First try at portrait generation on Z-Image: running Z-Image Turbo locally on a Mac Studio
A user ran Z-Image Turbo locally via ComfyUI on a Mac Studio 64GB for pure text-to-image, and the portrait results were satisfying.
A user ran Z-Image Turbo locally via ComfyUI on a Mac Studio 64GB for pure text-to-image, and the portrait results were satisfying.
A user conducted a comparison test on Moss Image 2MP: compared with 50 steps euler + sgm_uniform, the 8-step turbo LoRA (weight 0.8) was 10 times faster.
A user tried the ComfyUI 0.42.1 portable version for the first time, chose the Wan motion transfer template, and after installing all models, running it failed immediately: either it disconnected directly with no error, or it failed at the sampler stage, and the same happened with the regular version. The device is a Legion Pro 7 with 32GB RAM and a 5090 24GB VRAM, the NVIDIA driver has been updated, and the user is unsure whether they should run the 3 update bat files.
DramaFlow Studio has released version v0.1.15, with the full changelog covering v0.1.14 to v0.1.15. The original text does not disclose specific feature changes, parameters, or availability details.
ComfyUI workflow templates has released v0.11.91, installable via pip install comfyui-workflow-templates==0.11.91. This version marks the new partner node templates as local-only.
HiDream.ai co-founder and COO Wang Ke said during the Yunqi Conference that AI video generation is moving from "usable" to "easy to use," and that over the next 1–2 years the core of competition will shift from model capability contests to workflow closed-loop capability. Its video model HiDream-O1-Video-1.0 adds a front-end intent understanding module and adopts a self-developed native omni-modal UiT architecture; the film and television creation collaboration agent Zhenzan supports script analysis, storyboard generation, and character design, and a 5-person team completed 130 minutes of content for the 24-episode short drama "Yanmo's Beloved Concubine" in 15 days.
SunoMV sampled 1,531 storyboard projects and 65,853 shots and found that without reason constraints, AI defaults back to medium shots (32.6%), eye-level angles (50.1%), and push-in shots (28.4%). Its solution is to write one song-related camera-movement reason for each line of lyrics, then derive shot size, angle, and camera movement from the reason, and first lock characters and locations with character cards and scene cards. The next version of SunoMV will have the AI attach one reason to each shot when drafting storyboards, displayed on director mode and mobile storyboard cards.
A user on Reddit is asking for experience training Fizgig MOSS image LoRA: they tried 30–70 epochs, 0.25–0.5 MP, 20–40 images, alpha rank 16, learning rate 0.0001, and added block 22 to the ComfyUI workflow.
A user shared a ComfyUI node configuration for accelerating Moss image generation on an RTX 3060 12GB: enable the node's ultra fast setting, pair it with the latest 8-step LoRA and set the weight to 0.2-0.4, set the sampler to 50 steps, CFG 4, Res_multistep, Simple, 2mp. The LoRA is used to compensate for the skipped sampling steps.
The tutorial explains how to install the MiniMax H3 Director plugin in ComfyUI, drag shots, reference images, and audio onto the timeline, and have the plugin compile in real time the storyboard prompt that H3 reads during training.
Why it matters: The tutorial presents the installation of the H3 timeline plugin, model directories, and storyboard compilation as reproducible steps, and lists prerequisites and restrictions such as the training interval and commercial licensing.
Kapwing released a tutorial demonstrating using GPT-6 Astra to connect with Blender to build a dragon and scene's gray model, animation and camera shots, then using GPT Image 2 to generate design drawings and a photorealistic first frame, and finally Seedance 2.5 to generate a 10-second continuous shot.
Why it matters: The author tested using GPT-6 Astra to drive Blender to build a 3D scene, then went through image-to-3D and Seedance 2.5 to produce the video, and listed a comparison of three 3D models and itemized costs.
Kapwing released a tutorial introducing the use of Claude skills to generate camera instruction prompts, then using Seedance 2.5 in Kai to reshoot a single fixed-camera piece of footage into a multi-camera edit.
Why it matters: The author tested single-camera to multi-camera conversion with Seedance 2.5, providing the complete prompt and failure cases, which can be transferred to multi-camera editing for short dramas.
