Qwen-Image-2512 Trainer V2
Fine-tunes the Qwen-Image-2512 model using LoRA to teach it new subjects, objects, or styles from your images and captions.
image_data_url (required)
URL to a zip archive containing image + caption pairs. Each image (e.g., photo.jpg) should have a matching caption file (e.g., photo.txt).
Supported formats: .jpg, .jpeg, .png, .webp, .heic, .heif, .avif, .bmp, .psd, .tiff
default_caption (default = null)
Fallback caption for images without a .txt file. If not set and a caption is missing, training fails.
steps (default = 2000)
The ideal number of training steps depends on the difficulty of your task.
Simple tasks (e.g., learning a basic object or loose style) often work well with 1000–2000 steps.
Complex tasks (e.g., accurate character likeness, detailed faces, or precise artistic styles) usually require more steps — commonly 2000–4000+.
Tip: Start with the default (2000) and experiment. Train a few versions with different step counts, then compare the results. More steps can improve learning but may cause overfitting if your dataset is small or not diverse enough.
learning_rate (default = 5e-4)
Use 1e-4 for slower/conservative learning, 1e-3 for faster/aggressive learning.
How the Training Works
Image-caption pairing: Each image (e.g., photo.jpg) is paired with its caption file (photo.txt or photo.TXT).
Aspect ratio bucketing: Images are assigned to the nearest bucket matching their aspect ratio, preserving natural proportions.
| Bucket (H×W) | AR | Orientation |
|---|
| 1344×576 | 3:7 | Portrait |
| 1280×720 | 9:16 | Portrait |
| 1248×832 | 2:3 | Portrait |
| 1152×864 | 3:4 | Portrait |
| 1152×896 | 7:9 | Portrait |
| 1024×1024 | 1:1 | Square |
| 896×1152 | 9:7 | Landscape |
| 864×1152 | 4:3 | Landscape |
| 832×1248 | 3:2 | Landscape |
| 720×1280 | 16:9 | Landscape |
| 576×1344 | 7:3 | Landscape |
Caption dropout: 5% of the time, captions are dropped during training to help the model generalize.
Tips for Good Results
Dataset
- Size: 5-10 images minimum, 15-30 optimal for subjects, 20-50 for styles.
- Quality: High resolution images (1024px+) with good lighting and no watermarks.
- Variety: Different poses, angles, lighting, and backgrounds for better results.
Captions
Write clear, descriptive captions in a consistent style:
- Good:
a golden retriever sitting on a red couch, soft afternoon lighting
- Bad:
dog
Tip: Use an image caption model like Moondream-3
Trigger phrases: For subject training, include a unique trigger word or phrase in every caption (e.g., a photo of ohwx person). Use the same trigger when generating new images later.
Troubleshooting
| Problem | Signs | Solution |
|---|
| Overfitting | Outputs match training images exactly, poor generalization | Reduce steps, add more varied images |
| Underfitting | Outputs barely resemble your training data | Increase steps, improve captions or dataset quality |
| Caption errors | Training fails | Add missing .txt files or set default_caption |
fal-ai/qwen-image-2512-trainer-v2