QWEN IMAGE 2.1 Me v0.1 (Experimental Checkpoint)

QWEN IMAGE 2.1 Me v0.1 is an early-stage experimental checkpoint tailored for photorealistic rendering and rich atmospheric scenery. Built on the Qwen image architecture, this model supports both Text-to-Image generation and Image-to-Image / Image Editing workflows.

🧪 Model Status & Dataset

Version: v0.1 (Alpha / Experimental)
Focus: Realism, skin textures, micro-details, and immersive landscapes/scenery.
Dataset: Small, highly curated experimental dataset.
Note: As a test release, output consistency may vary. Community feedback and generation samples are welcome to help improve future iterations.

🎨 Core Capabilities

Text-to-Image: Generates detailed photorealistic scenes and portraits from text prompts.
Image-to-Image & Editing: Seamlessly handles image-guided generation, style transfers, and targeted image edits.
Scenery & Realism: Fine-tuned to capture natural lighting, environmental depth, and lifelike material textures.

⚙️ Recommended Generation Settings

To achieve optimal results with this checkpoint, use the following baseline parameters:

Sampler: euler
Scheduler: simple
CFG Scale: 3.5 - 5.0
Sampling Steps: 20 - 30 (Recommended for balanced speed and detail quality)

📝 Prompting Tips

Positive Keywords: RAW photo, photorealistic, cinematic lighting, highly detailed scenery, natural textures, sharp focus.
Image-to-Image: Keep denoising strength around 0.4 - 0.6 for balanced structural preservation during edits.

Qwen Image 2.1 Me - v0.2 fp16