ComfyUI v0.43.0: MOSS-image support, new partner nodes and Pad Image

The ComfyUI launcher tool has released v0.43.0. The changelog on GitHub lists MOSS-image support with color depth control and rope scaling, new partner nodes for HeyGen, ClarityAI, BFL and Bria, a Pad Image node with mirror padding, a top-p sampling fix, and a frontend update to 1.57.1.

What it means

For those running models locally, the main interest is in the changes to MOSS-image: rope scaling and color depth control have been added. This expands the image generation capabilities of this model on your own machine. Among the partner nodes, HeyGen Voice instant clone and text-to-speech have appeared, along with Crystal Video Upscaler from ClarityAI, 2K/4K resolutions for FLUX 3 Video and FLUX 3 Max nodes from BFL, and FIBO 2.0 Image Generate/Edit from Bria. Partner nodes require the corresponding accounts with these services. The Pad Image node with mirror padding and alignment to a multiple size will come in handy when preparing images for models with fixed resolution requirements. The top-p sampling fix affects sampling correctness. The developer does not specify the hardware requirements or VRAM amount for the new features. You can follow news about tools for running neural networks locally in the CPU3D news feed.