ComfyUI workflow templates v0.11.86 released: adds FLUX 3 Klein distilled T2I baseline workflow
ComfyUI workflow templates released v0.11.86, installable via pip install comfyui-workflow-templates==0.11.86.
ComfyUI workflow templates released v0.11.86, installable via pip install comfyui-workflow-templates==0.11.86.
A creator asked for help on Reddit, hoping to build a reusable workflow with ComfyUI and RunPod to batch-generate roughly 15-second 9:16 vertical short videos, always featuring the same photorealistic arborist character, changing only the scene, tree species, dialogue, and camera movement.
The Alibaba TaoLive AIGC team released TaoMate-H3-FL2VA on October 15, a 3-step LoRA that adds first-frame and last-frame control to MiniMax H3's streaming TaoMate-H3 runtime; the team calls it the first known real-time streaming first-and-last-frame-to-audio-video (FL2VA) model.
Why it matters: TaoMate-H3-FL2VA integrates first-and-last-frame control into streaming audio-video generation, and a ComfyUI conversion version is already available to run locally, making it easy for creators to evaluate whether to incorporate it into their existing production pipelines.
A ComfyUI user running the Krea2 4bit quantized model via multigpu nodes in a dual-GPU configuration (RTX 3060 + RTX 2060 Super) keeps encountering the error "CUDA error: an illegal memory access was encountered," which interrupts generation.
Claude itself cannot generate video, but it can call video models via MCP and return a finished MP4. In a four-node pipeline built on Wireflow, Nano Banana Lite generates keyframes, Gemini Omni Flash 1.1 and Kling O3 each produce 5-second clips, and Video Editor composites a 1920x1080 finished film. The whole process takes 101 seconds and consumes 176 credits.
Using Wireflow's node-based canvas to build an AI video workflow, a 9.3-second, 1080x1920, 30fps two-character fruit short drama was produced at a total cost of 164 credits, taking 6 minutes 50 seconds from first run to finished MP4.
Conradlocke released the second-generation Krea 2 Edit LoRA on October 16. The biggest change is structural: it is split from a single LoRA into two weights each with its own role: Krea 2 Identity Edit V2 is responsible for transferring the person in a single photo to a new scene, pose, clothing or lighting, while Krea 2 General Edit V2 handles instruction-based edits such as object removal, color changes, relighting, background replacement, object addition, rewriting sign text and stylization.
Why it matters: Krea 2 Edit LoRA is split into two weights, identity and general, and provides tested parameters for reference image placement and mixing the two LoRAs.
OpenMOSS released the 5B text-to-image model family MOSS-Image on October 16, including two checkpoints: the pretrained version MOSS-Image-Base and the general-user-oriented MOSS-Image-SFT.
Why it matters: The 5B text-to-image model focuses on Chinese and English text rendering and JSON layout control, with same-day support from ComfyUI core nodes, which can be used to evaluate layout-oriented image generation workflows.
Creating a faceless YouTube channel requires choosing a specific audience, planning original videos, and telling the content through voiceover, screen recording, animation or stock footage rather than appearing on camera. The article uses a fictional channel Clear Video Lab and its first episode "Why your tutorial feels rushed" as an example to demonstrate an animated faceless format, and gives specific steps for channel positioning, topic validation and visual identity setup.
Plenio Music Production System 1.0 was released on October 16, turning ComfyUI into a local music studio: a language model writes lyrics, and YuE2 or MiniMax Music 3 sings.
A generative AI workflow is a creative chain that connects multiple model nodes with links, where each node does only one thing: hold a prompt, run a model, or pass a file.
Comfy offers a $25,000 prize on the Comfy Developer Platform, calling for open-source rebuilds of the Higgsfield creative app over two weeks, requiring public repositories that anyone can run and fork.
F3K's default ComfyUI/Diffusers templates lack the prompt upscaler they depend on, resulting in poor generation quality. This upscaler is a fine-tuned version of Qwen3.5-9B, responsible for converting natural language into the specific style prompts required by the model, and can be run in ComfyUI through an LLM node. Users should pay attention to CFG settings (about 1 for the distilled version, about 3.5 for the base version) and use reasonable resolutions to generate images.
Higgsfield's official blog rounded up 9 AI tools for motion designers in 2026, covering Higgsfield AI, Adobe Firefly, Runway, Kling AI, Adobe After Effects, Magnific, Luma, Pika and Krea.
User CeFurkan shared a ComfyUI preset for FLUX 3 Klein on the comfyui subreddit, along with a link where the changelog can be viewed.
Dreamer 2.0.0 has been released for Apple Silicon Macs; the plugin itself is completely free, login is optional, and no activation code or device binding is required. The new version adds a Retouch Workshop and a global sidebar, and model thinking settings are displayed and saved per specific model, shared between Chat and image viewing. This version is a pre-release package; Photoshop reinstallation, panels, and paid model acceptance testing have not been completed, and Intel Macs have not been verified.
ComfyUI-TagForge removed the bundled pmp/prmpt.yaml and its fallback list entries because rndm_ponyxl was unused and contained invalid wildcard references. Deployed copies will be automatically cleaned up via Garage boot-sync.
The author released Model Bending, a ComfyUI technique that intervenes in internal activations midway through diffusion model sampling; with the same prompt, seed, and model, changing only the bending method of a certain layer yields different images.
A recent update to the Linux version of ComfyUI requires PyTorch 2.7, which prevents 10-series Nvidia GPUs from running, because the last PyTorch version compatible with 10-series GPUs is 2.6.
The author of AntiMatter Comfy released a free English-language MiniMax H3 / ComfyUI learning handbook, including a 32-page PDF, a Markdown version with copyable prompts, three Comfy-Org starter JSON files (T2V, I2V, R2V), model links, and troubleshooting notes, with no paid membership required.
