Startup Leverages NVIDIA RTX-Powered AI for Innovative Cooler Designs

Mark Theriault, the founder of FITY, envisioned a line of clever cooling products that keep beverages cold for longer without the mess of ice. Starting with 3D prints in his basement, Theriault faced the challenge of scaling from preliminary sketches to production-ready designs. To bring his creative vision to life, he relied heavily on AI and his NVIDIA GeForce RTX-equipped system.

Theriault's journey from concept to completion is a testament to the power of AI in product development. He tinkers with potential designs using traditional methods, from sketches to computer-aided design (CAD) to rapid prototyping, until he finds the right vision. A standout feature of the FITY Flex cooler is its customizability with fun, popular shoe charms.

For packaging design inspiration, Theriault uses the generative AI model Stable Diffusion XL, accelerated by the NVIDIA TensorRT software development kit, through the modular, node-based interface ComfyUI. ComfyUI provides granular control over every step of the generation process, from prompting to post-processing, making it ideal for advanced users like Theriault.

NVIDIA and GeForce RTX GPUs, based on the NVIDIA Blackwell architecture, include fifth-generation Tensor Cores designed to accelerate AI and deep learning workloads. These GPUs work with CUDA optimizations in PyTorch to seamlessly accelerate ComfyUI, reducing generation time on FLUX.1-dev from two minutes per image on the Mac M3 Ultra to about four seconds on the GeForce RTX 5090 desktop GPU.

ComfyUI can also add ControlNets, which help control image generation tasks like guiding human poses, setting compositions via depth mapping, and converting scribbles to images. Theriault even creates his own fine-tuned models using low-rank adaptation (LoRA) to keep his style consistent and minimize compute costs.

"Over the last few months, I’ve been shifting from AI-assisted computer graphics renders to fully AI-generated product imagery using a custom Flux LoRA I trained in-house. My RTX 4080 SUPER GPU has been essential for getting the performance I need to train and iterate quickly," said Mark Theriault, founder of FITY.

Theriault also uses generative AI to create marketing assets like FITY Flex product packaging. FLUX.1 models, which excel at generating legible text within images, are accelerated by NVIDIA and Black Forest Labs using quantization and TensorRT, reducing model size to just over 11GB of VRAM and improving performance by 2.5x.

To simplify using these models in ComfyUI, NVIDIA created the FLUX.1 NIM microservice, a containerized version of FLUX.1 that supports FP4 quantization and TensorRT. Theriault uses the Blender Cycles app to render final files and NVIDIA's AI Blueprint for 3D-guided generative AI to ease the positioning and composition of 3D images.

Finally, Theriault leverages large language models to generate marketing copy, tailored for SEO, tone, and storytelling, as well as to complete his patent and provisional applications, tasks that usually cost thousands of dollars in legal fees and considerable time.

"As a one-man band with a ton of content to generate, having on-the-fly generation capabilities for my product designs really helps speed things up," added Theriault.

Every texture, every word, every photo, and every accessory was a micro-decision, and AI helped Theriault survive the 'death by a thousand cuts' that can stall solo startup founders.