OpenScreen released v2.1.2, with the full changelog covering v2.1.1 to v2.1.2. This version was released by EtienneLescot on October 19, and the project also called on users to star the repository to help more people discover OpenScreen.
AIArtEngine released v8.0.0, with the full changelog covering changes from v7.8.0 to v8.0.0. The original text did not disclose specific features, parameters or availability details.
CG Channel summarized CG tool updates from the past week that were not separately reported: Maxon's free motion graphics software Autograph 2027.0 adds an MCP Server that can connect to external AI tools such as Claude and ChatGPT.
Higgsfield officially released a CLI usage guide explaining how to generate images and videos in the terminal, scripts or coding agents without opening the web interface. The CLI logs in via browser without relying on an API key, shares the same account and credits as the web version, supports models such as Seedance 2.5, Kling v3.0, Veo 3.1 and Wan 3.0, and allows checking credit prices first with generate cost and outputting with --json for script invocation.
KohakuMoss released 0.1.0 on PyPI on October 18, 2026. It is a fine-tuning framework built by KohakuBlueleaf, the author of LyCORIS, for MOSS-Image-Base and MOSS-Image-SFT, filling in the self-owned image training step that neither the model release nor the community few-step adapters covered.
Why it matters: Addressing the gap left by MOSS-Image's lack of a dedicated trainer, this framework provides a complete path from a layered 6GB VRAM option to directly exporting LoRA in ComfyUI format.
LTX Desktop 1.3.1 has been released, adding IC-LoRAs AlphaGen green-screen-free, prompt-free matting that can output transparent WebM and matte MP4. This version also adds motion trajectory control with up to three paths, Pose/Depth/Canny guidance video, compositing with up to 10 reference images, and opens 1440p and 4K for text-to-video, image-to-video and audio-to-video, with 1440p up to 20 seconds and 4K up to 10 seconds.
AusBoss's ComfyUI creative project ausboss-nodes adds a Layout Boxes (Utility) tool that lets you draw boxes on the canvas and specify what goes in each box—a certain object, or text written into the image.
mlx-audio released v0.6.0, adding support for the Fish Audio Drama 3 speech model, Phonon-2 speech recognition and MiMo-V2.5-ASR STT, plus Granite Speech 4.1 Plus. This version improves Breeze-TTS and VibeVoice generation performance, and fixes Qwen3-ASR streaming timestamps, Qwen3-TTS prefill, and the fallback when Metal is unavailable.
ComeCut v0.2.1 released its first Windows desktop version, only about 30MB in size, with fast startup and low resource usage. It adds a multilingual interface, automatic dubbing (supporting Chinese, English, Korean and Japanese, requiring the corresponding TTS), and auto-save when exiting the editing editor, and fixes the issue of audio being lost after separation, while displaying vocals and background audio in different colors.
AI Canvas v0.10.11 improves macOS media resource usage and false triggers of scroll-wheel zoom, and can prompt to reopen a saved project after an abnormal page exit. "Settings → Appearance" adds interface layout and background color previews, allowing adjustment of the toolbar area position, opacity, and frosted-glass blur. The assistant supports uploading PNG / JPEG / WebP reference images, and the desktop Skill can import from ZIP, tar.gz, and tgz; this macOS version only provides an Apple Silicon installer.
Canvas Agent v1.6.8.12 supports importing ZIP animation packages containing avatar.json and multiple MP4s, can switch animations by running state, and includes built-in multi-state animations as the default avatar. Canvas appearance adds presets and separately remembers light and dark color schemes, while also improving batch tables, image tools, and viewport interaction (#815, #816). This version also fixes issues such as the completion status of generation tasks not being restored after re-entering the canvas (#813), and the upgrade involves no database schema migration.
