SPECTRUM版本spectrum_v2.0 (ID: 601750)

"This is a collection of highly realistic SD1.5 models, featuring both artistic and photorealistic styles. The model has been carefully tuned for maximum photorealism with minimal prompting needed to achieve stunning results. All showcase images have been generated without the use of negative images to demonstrate what is possible on a bare prompt. The model possesses a deep understanding of what a photo is and can accurately interpret even fantasy-like prompts. However, if the prompt tends to be more fantastical in nature, it may deviate from the photo and require additional prompting or the use of negatives. A VAE model has already been baked into the system."

描述:

This is a merge model that combines various of my favorite realistic SD1.5 models with some more artistic ones for style and consistency. You are particularly pleased with the output generated by this model, which has been highly tuned for photorealism with minimal excessive prompting needed to achieve stunning results. The showcase images have been generated without negatives to demonstrate what is possible on a bare prompt. Since the model is knowledgeable about what a photo is, you can avoid using photos when prompting if preferred. However, if your prompt tends to be more fantasy-like, the model may turn away from photo and you will need to tweak your prompt or add negatives accordingly. The VAE is already baked in, meaning it has been pre-trained on a large dataset of images and can generate high-quality output based on input prompts.

训练词语: