Skip to content
Date and source

Dates follow the news timeline in Beijing time and include both endpoints.

Nov 15Sun
  1. ComfyUI Wiki · 工作流与开源创作分享72

    ComfyUI adds a Video Deflicker node to fix flicker from per-frame processing

    ComfyUI core adds a Video Deflicker node, merged via PR #17754, which can fix flicker after per-frame upsampling and other processing. The node requires downloading RAFT-large weights to ComfyUI/models/optical_flow/, using a stable frame as guide to read motion, preserving real changes. The same PR replaces torchvision raft_large with a native implementation, about 2.3x faster.

    Why it matters: The new node provides a fix for flicker after per-frame upsampling and other processing, using a stable frame as guide to preserve real changes, allowing creators to optimize existing workflows accordingly.

  2. SuSu_酥酥73

    MiniMax H3 Image quantized version reduces VRAM from 68GB to 6.4GB; 8GB graphics cards can run 4K image generation.

    The open-source community has quantized MiniMax H3 Image to Q2_K, reducing VRAM requirements from 68GB to 6.4GB, allowing 8GB graphics cards to run 4K image generation locally. The quantized version natively outputs 4096×4096 and 21:9 widescreen, official distillation produces images in 6 steps, integrates generation and editing, and supports up to 16 reference images.

    Why it matters: The quantized version reduces VRAM requirements from 68GB to 6.4GB, allowing 8GB graphics card users to run 4K image generation locally, which can be used to evaluate whether to replace existing image generation workflows.

  3. Monster社区 · 中文ComfyUI实测与工作流54

    MiniMax H3 and MOSS Image Generation: Model Comparison and Accelerated LoRA Trade-offs

    Author T8star-Aix compares MiniMax H3 and MOSS image generation models, showcasing ComfyUI workflows and accelerated LoRA trade-offs. The article provides a checking sequence: establish a baseline with the same prompt, distinguish style from structure, then confirm whether the acceleration gains are worth the detail loss. It also notes that 6 steps, 15 steps, and 8 steps cannot be blindly copied without considering the model.

  4. note · akb428 · 图像视频创作分享29

    Create beautiful girl character parties with Anima and Pixubus 4: usable for novels, event CG, and animation

    Pixubus 4 and Anima can one-click generate a party of 1-5 beautiful girl characters, supports Japanese prompts, and no English tags are required. The party can be imported into Inkubus v4.5 for novel writing and output as EPUB, or via Vidubus 1.5 generate a 15-second animation with sound. Anima was developed by CircleStone Labs and Comfy Org, trained on millions of anime images and approximately 800,000 art images.

  5. Reddit StableDiffusion · 创作者与工作流分享62

    Styles Atlas releases nodes for ComfyUI, providing 1000+ styles and 800+ characters.

    Developer releases Styles Atlas, three ComfyUI nodes convert Excel tables into prompts, providing about 1,600 rendering styles and 827 known characters. Nodes can be chained, support random, wildcards, and hover thumbnails; data is stored in .xlsx files and can be edited by users. The project has been published to GitHub: https://github.com/atlasimp/Style_Atlas

  6. MiniMax H3 Novel Pipeline · ComfyUI创作项目发布36

    MiniMax H3 Novel Pipeline updated to 2.7.1: enhances cinematic chapter generation and diagnostics

    MiniMax H3 Novel Pipeline updated to 2.7.1, enhancing cinematic chapter generation, H3 input preparation, and diagnostic capabilities. Adds structured temporal estimation, retry grouping, and diagnostic sidecar files, with the default adapter token budget set to 16384. The project is released as a ComfyUI creation project, verified through offline integration tests and simulated regression coverage.

Nov 14Sat
  1. Reddit StableDiffusion · 创作者与工作流分享54

    Powder Nodes 2.1.0: Explore Prompts, Styles, and LoRA, and Refine Images

    Powder Nodes 2.1.0 is released, a free and open-source ComfyUI toolkit that separates prompts, styles, and LoRA for exploration and supports image refinement. It supports exploring variants using prompt lists, reusable styles, and LoRA combinations, comparing results in an image grid, and refining with Canvas painting, masks, and layers, providing 14 ready-made workflows. GitHub: https://github.com/E2GO/e2go-comfyui-nodes

  2. ComfyUI-HotSwitches · ComfyUI创作项目发布30

    ComfyUI-HotSwitches fixes the issue where model unload state is lost after clearing the cache.

    ComfyUI-HotSwitches fixes the issue in version 0.3.0 where the switcher forgot the last model after ComfyUI cleared the cache. After the fix, switching models still only unloads the previous model; running the same model again retains the weights, and this record is independent of the output cache. The fix has been verified by actual generation tests with BF16 and NVFP4 Qwen, and a restart of ComfyUI is required after updating for it to take effect.

  3. ComfyUI-HotSwitches · ComfyUI创作项目发布46

    ComfyUI-HotSwitches update: Hot Lazy Switch adds model unload option and fixes label reset

    ComfyUI-HotSwitches update: Hot Lazy Switch adds a default-enabled "Unload previous model" option, which automatically unloads the previous model when switching models, while shared text encoders, VAEs, and unrelated models are unaffected. It also fixes a bug where the last custom label would reset to its default name after refreshing the page or reopening the workflow. After updating, you need to restart ComfyUI and hard refresh the page (Ctrl+F5).

  4. Monster社区 · 中文ComfyUI实测与工作流39

    Seedream 5.5: ComfyUI Image Generation, Reference Editing, and Layering Tutorial

    Seedream 5.5 can follow three routes in ComfyUI: text-to-image generation, reference-image editing, and poster layering. The demonstration outputs 720×1280 PNG and emphasizes first distinguishing t2i and i2i tasks. Reference editing requires checking the face, outline, and pose, while poster layering requires checking text, edges, and inpainted areas.

  5. Reddit comfyui · 创作与工作流分享67

    Making a 10-minute dialogue-free animated film in ComfyUI using MiniMax H3, LTX-2.5, and Wan 2.2; the author shares problems encountered and solutions.

    The author used MiniMax H3, LTX-2.5, and Wan 2.2 in ComfyUI to complete a 9:51 dialogue-free animated short film 'DAY 731', with most shots generated by H3.

    Why it matters: The author publicly shared parameters and pitfalls-to-avoid experience for making a long-form film on a 3060 12GB, including specific issues such as resolution, sampler, and VAE redraw.

  6. Reddit comfyui · 创作与工作流分享61

    Hands-on test and sharing of sync.so's ComfyUI lip-sync workflow

    The author tested sync.so's ComfyUI lip-sync workflow and shared installation steps, model prices, and duration limits. Two usage methods: built-in nodes and custom nodes. lipsync-2 is about $0.05/second, free quota is 3 times per month for 20 seconds. In the test, only the speaking actor's mouth was reanimated; the rest of the footage and the reference dialogue were unaffected.

  7. Reddit StableDiffusion · 创作者与工作流分享18

    Has anyone tried building a one-click AI short film with ComfyUI?

    A Reddit user attempted to build a one-click AI short film workflow with ComfyUI, using Qwen 2.1 for image generation and face swapping, MiniMax H3 for video generation, and LTX 2.5 for video upscaling. The workflow runs normally, but the output is still 'AI SLOP'. The author believes the problem lies in the prompts, and used Claude AI to create nodes that automate the three workflows into one click.

Nov 13Fri
  1. Reddit StableDiffusion · 创作者与工作流分享62

    RTX 5070 Ti ComfyUI benchmark: overclocking reduces generation time by about 10%, tuned parameters cut H3 video time by more than half

    The author tested the effects of GPU overclocking, ComfyUI startup parameters, and connecting the display to the iGPU on generation speed on an RTX 5070 Ti. Overclocking consistently reduced generation time by about 9-12%, and tuned parameters cut H3 video time by more than half.

  2. Reddit StableDiffusion · 创作者与工作流分享15

    How to maintain facial consistency in MiniMax H3 video during dynamic camera motion?

    When users generate dynamic camera motion videos with MiniMax H3 video, the face and eyes exhibit slight flickering/jitter. The configuration is a single-chain exact latent continuation in ComfyUI at 1344×768 @ 24 fps, 192 frames (8 seconds), using prompt + start_image + 2 reference images (Qwen3-VL encoder), sampler res_multistep 32 steps, cfg 1, fixed seed, but the jitter still persists.

  3. note · miyuu_2026 · 图像视频创作分享47

    11/13 [Wan2.2] Fashion Hoodie Edition

    The author used ChatGPT to generate reference images, Wan2.2 14B to generate video, and Capcut to edit, producing an 18-second AI short drama. In ComfyUI, Wan2.2 14B generated 73 frames at 1024x576px resolution, consisting of six 3-second shots. The production environment was Windows 11 Home, Ryzen 7 7700, RTX 5060 Ti 8GB, 32GB RAM.

  4. Reddit comfyui · 创作与工作流分享53

    Finally, a way to increase batch size in MiniMax H3/Pruna H3 image and video workflows

    The author used Claude Code to develop two custom nodes for MiniMax H3 image and video workflows, supporting generation of 2 or more different images or videos in a single run, equivalent to increasing batch size. The nodes are MIT-licensed and have been published on the ComfyUI Registry; the GitHub page provides more information. Previously, MiniMax H3 could not increase batch size like Flux.2 Klein 9b, so the author developed these nodes.

  5. note · enigmind · 图像视频创作分享18

    MiniMAX H3 suddenly fails to run: causes and countermeasures

    When running ComfyUI - MiniMAX H3 i2v batches on an RTX 2080 Ti, an error occurred. The cause was Windows 11's Smart APP Control blocking system files. After disabling that feature, it returned to normal. In a no-prompt test, Kling AI 2.5 Turbo's operations were more detailed, and MiniMAX H3's focus movement was comparable.

  6. Reddit StableDiffusion · 创作者与工作流分享46

    ComfyUI-FileExplorer

    The ComfyUI-FileExplorer node made by ChatGPT provides an experience similar to Windows File Explorer, supporting multiple outputs and image, video, audio, and .txt files. It can load multiple files from the same directory (such as multiple reference images for image editing). Compared to native load/upload nodes, it loads files directly from the full path, avoiding rapid expansion of the input folder.

  7. Reddit StableDiffusion · 创作者与工作流分享63

    WakeGPU released: free open-source desktop app, remotely runs ComfyUI/Colab GPU and supports automatic idle shutdown and batch queuing

    WakeGPU released, a free open-source macOS/Windows desktop app for remotely running ComfyUI/Colab GPU, with support for automatic idle shutdown and batch queuing. The tool is MIT-licensed, with no paywall, subscription, or telemetry, and binaries are published on GitHub Releases.

  8. Reddit comfyui · 创作与工作流分享64

    Share a ComfyUI workflow: using MiniMax Fun Control to create 360-degree orbit and looping animations

    The author shared a ComfyUI workflow that uses MiniMax Fun Control to composite a 360-degree character orbit with a 360-degree environment, and supports seamless looping animations. The process includes character pose refinement, 4-angle lighting, a 5-frame orbit video, environment 360-degree slice compositing, and using the composite video's depth as the control video and the environment video as the reference video to drive MiniMax Fun Control.

  9. Reddit StableDiffusion · 创作者与工作流分享52

    H3 Image Character Sheet Builder

    The author shared an automated character sheet generation workflow based on H3 Image, which can generate a character from different angles at once while maintaining similarity and quality. The workflow supports different character types such as vehicles and quadruped creatures, and has been submitted to Civitai.

  10. comfyorg · YouTube 创作视频68

    LTX's VFX nodes are now available in ComfyUI via Partner Nodes.

    LTX's VFX nodes are now available in ComfyUI via Partner Nodes, allowing post-processing to be completed without leaving the node graph. Supports recomposition (1:1, 4:5, 5:4, 9:16, 16:9), upscaling to 4K/8K, SDR to 10-bit HDR (BT.2020/HLG), and per-frame alpha generation up to 4K.

    Why it matters: LTX VFX nodes allow post-processing to be completed within the ComfyUI graph, enabling creators to evaluate whether to replace their existing export-then-post-process workflow.

Nov 12Thu
  1. Reddit comfyui · 创作与工作流分享7

    What are the best settings in ComfyUI for producing a roughly 3-minute, 1080p+ short film from live-action footage?

    The user asks whether there is currently an available ComfyUI workflow that can convert live-action footage into a roughly 3-minute, 1080p+ sci-fi short film, rather than rendering everything from scratch. He mentions that when he used Ebsynth in the early days, the pipeline was not yet mature, and hopes for recommendations for the best settings, with storyboard and cross-scene style consistency capabilities.

  2. Reddit comfyui · 创作与工作流分享50

    Action Director standalone version released, supporting 3D posing, camera composition, and sending settings back to ComfyUI

    Action Director has expanded from a ComfyUI node into a standalone application, providing a dedicated 3D viewport for character posing, camera setup, and scene composition. It sends settings directly back to ComfyUI through a live link and can continue to be used alongside existing workflows. The author says it is still being refined and provides GitHub and Microsoft Store entry points.

  3. Reddit comfyui · 创作与工作流分享43

    ComfyUI-Cleaner — Find and remove models no longer needed

    ComfyUI-Cleaner 1.0.0-beta released, this is a Python tool coded with AI assistance, for finding and removing models no longer needed. It can identify models not referenced by saved workflows, duplicate models in different locations, and models stored in unexpected folders, and checks potential download sources on Hugging Face and Civitai. Before deletion, suggested changes need to be reviewed; the script supports dry-run mode and operation logs; however, some seemingly unused models may still be needed by custom nodes or templates, so careful review is needed before deletion.

  4. 蓝鲸新闻 · 搜狐公开专栏77

    160 million views earned 390,000 yuan: AI short dramas' exorbitant profits all went to cloud providers

    160 million views only brought in 390,000 yuan in revenue sharing, while cloud computing trial-and-error costs often range from tens of thousands to hundreds of thousands of yuan, making the pure cloud-based AI short drama business model untenable. The article presents a layered hybrid workflow—cloud-based scriptwriting, local image generation to lock character designs, low-resolution trial-and-error, and post-production super-resolution—which reduces the comprehensive cost per shot from over 20 yuan to a few jiao.

    Why it matters: This layered hybrid workflow reduces the trial-and-error cost per shot from over 20 yuan to a few jiao, providing a transferable path for cost reduction in AI short dramas.

  5. Reddit StableDiffusion · 创作者与工作流分享57

    MiniMax H3 Image Free Simple Workflow Sharing: Realistic 1Girl Text-to-Image / Image-to-Image

    Reddit user Cold_Pudding5326 shared a simple workflow for MiniMax H3 images, supporting text-to-image and image-to-image. The author says the model has outstanding facial consistency and can produce stable portraits without LoRA, but prompt adherence is currently poor. The workflow has been uploaded to Civitai.

  6. Reddit comfyui · 创作与工作流分享56

    Camera drift stems from restarting each time; link the last frame of the previous shot to maintain continuity.

    The author proposes using the actual last frame of the previous shot as the first frame of the next shot to maintain shot continuity, and gives three rules: change only one element at a time, reset to the master reference every three to four shots, and keep a link log so that only broken links need to be redone. Linking sacrifices the freedom to reimagine each shot, and bad frames contaminate subsequent shots.

  7. Reddit comfyui · 创作与工作流分享54

    WASasquatch releases ComfyUI Vast.AI template with Open Manager

    WASasquatch releases a ComfyUI template for Vast.AI with Open Manager built in. After launching an instance, users can directly load workflow templates within ComfyUI and download required models from the 'Not Downloaded' tab in the model library; the desktop version also provides in-app file management tools, without leaving ComfyUI or using the command line.

  8. ComfyUI Wiki · 工作流与开源创作分享64

    MiniMax-H3-Image integrates pDMD 2-NFE LoRA, completing sampling in 2 steps

    akhaliq merged the pDMD 2-NFE LoRA into the Comfy-Org checkpoint, reducing MiniMax-H3-Image sampling from 6 steps to 2 steps; a single 4096x4096 image on H200 went from 25.1 seconds to 17.4 seconds. The LoRA has been pre-merged into the 40GB weight file, or the 2GB LoRA can be loaded separately into an existing checkpoint; 2 steps work for most prompts, but graphic prompts are recommended at 3 steps.

Nov 11Wed
  1. Reddit comfyui · 创作与工作流分享39

    ComfyUI's Open-H3-Regenerate-2K node update adds IC support

    ComfyUI's Open-H3-Regenerate-2K node update adds NVIDIA IC model support; in testing, details are clearer and facial shadows are closer to the source image. The update includes a standalone IC loader and video refinement node, adjustable IC/LMS Sharp strength, and three/four-step sampling; it requires a newer ComfyUI and Ref2VA model, and a one-time preparation script reads about 26 GB of network data and saves a 185 MiB timestep file.