ComfyUI Outpaint Mask Editor released
The ComfyUI Outpaint Mask Editor has completed its first registry release, supporting cropped-coordinate composite output. The tool is used for mask editing and compositing in outpaint scenarios.
The ComfyUI Outpaint Mask Editor has completed its first registry release, supporting cropped-coordinate composite output. The tool is used for mask editing and compositing in outpaint scenarios.
Creator Animation offers an on-site AI Film Studio where creators can use prepaid credits to directly generate multi-scene films and keep 80% of their work's revenue under an 80/20 split. The platform has no ad interruptions or algorithmic review thresholds, viewers can watch for free, fans can buy a $9/month Download Pass for offline saving, and creators start with 10GB of free storage.
Creator Animation launched a direct distribution platform for AI film creators, where creators retain intellectual property and take 80% of revenue. The platform offers 10GB of free storage, $0.40/GB expansion, and tiered plans: Indie (100GB), Studio (500GB), Feature (2TB), and Filmmaker (5TB); viewers can stream for free without an account, and a $9/month Download Pass unlocks high-bitrate offline downloads.
Among the seven white-label AI SaaS tools for agencies and software founders in 2026, Wireflow ranks first for hiding the entire generation layer behind the company's own API, so clients cannot see which models are running underneath.
Technical director Alexandre Grégoire built a purely gesture-controlled AI installation with ComfyUI during the first session of the Generative AI Bootcamp for Film and TV, with no mouse, keyboard, or controller.
The Vimeo Enterprise August update is live, bringing 10 improvements to Webinars, Live events, and Vimeo AI. Registration forms support AI translation into up to 8 languages, registrants receive personal Magic links for one-click entry, and event emails can be set with an independent sender identity.
StabilityMatrix v2.16.4 has been released, focusing on fixing the issue where importing a package silently deleted existing files; it now shows the folder size and file count and asks for confirmation. When the shared model folder is turned off, only links created by Stability Matrix are removed, and the real directory is no longer touched; the issue where self-nested workflow library links froze the page has also been fixed.
Trendlume v0.2.0 adds a "Trending Hub" (/trends) that aggregates public hot lists from 7 major platforms—Weibo, Douyin, Zhihu, Toutiao, Xiaohongshu, Bilibili, and Baidu—using a dual-source fault-tolerance mechanism with the primary source 60s API and the backup source xxapi.
College Path Studio used Pictory to scale video content output without adding staff, boosting team efficiency by 450% and saving 250% on video production costs. This small team of about four people used Pictory to build videos scene by scene, covering free and paid courses, product demos, YouTube product shorts, and long educational videos, while keeping the team off camera.
Embedded text is exactly why you're still opening Photoshop. Layers can now extract it. Select a layer, click Extract, and the copy becomes live text, already automatically matched to the closest font in the font library. Retype. Change the font. Re-layout. Changing a line of copy takes two clicks instead of regenerating. Layers, powered by Uni-1.
The Hugging Face team rebuilt most of AUTOMATIC1111's functionality with Gradio Workflow, creating a single-canvas Workflow1111 composed of 11 media pipelines and 73 nodes.
Why it matters: Replicating all of A1111's functionality with 73 nodes offers another way to organize node-based workflows, comparable to ComfyUI.
Luma enterprise governance features are now live. Team isolation. Set spending caps per project. Billing doesn't stall because of spending caps.
Google launched Google Pics, an image creation and editing tool under Workspace, built on the Nano Banana image generation and editing model. It will roll out over the coming weeks to all Google AI Pro and Ultra subscribers and most Workspace business customers.
Why it matters: Google is bringing image generation and editing tools into Workspace apps, giving teams a basis to decide whether to migrate their image production workflows.
Gradio's built-in gr.Workflow describes AI pipelines as graphs of typed nodes, offering a drag-and-drop canvas where each node can run and intermediate results are visible; the same graph is also a REST API and can be deployed to Hugging Face Spaces with one click.
Why it matters: Gradio turns multi-step AI pipelines into a visual graph and automatically exposes REST interfaces, giving teams a basis to assess migration costs against their existing production workflows.
NVIDIA released the Magpie TTS Multilingual open-weight model with 364M parameters, supporting 12 languages. This update adds Modern Standard Arabic, Korean and Brazilian Portuguese, and improves multilingual synthesis quality through updated training data.
A Hugging Face blog post explains that Diffusers now natively supports Nunchaku's 4-bit diffusion inference; loading a quantized checkpoint only requires calling from_pretrained(), with no local CUDA compilation needed.
Why it matters: Nunchaku's 4-bit quantized inference is now natively integrated into Diffusers, allowing creators to evaluate the trade-off between VRAM usage and speed for image generation on consumer-grade GPUs.
Hugging Face and Cerebras jointly demonstrated a real-time speech-to-speech pipeline, using Cerebras to accelerate Gemma 4 31B inference, Nvidia's Parakeet for speech recognition, and Alibaba's Qwen3TTS for speech synthesis, with each layer being modular and open-source replaceable.
OpenAI rebuilt its WebRTC technology stack to support low latency, global scale, and smooth conversation turn-taking for real-time voice AI.
ChatGPT supports creating and optimizing images through clear prompts, allowing users to iterate on designs and generate high-quality visual content in minutes.
Descript leverages OpenAI reasoning models to achieve large-scale multilingual video dubbing, automatically localizing large content libraries while preserving the original timeline and semantics.
The Gradio team open-sources Daggr, a Python library that lets you chain Gradio apps, ML models, and custom functions with code and auto-generate visual canvases. It supports inspecting any node's output, re-running individual steps, configuring fallback nodes, and automatically saving workflow state and cached results. Daggr requires Python 3.10 or later, can be installed via pip, and workflows can be deployed to Hugging Face Spaces for sharing.
Higgsfield leverages OpenAI's GPT-4.1, GPT-5, and Sora 2 to let creators produce cinematic, social-media-ready videos from simple inputs.
Hugging Face AI Sheets releases a major update adding vision support, allowing users to directly view, analyze, and extract information from images, as well as generate and edit images in spreadsheets without writing code. Powered by Inference Providers, the tool can call thousands of open-source models, supports extracting structured data from receipts, documents, and charts, and can generate images from prompts and produce batch variants. Results can be exported as CSV or Parquet.
Hugging Face adds a watermark parameter to Gradio, allowing developers to add visible watermarks to AI-generated images, videos, and text with just one command. For images and videos, the watermark file is specified via the watermark parameter of gr.Image and gr.Video, supporting images or numpy arrays, and QR watermarks can also be used; for text, the watermark parameter of gr.Chatbot automatically appends an attribution when users copy AI responses.
By connecting Claude to Hugging Face Spaces through the Hugging Face MCP Server, you can call models such as FLUX.1 Krea [dev] and Qwen-Image to generate images.
Invideo AI leverages OpenAI's GPT-4.1, gpt-image-1, and text-to-speech models to turn creative ideas into professional videos in minutes, increasing video production speed by 10×.
Hugging Face launches a new OpenAI Whisper deployment on Inference Endpoints, implemented with vLLM, achieving nearly 8× improvement in RTFx over the previous version with no loss in transcription quality.
ChatGPT image generation supports creating brand visual assets and product mockups from prompts, sketches, or uploaded images. This feature is designed for brand visual and product mockup scenarios and can generate based on existing assets.
Hugging Face and Cloudflare have reached a partnership. FastRTC developers can access Cloudflare's TURN network, which spans over 335 locations worldwide, using a Hugging Face Token. They can stream 10GB of data per month for free with no credit card required. This integration will be available in FastRTC version 0.0.20 and above, and users can switch to a Cloudflare account to expand beyond the limit.
Hugging Face releases FastRTC, a real-time communication library for building real-time audio and video AI applications in Python.
Why it matters: FastRTC wraps WebRTC real-time audio and video communication into a Python library, eliminating the upfront costs of building real-time infrastructure from scratch.
The Diffusers team experimentally introduces Remote VAE, which offloads the VAE decoding process of diffusion models to remote Inference Endpoints, reducing the VRAM pressure on consumer-grade GPUs for high-resolution image and video synthesis.
Hugging Face Hub adds three serverless inference providers: Hyperbolic, Nebius AI Studio, and Novita, supporting models such as DeepSeek-R1 and Flux.1, with integrated JS and Python client SDKs. Users can configure their own API keys in account settings to connect directly to providers, or route calls through HF and be billed at standard API rates with no markup. PRO users receive $2 in inference credits per month.
HuggingChat launches Community Tools, allowing users to convert any public Space on Hugging Face into a tool that models can directly call, thereby expanding multimodal capabilities such as image understanding, video generation, and text-to-speech. Users can also build custom tools using Gradio and add up to 3 tools in their assistant. This feature is still in experimental stage.
OpenAI publishes a detailed explanation of the technical principles and safety research behind its text-to-speech model Voice Engine. Voice Engine can generate speech from text, and the official team also introduces safety research conducted around this model.
Hugging Face's PEFT library adds new merging methods for LoRA adapters, including cat, linear, svd, and TIES-based ties and ties_svd.
Segmind releases the SegMoE framework, which can create Mixture-of-Experts (MoE) diffusion models from scratch and has been integrated into Hugging Face diffusers. This release includes three model versions—SegMoE-4x2, SegMoE-2x1, and SegMoE-SD4x2—along with the segmoe code package for building custom MoE models. The models are licensed under Apache 2.0.
Why it matters: The release provides code for building MoE diffusion models from scratch along with three merged versions, enabling creators to combine multiple pretrained models into new ones.
ONNX Runtime's CUDA and TensorRT execution providers can significantly accelerate SD Turbo and SDXL Turbo inference on NVIDIA GPUs, with throughput improvements of up to 229% (SDXL Turbo) and 120% (SD Turbo) compared to PyTorch.
Hugging Face Diffusers now supports running SDXL inference on Cloud TPU v5e via JAX, achieving high-performance, low-cost deployment with JIT compilation and XLA parallelism. The demo generates 4 images of 1024×1024 in parallel on multiple TPU v5e-4 instances in about 4 seconds, with pure generation time of about 2.3 seconds; TPU v5e costs less than half of TPU v4.
The Hugging Face Diffusers team, in collaboration with the T2I-Adapter authors, provides T2I-Adapter support for Stable Diffusion XL and releases checkpoints for conditions such as sketch, canny, lineart, depth, and openpose.
Why it matters: Comparing ControlNet's parameter count and inference overhead, it presents a T2I-Adapter controllable generation solution on SDXL and reusable training scripts.
Apple and Hugging Face have ported the Stable Diffusion XL base model to Core ML, usable on Apple Silicon Macs running the macOS 14 public beta; it currently uses the ORIGINAL attention implementation, and the refiner stage has not yet been ported.
Why it matters: Apple and Hugging Face jointly present a Core ML mixed-bit quantization solution for SDXL on Mac, with measured latency for various models and compression recipes.