April Wrapped: Every Launch From a Very Busy Month
ComfyUI released a dozen updates in April, including native video editing, a redesigned model manager, and canvas improvements. The changes target faster iteration for AI image and video workflows.
ComfyUI shipped more than a dozen feature updates and fixes in April 2024, spanning video editing, model management, and canvas tools. The release cadence reflects the platform's push to streamline node-based AI workflows for creators working with generative image and video models.
What's new
The biggest addition is native video editing inside ComfyUI. You can now trim, split, and arrange video clips directly in the node graph without round-tripping to external editors. The update also introduces a redesigned model manager that surfaces metadata, previews, and version history for installed checkpoints and LoRAs, making it easier to audit which models are in use.
Other notable changes include:
- Canvas pan and zoom improvements: smoother navigation in large workflows, with keyboard shortcuts for frame-all and reset-view.
- Batch processing speed-ups: parallel execution for certain node types, cutting render time for multi-image exports.
- Error reporting overhaul: inline warnings now show which node failed and why, reducing guesswork during debugging.
- Custom node search: a filterable sidebar that indexes community extensions by category (upscaling, inpainting, motion, etc.).
How it fits your workflow
If you're prototyping AI-generated video or iterating on Stable Diffusion outputs, the video editing tools let you stay inside one environment instead of bouncing between ComfyUI, Premiere, or DaVinci Resolve for basic cuts. Editors and motion designers testing text-to-video models (Runway, Pika, or local Stable Video Diffusion setups) will find the trim-and-preview loop faster.
The model manager update helps VFX artists and animators who juggle dozens of fine-tuned checkpoints. Instead of renaming files or keeping a spreadsheet, you get thumbnail previews and notes in the UI. It's similar to what Automatic1111 offers but integrated into the node graph.
Batch processing improvements matter most for creators rendering style frames or concept art at scale—think 50 variations of a character design in one pass. The parallel execution cuts wait time, especially on multi-GPU rigs.
Custom node search addresses a long-standing pain point: ComfyUI's ecosystem has hundreds of community extensions, and finding the right upscaler or controlnet node used to mean scrolling or guessing. The new sidebar groups nodes by function, so you can filter for "video" or "face" and see what's installed.
What it costs
ComfyUI remains open-source and free to download. You run it locally or on your own cloud instance. Hosting and GPU costs are separate and depend on your setup. Visit the ComfyUI site for installation guides and system requirements.
Read the original announcement on ComfyUI ↗