Train LORA to Qwen-Image
Hello, everyone.
I would like to train a LORA with Qwen-Image, but I can't find any information on how to do it. What should I do? I have the dataset with the captions created.
Thank you very much.
Great, thank you very much :-)
Create a photorealistic 3‑D rendering of a narrow office toilet, dimensions 2.7 m (length) × 1.0 m (width). Modern minimalist style. Dark‑gray matte floor (RAL 7016, 30×30 cm tiles, anti‑slip). Light walls (RAL 9001 or RAL 7035, smooth finish).
Layout (to scale):
- Entrance on the short side; immediately on the right is a wall‑hung hand‑washing sink (≈45 cm wide, mounted at 80 cm). Touchless chrome faucet, frameless LED‑backlit mirror above. Small wall shelf with soap dispenser and waste bin.
- In the centre of the room sits a built‑in toilet on a raised platform (step height ≈15 cm, platform ≈80 cm long × 60 cm wide). Platform covered with the same dark gray tile, edged with a light‑gray border for contrast. Wall‑mounted toilet flushes automatically, backs onto a matte black wall panel. Small side shelf on the left of the platform for toilet paper.
- Opposite wall is clean, with a quiet (≤35 dB) recirculating exhaust fan near the ceiling.
- Lighting: recessed ceiling LEDs (4000 K) all around, spot light above the toilet, subtle LED strip under the raised platform to highlight the step.
- Hygienic features: touch‑less soap dispenser, automatic hand dryer integrated into the right wall, motion‑sensor light switch near the door.
Render requirements: top‑down plan and a 30° perspective view, high‑detail materials (glossy chrome, textured ceramic, matte floor). Accurate color values: dark gray floor (RAL 7016), light walls (RAL 9001/7035). Emphasize contrast between the dark platform and bright walls, soft yet functional illumination.
