Hey everyone, welcome back!
Today we’re taking a close look at CogView 4, the latest image generation model out of China — and this one definitely deserves some attention.
I ran CogView 4 through a full side-by-side comparison with SDXL, Flux, and Lumina 2 using some really challenging prompts — everything from LEGO-style Einsteins to painting pandas in the bamboo forest. The goal? See which model actually understands prompts best, delivers sharp visuals, and works well in a real test environment.
I used ComfyUI and tested everything in a shared workflow, which you can check out for yourself. All the links are below:
🔗 Links & Resources:
CogView 4 Model: https://huggingface.co/THUDM/CogView4-6B
GitHub Repo: https://github.com/THUDM/CogView4
Online Demo (Miniapp): https://huggingface.co/spaces/THUDM-H...
ComfyUI: https://github.com/comfyanonymous/Com...
My Workflow: https://openart.ai/workflows/jaguar_l...
So the big question: Should we add CogView 4 to CogniWerk AI?
It’s not fully integrated yet, but the potential is definitely there — and I’d love to hear your thoughts. Try it out, compare results, and let me know what you think!
Thanks for watching — if you’re into AI model comparisons, deep dives, and testing the limits of creative tools, don’t forget to like, subscribe, and leave a comment with your favorite model.
Catch you next time!
– Benjamin from CogniWerk 🚀
コメント