Famegrid Qwen

Details

Model description

This is a first version of FameGrid for Qwen-Image, fine-tuned on a curated dataset social media influencer photos. It took tons of work and significant server costs to build this LoRA.

I'm it early so you can test and give feedback.

Help Us Finish the Final Version This beta is functional, but the full release needs more training, testing, and refinement. Server time is expensive your support gets the final version out faster.

If you find this useful, consider contributing: Donate via Stripe

Even small amounts help cover GPU costs and speed up development. Donations of $10 or more will get access to the premium version (trained at higher res and lora rank/size) of the model when it becomes available. 🙏

Donations over $5 will recieve 600+ PNG (from this set) with complete workflows and prompts for Famegrid Qwen.


Important Notes

  • Download Workflow here

  • Start your prompts with "A iPhone photo of a IGMODEL woman (detailed skin:1.3)...."

  • Lower CFG enhances realism; higher CFG improves sharpness but may reduce naturalness.

  • CFG and weight are your levers. If you use low CFG then user low weights and higher weight for high CFG.

  • This model loves long and detailed prompts, I suggest using an LLM to help craft or expand on your prompts. 800 characters minimum.

  • Boost the strength if your photos aren't coming out realistic.


Who is this for? The training data was hand-selected for:

  • Natural, confident poses

  • Clean composition and lighting

  • Modern, polished social media aesthetic

It works well for:

  • Lifestyle and influencer-style photos

  • Product shots with human context

  • AI character design and virtual photography

  • Social media content creation

Trigger word: IGMODEL


Recommended Settings (Qwen-Image)

  • Model Flow: 2

  • Sampler: res 2s

  • Scheduler: simple

  • CFG: (1-4) 1 recommended

  • LoRA Weight: 1.0 – 2 (1 recommended, boost of image loses realism)

FameGrid Qwen Checkpoint (Coming Soon) – Final version will be uploaded to CivitAI once training is complete.

Images made by this model

No Images Found.