The LoRA was built from a series of 26 3D renderings from a Blender file of a Pepsi product. Each rendering was created at 1024 x 1024 pixels with 256-512 steps. The LoRA, based on FLUX.1 [dev], is intended to demonstrate the challenges of accurately reproducing a specific product through the use of fine-tuning. Challenges include the accuracy of the product's logo, graphics, typography, text, proportions, dimensions, and color in generated copies.