v6.14.0-rc2
Breaking Changes📦 invokeaiView on GitHub →
⚠ 1 breaking✨ 15 features🐛 4 fixes🔧 17 symbols
Summary
This release introduces significant new features including video generation, support for multiple new AI models (Krea, Ernie, Ideogram, Flux.2), multi-GPU acceleration, and native Intel XPU support. It also brings performance improvements and enhancements to the Canvas and workflow management tools.
⚠️ Breaking Changes
- Not all LoRAs will currently load. Please report those that don't in Issues: https://github.com/invoke-ai/InvokeAI/issues.
Migration Steps
- Configure InvokeAI for multi-GPU generation if you have two or more GPUs installed: https://invoke.ai/configuration/invokeai-yaml/#multi-gpu-generation
- Set VAEs to run on the CPU via the Model Manager to reduce VRAM contention.
✨ New Features
- Video generation support via Wan 2.2 (text-to-video, image-to-video, image-to-image-video).
- Support for Krea.2-Turbo, Krea.2-Raw, Ernie Turbo, Ideogram 4, and Flux.2 Dev models.
- Anima controlnets and inpainting support.
- Flux.2 PiD support for super resolution up to 4K.
- Multi-GPU support for parallel generation across multiple GPUs.
- Native Intel XPU support.
- VAEs can now be run on the CPU to reduce VRAM contention.
- Improved support for LoRA sliders, including UNet-only sliders and editing min/max/default values.
- Pressure-sensitive brush opacity on tablets and touch screens in Canvas.
- Fine-grid hint added to the move tool in Canvas.
- New transparency lock for Gradients and Shapes in Canvas.
- Clip Strokes extended to Bbox and Rect/Oval shapes in Canvas.
- Optimized graph execution for faster complex workflows.
- Added a LoRA collection picker node to stack multiple LoRAs.
- Workflows can now call other workflows.
🐛 Bug Fixes
- Fixed ComfyUI model_sampling keys when loading Anima checkpoints.
- Relaxed 5s pytest-timeout marks that flake on slow CI runners.
- Ignored ComfyUI model_sampling keys when loading Anima checkpoints.
- Don't hold VRAM in an idle server (CUDA context at startup).