceptable risks.
@thelustriai It uses Discord's own API for login. It's standard practice for OAuth login verification. If you're unsure, you can log into Discord first and then visit the site; it should still work.
@UltraMuse It should not be required at all to visit an external service or webseite just to download a workflow. Why not publish it here? If you use that method to lure people into your Discord, it won't work. People will not hang out there longer than necessary if you enforced the visit.
Can you share the workflow?
Hi it's in the discord free.
www.ultramuse.art
@UltraMuse even with that simple workflow of z-image of urs doesnt give this quality of ur image
@markhassain3712 I uploaded a new checkpoint see if that works better.
Details
Files
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.