This workflow compares the performance and output quality of CogView 4 with SDXL, FLUX, and Lumina using RealVis 5 SDXL, Flux Dev and Lumina 2 (Alpha VLLM) as base models.

There is no ComfyUI integration for CogView 4 yet, so it uses a wrapper.

This workflow is ideal for benchmarking different AI models, comparing realism and coherence in generations, and evaluating processing speeds for different pipelines.