An early test I did. The top left tile was generated by Flux Fill, pretty much identical to the others. This can potentially be used for upscaling with high fidelity.

Suno v4.5 sounds great.

Now available! PixelWave FLUX.1-schnell 04 is available!

I spent the past months building OpenMuse.ai. Growing from the Banodoco community, my hope is that it’ll become a place people find extremely inspiring/empowering: a worthy home for the open source AI art movement🌱 Beautiful example of an LTXV LoRA by Mohamed Oumoumad:

📢 Comfy Town Hall 🤝 Special Guest: POM 🎤 Host: Purz.ai ⌚ 5pm EST / 2pm PST Today we sit down with Peter O'Malley — the visionary behind Banodoco, a thriving Discord community that empowers open-source devs and artists to build and learn together.

Benny Govaerts I just uploaded it to Hugging Face with some instructions on how to use it in ComfyUI. If you want to get the highest quality depth maps with Depth Anything V2, you should run the FP32 models, and the giant model will provide more details. huggingface.co/Nap/depth_anyt…

Looks like Qwen2VL-Flux ControNet is actually the best Flux ControlNet for depth. At least in the limited tests I ran. It even surpasses InstantX Union Pro 2.0 Depth. All tests were done with the same settings and the recommended ControlNet values from the original projects.

AI ad (scam?) slop spotted in the wild.

HiDream I1 Portraits - Dev vs Full Comparisson - Can you tell the difference? The Dev version is surprisingly better than the Full version. And many cases, despite Full taking more than 3x the time per image. Took ~4m27s to generate both images on a 4090. Made in ComfyUI

𝙷𝚒𝚌𝚔𝚖𝚊𝚗 I always wondered about that myself. What harm does it do to the human spirit when social norms say you are supposed to completely ignore the presence of another human being around you? Even on a bus, elevator or airplane. Really makes it feel like we live in a human zoo.

Ethan Mollick You are the one generating the thoughts, not the LLMs. The LLMs are just arranging symbols in a particular order to satisfy YOUR requests. If you don’t read the LLMs’ reply, no thoughts have been generated. Is that really a mystery or is it an illusion?