Type | |
Stats | 61 3 |
Reviews | (4) |
Published | Mar 26, 2025 |
Base Model | |
Training | Steps: 4,800 Epochs: 5 |
Usage Tips | Clip Skip: 1 Strength: 0.8 |
Hash | AutoV2 8FAFBA413F |
This model is specifically designed to create modifications to your architectural images, converting them into winter renders. The best use case for this model is to use it in conjunction with ControlNet. You can specify both a strong ControlNet to preserve the details of the original image by only changing the time of year and lighting of the scene, and a weak one to get modifications to the architectural object. In any case, ControlNet will not preserve absolutely all the details you want, and for optimal results you may need to post-process in Photoshop, with the original image and AI result overlaid, after which you can erase the wrong details and match the images in colors and brightness.
The optimal LoRA weight is 0.8-0.9.
This is a workflow you can start with:
https://civitai.com/models/1401477