ComfyUI workflow templates released v0.11.90, installable via pip install comfyui-workflow-templates==0.11.90.
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.
TorchNeuron provides a native backend for PyTorch on AWS Trainium, allowing HF libraries such as Transformers and Diffusers to run models directly.
Stable Diffusion Art released a complete ComfyUI workflow tutorial for MiniMax H3 reference-to-video (ref2va), which can combine up to 9 reference images, 3 reference videos and 3 audio clips to generate a 24fps multi-shot video with synchronized stereo dialogue, sound effects and music in one pass.
Why it matters: The tutorial provides the complete node configuration and prompt structure for MiniMax H3 reference-to-video in ComfyUI, which can be transferred to multi-shot character consistency workflows.
ScreenWeaver's co-founder published an item-by-item comparison of ScreenWeaver and LTX Studio, with the conclusion: starting from a concept, client brief or image, and needing to deliver an ad or edited film this week.
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.
lodestones released Kromoss on October 19, a fully fine-tuned model of MOSS-Image-Base aimed mainly at the Base model's blurriness issue. It uses the Qwen3.5-9B text encoder and FLUX.2 small VAE, and can be dropped directly into existing MOSS-Image workflows.
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.
Luma launched Luma Variants on October 1, 2026, which can automatically adapt one approved static ad into five sizes—Story 9:16, Portrait 4:5, Square 1:1, Medium Rectangle 6:5 and Widescreen 16:9—and translate the copy, but the company has not announced pricing.
Topaz Labs provides an official Image API and Video API, called with an X-API-Key and billed by credits: Gigapixel costs 1 credit per 24 MP output, a 10-second 1080p Proteus video costs 4 credits, and credits are priced at $0.08 to $0.12 each.
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 author trained a style LoRA "SepiaBlue Luminous" for the local image generation model MOSS-Image and released it publicly on Civitai, with the trigger word qsv2style. Training used 100 ChatGPT-generated images sharing only the style, with different characters and scenes, based on MOSS-Image-Base, at 1024 resolution, rank 16, learning rate 1e-4, for 1000 steps, and the 400-step version was ultimately chosen.
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.
The user is using the MiniMax H3 video extender (backported) for video extension; after the first generation, during upscaling in the second generation, the sample custom advanced node inside the model's latent upscale subgraph reports an error. The poster asks about the cause and hopes for suggestions on alternative nodes.
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.
A ComfyUI desktop newcomer wants to load a self-downloaded LoRA locally in a Wan image-to-video workflow, but after replacing the LoRA parameters the results are abnormal, and manually adjusting after unpacking the subgraph reports "not enough memory", while it runs normally at low resolution in the packed state. Their goal is a simple IMAGE → PROMPT → RESULTS flow that reproduces the model page's example results.
A user reports that ComfyUI startup has slowed due to installing a large number of custom nodes over the past few months and wants to know which nodes can be deleted. They found a cleanup solution that looks promising and ask whether anyone has used it, whether it works, and whether there is a faster and simpler way, or simply removing everything and reinstalling one by one based on workflow errors.
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.
OpenMOSS released MOSS-Image, a 5B DiT model using the Qwen3.5-9B text encoder and FLUX.2 VAE, available in Base and SFT versions under the Apache 2.0 license.
Why it matters: MOSS-Image is open-sourced under Apache 2.0 and supports structured layout control, allowing evaluation of text rendering capability against existing open-source image models.
The author released an Evangelion fan concept proof of concept, placing an AI-rendered Sachiel and a Redshift-rendered CG Unit-01 in the same shot to achieve physical contact, with the contact point maintained 1:1 frame by frame during complex camera movement, with no drift.
Why it matters: The author uses live-action-grade compositing of AI characters and CG characters aligned frame by frame to demonstrate a concrete approach for integrating AI layers and depth channels into a traditional VFX pipeline.
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.
After initial testing, a user considers the MOSS model the strongest inpainting model they have tried; the workflow has embedded example images, which can be downloaded via a Civitai link if missing.
Students can get one year of Google AI Pro for free and use A100 for up to 35 hours per month for free on Google Colab. The accompanying ComfyUI Notebook automatically switches to the corresponding model based on the T4 or A100 provided by Colab Pro.