April 2026 Summaries

ComfyUI has integrated support for HappyHorse 1.0, a cinematic video generation model developed by Alibaba, which is tailored for high-quality storytelling and creative workflows in video production. This model is particularly suited for creators focused on producing advertisements, e-commerce visuals, and social marketing videos, thanks to its emphasis on aesthetics, multi-shot sequencing, and editing capabilities. HappyHorse 1.0 provides various creation paths, including text-to-video (T2V), image-to-video (I2V), and subject-to-video (S2V), and offers cinematic features such as wide-aperture framing and atmospheric moods. The model can generate multi-shot outputs of up to 15 seconds at 1080p resolution, with a focus on maintaining consistency across transitions. It also supports editing workflows that allow for footage transformation and subject manipulation, ensuring that motion and composition are preserved. Users can access these features by updating to the latest version of ComfyUI or using Comfy Cloud.

ComfyUI, initially a community-driven open-source project, has secured $30 million in funding at a $500 million valuation, led by Craft with participation from several investors, bringing its total funding to $47 million. The platform, known for empowering creators with AI-driven workflows across various media, is now used by over 4 million users, with substantial contributions from the community, including more than 60,000 nodes and 150,000 daily downloads. ComfyUI is widely recognized in creative industries, utilized by studios and agencies for projects like AI-generated Super Bowl ads, and has become a significant skill in the job market. The new funding will enhance its development, focusing on cloud infrastructure, collaborative workflows, local experience, ecosystem stability, and model support, while maintaining its commitment to remaining open-source and community-driven. This approach allows for rapid innovation and scalability, ensuring that the platform continues to grow and evolve with its user base.

Seedance 2.0, ByteDance's latest video generation model, enables users to create videos featuring real people with consistent identity and authentic facial expressions. This model enhances video production with industrial-grade controllability, native audio-video synchronization, and cinematic control over camera movements and lighting, ensuring a stable appearance across transitions. Users can mix text with images, videos, and audio clips for comprehensive directing, while a robust verification process prevents unauthorized use of likenesses, aligning with AI transparency regulations. Once verified, users receive a Group ID and Asset ID, facilitating easier future use without needing repeated verification.

OpenAI's GPT Image 2.0, now integrated into ComfyUI through Partner Nodes, represents a significant advancement in image generation by introducing reasoning before creation, which allows for more complex and accurate outputs, such as dense text and detailed infographics. Unlike previous models that struggled with edit-based workflows, GPT Image 2.0 maintains structural fidelity during targeted edits, ensuring that changes do not inadvertently affect other parts of the image. It can produce up to eight consistent images from a single prompt, preserving continuity across a series, which is particularly useful for tasks like storyboarding and creating product variants. This model seamlessly integrates into hybrid workflows, allowing users to utilize it for text-heavy frames and then transition to local models for further processing.

Quiver is an advanced tool for structured SVG generation within ComfyUI, offering capabilities such as text-to-SVG and image-to-SVG conversion, which are particularly beneficial for creators seeking vector deliverables instead of pixel-based graphics. This tool, which acts as a Partner Node in ComfyUI, is designed to generate high-quality, scalable vector graphics that can be used for logos, icons, illustrations, and converting raster images into vectors. It allows users to explore various design directions and refine elements like typography and geometry through editable vector files. Quiver simplifies workflows by enabling the direct conversion of images and sketches into vectors, eliminating the need for manual tracing or third-party tools, and is optimized for design and development processes.

ACE-Step 1.5 XL is an advanced open-source music generation model that leverages a 4-billion-parameter Diffusion Transformer decoder to produce high-quality audio locally on consumer hardware. This model, part of the ACE-Step framework, offers three variants: xl-base for versatility, xl-sft for peak audio quality, and xl-turbo for rapid processing, all commercially licensed under the MIT license and trained on legally compliant datasets. It showcases commercial-grade sound quality, comparable to leading music models, with ultra-fast generation times and the capability to create compositions ranging from 10-second loops to 10-minute pieces. Supporting over 1000 instruments and styles, as well as lyrics in more than 50 languages, ACE-Step 1.5 XL provides a flexible tool for music creators, enabling them to generate diverse and richly detailed audio outputs on platforms like ComfyUI or Comfy Cloud.

ERNIE-Image, Baidu's open-source text-to-image model licensed under Apache-2.0, is now available on ComfyUI, utilizing an 8 billion parameter Diffusion Transformer to deliver precise text rendering, strong instruction following, and structured visual generation across a wide stylistic range. It features a built-in Prompt Enhancer, which uses a 3 billion parameter model to expand short inputs into richer prompts, enabling the creation of complex and detailed artworks, from educational infographics to cinematic posters and editorial fashion photography. The model supports dense and layout-sensitive text in multiple languages and is compact enough to run on 24 GB VRAM, allowing for both realistic and stylized image generation. Users can download and explore different versions such as the Main SFT model for high-quality output and the ERNIE-Image-Turbo for faster generation, both easily accessible through Comfy Cloud, enhancing creative workflows for artists and developers alike.

Sonilo, an AI-powered service that generates frame-synced soundtracks for videos, is now integrated into ComfyUI via Partner Nodes, allowing users to produce synchronized music directly within their existing workflows. This integration enables creative professionals to generate music that aligns with the length, pace, and emotional arc of their videos, facilitating faster iterations and maintaining creative flow without the need for manual music sourcing or timing. Sonilo's generated music is commercially ready, eliminating additional licensing complexities and making it ideal for use in advertising, video production, visual effects, and content creation. By embedding this capability within ComfyUI, developers and pipeline engineers can automate soundtrack generation, streamlining the collaboration between creative and technical teams.

Seedance 2.0, a cutting-edge video generation model, is now available in ComfyUI, allowing users to transform text, images, video, and audio into high-quality videos with synchronized audio and cinematic camera control. This model excels in multimodal reference-based generation, drawing from up to nine images, three videos, and three audio files to produce coherent outputs that maintain object details, textures, and character features. It replicates camera movements, syncs visuals to audio at a phoneme level, and ensures stylistic consistency across videos. Seedance 2.0 also offers precise video editing tools, enabling users to replace subjects, edit objects, and inpaint scenes without needing to regenerate videos from scratch. Additionally, it supports seamless video extension, allowing clips to be naturally extended forward or backward in time while maintaining continuity. Users can get started by updating their ComfyUI or accessing Comfy Cloud to explore Seedance 2.0's capabilities.

Wan2.7, Alibaba's latest video generation model, is now accessible in ComfyUI through Partner Nodes, offering a significant upgrade over version 2.6 by enhancing image quality, audio, motion dynamics, stylization, and consistency. This update introduces a full suite of creative workflows, allowing for up to 5 real-person image inputs, vocal timbre references, and 3×3 grid-based image generation. The model supports a multimodal video pipeline that includes text, image, audio, and video inputs across four task types: Image-to-Video, Text-to-Video, Video Continuation, and Reference-to-Video, as well as Video Edit capabilities. Users can explore various workflows such as creating video from images, generating video from text prompts, continuing existing clips, and editing or replicating videos via text prompts or style transfer. To utilize Wan2.7, users need to update ComfyUI to version 0.18.5 and select the appropriate workflow from the Template Library.