The open-source image and video management tool PixlStash released 1.12, which can directly run ComfyUI workflows; right-clicking a character with a LoRA loads the LoRA by character and generates an image. The new version groups different checkpoints and character LoRAs of the same workflow into one workflow card, supports setting LoRA strength and trigger words for each of two people in the Run window, and allows opening the workflow from PixlStash for editing and saving as a new version with Ctrl+S.
ComfyUI Portable was updated to v0.42.1, fixing crashes some users experienced when running MOSS-Image template workflows. The community added two visual layout editors, MOSS-Layout-Editor and ComfyUI-MOSSLayout, and there appeared a LoRA, workflow, and img2img node for using MOSS-Image for line art coloring.
Backlot is an open-source single-window studio that runs on your own GPU without a node graph, supporting text-to-video, image-to-video, reference mode (keeping the same person or product consistent across shots), 360 orbit generation from a single photo, latent chaining for long scenes, and still-image features with identity editing, 96 camera angles, and SeedVR2 4K.
Why it matters: The open-source project brings multi-model video and image generation into a single-window interface and provides measured timings on a 16 GB GPU, making it easy for creators to assess local deployment feasibility.
The developer launched Yeufonic, a free open-source frontend for YuE2/ComfyUI, aiming to hide ComfyUI's complex node wiring so musicians can focus on content creation and management. It provides a song library management and song generation, and allows training your own LoRA with local materials, along with examples and downloadable LoRAs.
BeforeShare can check embedded metadata before sharing an image and save a cleaned copy; file contents are processed locally in the browser, free and with no account required. Its public test used two synthetic PNGs (containing ComfyUI-style prompt, seed, and workflow data) to verify: both workflow containers were removed, while the image stream and decoded pixels remained unchanged. The tool also supports removing some C2PA containers, but does not remove SynthID or visible watermarks.
Perplexity updated Computer on October 19, offering Nano Banana 2.1 image generation and editing to users with access to Computer, with follow-up instructions in the same conversation to modify.
ComfyUI v0.42.0 was released on October 19, officially supporting OpenMOSS's 5B image model MOSS-Image, and adding Qwen 3.x series performance improvements Part 2.
The GIMP 3 open-source plugin OpenFill can perform generative fill, object removal, outpainting, and text-to-image locally via ComfyUI, with no cloud upload needed. The AI sound effect tool earshot v0.3 adds 201 code-synthesized sound effect libraries and can run offline on CPU alone.
The author rounded up seven ComfyUI tools around MiniMax H3, including H3-LongTakeNoCuts, which suppresses quality degradation in video extension, and obvpm-timeline, which generates extensions and intermediate segments on a timeline.
Why it matters: The seven ComfyUI tools around MiniMax H3 each have their own role, covering long-shot quality preservation, timeline generation, and VRAM optimization, and can be chosen according to one's own workflow.
Dreamer 2.0.1 is released for Apple Silicon Macs, fixing an issue where, when Photoshop natively opens the main panel for the first time, the display event occurs earlier than the plugin lifecycle listener registration, causing the startup logic not to trigger; the Host will read the panel menu state a limited number of times to restore the first display.
Dreamer 2.0.1 releases a Windows x64 version; the plugin itself is completely free, login is optional, and users can directly generate images and perform local color correction, alignment, and light re-projection without a Token Plan. In this version, Auto resolution for asset generation prioritizes the model-supported 4K, and candidates are archived to "Generated Assets" for reuse. The EXE has not yet been Authenticode-signed, so Windows may show an unknown publisher or SmartScreen warning; users need to verify the SHA-256.
Runway announces that Runway MCP is now available on Grok Bot, allowing users to directly use its generation capabilities within Grok conversations without leaving the chat interface. The company provides an entry link for users to get started.
ComfyUI author comfyanonymous released MOSS_image_MAX_ULTRA_hires_lora_test on October 19, a LoHa adapter for OpenMOSS's 5B text-to-image model MOSS-Image, aimed at raising generation resolution to 4 MP and above.
