Dreamina 3.1 vs Qwen Image on an abstract Bulbasaur prompt Think I prefer Dreamina here. Mostly because it kept his skin correctly colored. Going to run some more test.. feel free to give s your favourite prompts!

A welcomed feature I’ve been using for awhile now

Chances of GPT-5 having better image edit functionality?

In other news, just went out to clean this old commuter 2004 CRV I have in my driveway and someone has cut the catalytic converter off it 😬 That’s annoying … had planned to sell it this week

Seems good

New guide from Inner Reflections in the thread below

Nathan updated his scraped data from the Wan discord chat. Gold mine of searchable knowledge in here

With added audio via MMAudio v2 on FAL

How about a 1 step Shrek replacement A.I.Warper ?

Turbo Lora for Qwen image! huggingface.co/lightx2v/Qwen-…

Testing Runway Aleph. It really struggled with a lot of the things I was trying. It’s also incredibly slow often waiting well over 5mins for a 5s clip which deterred me from continuing my tests tbh. When it works, it works well.

Qwen image 2 step tests Using the 4 step lightning lora with res_2s sampler