A user tested anime and natural-light still-life styles in ComfyUI with FLUX.3 Image Klein (Base, NVFP4, about 5.3 GB); the model comes from black-forest-labs/flux-3-image-klein-comfy on Hugging Face.
User neptune765 figured out in ComfyUI how to use Qwen2.1 to improve detail in T2I workflows, finding that it is still not perfect after upscaling but already quite good for text-to-image. The next plan is to process in Photoshop and then further enhance with I2I.
Krea 3 has launched on ComfyUI via Partner Nodes; it is a new image model from Krea aimed at reference-image-driven generation and editing. The model focuses on faster iteration and lower per-image cost while maintaining image quality, and supports uploading multiple reference images that are called by upload order number, for example taking the person from image 1, the jacket from image 2, and the background from image 3.
Why it matters: Krea 3, which connects reference-image-driven generation and editing to ComfyUI Partner Nodes, lets you learn how to call multiple references by number and feed results back.
The author organized 23 MiniMax H3 custom nodes into six categories: long video and continuous generation, performance and editing, image quality correction, audio and dialogue, reference assets and character motion control, and assets and production support, with links to each repository's README and workflows.
A user compared the image editing results of Flux 3 and Krea 2 with a community Edit lora, using the prompt "make her lie on the floor," and the results showed that Flux 3 performed poorly. The test was based on the same source image: Flux 3 generated the second image, and Krea 2 with Edit lora generated the third image.
The author compared the bf16 base and distilled versions of Flux 3 Klein on an RTX A4500 20GB using the official ComfyUI workflow, with 50 prompts, the same seed, and about 2MP.
Valtz offers a workflow on Mac for generating AI videos from stick-figure animation: draw a simple stick figure on the canvas, insert Position and Zoom keyframes in the Crop and Position panel, use the joystick to move the Markup Layer and record displacement, then select Linear Interpolation on the first keyframe in the timeline, and Valtz automatically interpolates frames to generate smooth animation.
A user using Flux 3 Klein as an editing model in Comfy found results far worse than F2 Klein, questioning whether it is a model implementation issue or a problem with the model itself. Although the official benchmark shows F3K far surpassing F2K, the user's test prompts included having a man sit on a sofa in a modern minimalist living room and obtaining a right-side view of the scene, and neither produced satisfactory results.
Black Forest Labs releases the 5.6B-parameter open-weight model FLUX 3 Image Klein, supporting editing with up to 10 reference images and bounding box layout control, and it can already be invoked in ComfyUI via templates.
Maxon releases its 2026 fall ecosystem update, upgrading Cinema 4D, ZBrush, Redshift, Red Giant, and Autograph across the board to 2027.0.
The creator used ComfyUI with LTX-2.5 and MiniMax H3 to complete the image-to-video experimental short film ARCHIVE ZERO, which tells the story of an institutional archive gradually becoming impossible.
A new ComfyUI user running Krea 2 Turbo on an AMD Radeon RX 7600 (8GB VRAM) experiences a first-step wait of 6 to 7 minutes or even longer before sampling begins, but once sampling iterations start, the speed is fairly fast.
The author releases H3-Anti-Burn LoRA to mitigate the problem of footage becoming progressively over-sharpened and burned when H3 clips are chained into longer videos. The LoRA is loaded only on continuation clips, with an initial weight of 0.5. In a chained test of about 37 seconds, facial texture change for lightx2v 8-step dropped from +44% to 0%, HyperFlow from +25% to −1%, and Larry Turbo from +261% to −5%.
Why it matters: To address over-sharpening and burning after chaining H3 clips, the author provides a loadable LoRA with specific weight recommendations, along with multiple sets of before-and-after comparison data.
A user asked on Reddit how to remove text and text boxes from videos with Minimax H3. They mentioned that qwen edit 2.1 can be used to remove text frame by frame and then stitch the frames together, but they want to know the specific workflow for Minimax H3.
The author reproduces the common TikTok character-swap videos in ComfyUI using Minimax, replacing real-person dances with animals or other characters. The approach is to load MiniMax-H3-Character-Swap-LoRA on the default ref2va template workflow and first use the default Qwen 2.1 edit template to generate a single-frame guidance result.
ComfyUI releases v0.11.85, with the version number bump completed by @comfyui-wiki in #1335. This version was not published to PyPI because the PR lacks a release label; the label needs to be added and the workflow re-triggered for publication.
Ome Omy Productions has made the CUT version, previously part of the $5 Filmmaker tier, free. The author describes it as a local MiniMax H3 / ComfyUI timeline editor that can generate shots, preserve different takes, edit and trim footage, extend scenes, patch discontinuities, and arrange results on the timeline. It also supports saving projects, comparing takes, and exporting finished videos.
A user tested Z Image Turbo (bf16) with a custom character LoRA at 8 steps, res_multistep / simple, CFG 1, shift 3, and the same seed (710101), changing only the film-style line at the end of the prompt (Cinestill 50D / 35mm / Portra 400) to compare which produces a more realistic golden hour atmosphere. The author also asks if there are any prompt tips they might have missed.
A user encountered errors at the ksampler step when using the gguf quantized version of Qwen Image 2.1 for image editing in ComfyUI, regardless of which gguf quantization combination or gguf node branch they tried.
A user attempted to replicate the video character replacement effect of the Higgsfield Genjutsu model in ComfyUI, testing dozens of models including SCAIL_2_REPLACEMENT. Motion replication was perfect, but the output character's face differed noticeably from the source image. Adding the ReActor node still yielded poor results, and the user is seeking community help.