Krea has launched a Blender plugin that generates material references, environment concepts, lighting tests and multi-camera views in the 3D Viewport sidebar. It requires Blender 5.2 or above and supports macOS and Windows.
Krea has released a DaVinci Resolve plugin that can generate lighting, tone and atmosphere references, as well as backgrounds, environments and insert shots within Resolve Studio, with generated results entering the project as ordinary media.
Krea has released the Photoshop plugin Krea for Photoshop, which provides background replacement, relighting, multi-camera layout, image and video generation, real-time canvas and product mockups within Photoshop documents, with results returned as ordinary layers.
Why it matters: Krea brings image and video generation into the Photoshop panel, with results returned as ordinary layers, making it convenient to evaluate integration approaches against existing retouching workflows.
ComfyUI-CharacterCardStudio 1.2.1 was released on October 20. It is a custom node package that can generate character reference setting cards and slice and lay them out according to templates. The input is an identity-anchoring reference image or a character description, and the output is a multi-panel card; each panel is cropped to a fixed 768x768 after generation and placed back in its original position.
Krea has released a Premiere Pro plugin that can generate shots, insert shots, backgrounds and style options within the editing software panel; results enter the project as ordinary media assets and can be trimmed, composited, color graded and mixed as usual.
Why it matters: Krea embeds generation capabilities into the Premiere timeline, with finished output delivered as ordinary assets that editors can process directly using their existing workflows.
mlx-audio releases v0.6.0, adding support for Fish Audio Drama 3, Phonon-2 speech recognition, MiMo-V2.5-ASR, and Granite Speech 4.1 Plus. It also fixes multiple issues with Qwen3-TTS/ASR, Whisper, and others, and optimizes Breeze-TTS generation and VibeVoice performance.
Flagship Scene UI is a custom node interface running within ComfyUI that organizes video workflows by scene and clip for MiniMax H3. It supports importing JSON plans containing prompts, reference assignments and generation settings, and can regenerate an individual clip without rebuilding the entire project.
Wireflow reproduced the core task of Higgsfield Ads Studio using its own node canvas: generating 3 static ads from one product image of a terracotta mug, consuming 39 credits in total with model time under 1 minute, and laying out headlines as real text layers.
The author released the H3 Recast workflow, which can replace up to 3 characters in an existing clip in a single run, preserving motion, scene and original audio, and adds a FaceRefine lip-sync repair step. The process is: load the clip, grab a frame and use Qwen-Image or GPT Image to change it into the new look as a scene reference, then write one line of description per person—fill in as many lines as the number of people to replace.
Why it matters: The author provides a ComfyUI workflow that replaces up to three people at a time, and lists VRAM, RAM and measured runtimes on three tiers of GPUs to help assess local deployment costs.
Workflow Studio released v0.8.8, supporting MOSS-Image with ComfyUI v0.42.0; the generation UI tab correspondence check in the workflow tab is shortened from tens of seconds of waiting to about 2 seconds in perceived time.
Krea 3 was released on October 15; this article checks the official inputs and outputs, the web usage guide and the API documentation, and sorts out which details are preserved in the two types of editing examples, swapping furniture and swapping backgrounds. The release page says most native 2K edits take under eight seconds, $0.06 is the 2K base item, and the API separately lists $0.02 per 1K reference image and $0.07 per 2K reference image, with high mode adding another $0.02 per round.
InvokeAI 7.0.0 alpha 3 has been released, adding role-based board sharing, support for the Qwen-Image-2.1 image model, and the SageAttention acceleration backend for NVIDIA GPUs. This version speeds up full-image generation at 2048² resolution on an RTX 4090 by 15–35% and video generation by 12–35%. Note that v7 performs a one-way upgrade of the v6 database, and the official recommendation is to install and test with a brand-new root directory.
The author has released the open-source project freavy, a free local-first node canvas running on ComfyUI for AI image and video generation, with assets and rendering done on your own GPU, requiring no subscription and uploading no content.