Skip to content
Date and source

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

Nov 3Tue
  1. Reddit StableDiffusion · 创作者与工作流分享12

    How to best combine pose, identity, clothing, and background with Krea 2 in ComfyUI?

    A ComfyUI user built a workflow using Krea 2 Turbo, identity LoRA, clothing reference, background reference, and DWPose/OpenPose LoRA, hoping to preserve exact face, clothing, and room while following a completely different pose reference, but current pose consistency is unstable, and clothing and background details also vary. The user asks whether to use a two-pass workflow (first establish pose and composition, then refine identity, clothing, and background) or whether there is a more reliable single-pass conditioning method for Krea 2.

  2. Reddit StableDiffusion · 创作者与工作流分享8

    What is the current best open-source img2img solution for facial identity preservation? Is Qwen Image 2.1 still the king?

    Reddit users compare open-source img2img facial identity preservation solutions: dedicated face-swap LoRA (such as BFS) and Qwen Image 2.1's native capabilities, as well as zero-shot adapters like PuLID, InstantID, and ComfyUI attention mask inpainting, in terms of their performance in preventing facial drift.

Nov 2Mon
  1. 电影界 · 短剧经营与内容趋势37

    Micro-drama forum opens in Beijing on November 5, Pinduoduo enters core view for the first time, micro-dramas transform from content products to super connectors.

    On November 5, the Micro-Drama Forum of the 2026 China Radio and Television Excellent Works Creation Conference will open at the Beijing Conference Center, with Pinduoduo entering the forum's core view for the first time. The forum is themed "Refinement, Popularization, Internationalization" and is hosted by the Network Audiovisual Department of the National Radio and Television Administration and the Beijing Municipal Radio and Television Bureau, with Hongguo Short Drama and others as co-organizers. Micro-dramas are transforming from content products into "super connectors" that connect consumption and activate industries.

  2. Reddit StableDiffusion · 创作者与工作流分享19

    The Best Way to Create Clear Videos with MiniMax H3

    A Reddit user shares their experience of making clear videos with MiniMax H3: using DMD, DMAD, step600 ema, and Pika Flex, especially liking Pika Flex's realism of characters, but the 2mp output still has noise and ghosting. The user asks if anyone has tried combining 20~50 high step counts with high-quality turbo LoRA (such as Pika Flex), and says generation time is not an issue.

  3. Reddit StableDiffusion · 创作者与工作流分享16

    What is the current state of video generation on the AMD 9070 XT?

    AMD 9070 XT user's real-world test shows MiniMax H3 video generation speed is terrible: generating a 5-second 480p video takes 8 minutes. The user says image generation has been running stably through models such as zIT, MOSS, and krea2, but video generation still awaits an acceleration solution, and asks whether there is a model suitable for AMD graphics cards.

  4. Reddit comfyui · 创作与工作流分享3

    ComfyUI Model Selection Help: LTX 2.3 Underperforms at Low Resolution

    A ComfyUI user, limited by system performance, can only generate with LTX 2.3 at low resolution, resulting in poor quality. The user prefers image-to-video and plans to produce a 15-20 minute YouTube series. Their device is an ASUS TUF Gaming A15, and they are seeking recommendations for models and configuration.

Nov 1Sun
  1. Reddit StableDiffusion · 创作者与工作流分享12

    Qwen Edit 2.1 character reference image placement is not effective, ComfyUI storyboard production help request

    The user tried to use Qwen Edit 2.1 in ComfyUI to place character reference images at specified positions and poses, but the results were poor—the characters appeared as if directly pasted into the scene, with faces often facing the camera. The user plans to create storyboards for a short film and needs to place multiple characters in different positions, so they are asking what model should be used.

Oct 31Sat
Oct 30Fri
  1. Reddit comfyui · 创作与工作流分享15

    H3 Character Replacement

    The user attempted to perform full-body character replacement using H3, but was unable to find a workflow that preserves the original motion. Other models also have issues: scail2 has eye defects, wan22 has limb regression, and ltx 2.3 has poor quality. The user is seeking advice on precise motion transfer based on motion capture rather than motion control LoRA.

  2. Reddit comfyui · 创作与工作流分享8

    RTX 3090 user seeking i2v workflow suggestions

    A 3D visual artist using an RTX 3090 seeks image-to-video (i2v) workflow suggestions for architectural visualization (Archviz) production, requiring only camera push/pull (dolly) without distorting or blurring structures and faces in the frame. The user wants the workflow to run properly on their PC.

  3. Reddit comfyui · 创作与工作流分享9

    LTX 2.3 does not work on ComfyUI Desktop.

    Reddit users tried to use LTX 2.3 on ComfyUI Desktop but it didn't work; even after reinstalling, it still didn't work. Image Z Turbo loads fine, but almost all other models fail to load.

  4. Reddit StableDiffusion · 创作者与工作流分享12

    How to create a high-quality dataset from limited reference images to train a LoRA?

    A user posted asking for help on how to create a high-quality LoRA training dataset from limited reference images, while maintaining facial identity, body consistency, and high detail levels. The dataset generation workflow they tried tends to reproduce the original facial angles, and the output has an obvious AI-generated look. The user asks whether face swapping is the best method, or if there are better alternatives.

Oct 29Thu
  1. Reddit comfyui · 创作与工作流分享10

    A LoRA trained on Krea 2 has skin texture issues—how can this be resolved?

    A ComfyUI user reports skin texture issues with a LoRA trained on Krea 2; testing all step counts produced the same result. The LoRA was trained on ostris for 3000 steps, and images were generated with Sick Ollie; another LoRA also trained on Krea 2 for 1000 steps showed similar issues. The poster says they are unsure of the cause and is seeking help.

Oct 28Wed
  1. Reddit comfyui · 创作与工作流分享11

    Wan2.1 First Experience: Generated First Clip but Results Fell Short of Expectations

    The user generated their first clip on Wan2.1, but the results fell short of expectations, mainly due to a lack of knowledge in maintaining image-to-video clarity and fidelity. The user is currently trying the Reactor face swap workflow and is curious about LoRAs, hoping to create realistic models and increase the realism of personal photos. They also want to add family members to clips, but find it difficult to preserve the original image appearance.

Oct 27Tue
  1. Reddit comfyui · 创作与工作流分享10

    How to choose the best workflow for RTX 5050?

    A user with an RTX 5050 (8GB VRAM, 16GB system memory) asks about the best I2V/T2V workflow in ComfyUI, seeking a good balance between performance and quality. The user mentions that Wan 2.2 5B can generate decent videos, but hopes to avoid downloading many incompatible dependencies and simplify the process.

Oct 26Mon
  1. note · lofty_eagle5366 · 图像视频创作分享8

    nijijourney and Midjourney illustration creation notes: purple-toned gender-neutral characters and the "Dawn" series

    The author used nijijourney and Midjourney to generate illustrations, attempting to depict gender-neutral characters in purple tones without specifying gender, and formed a recent creative theme around the "Dawn" prompt. Among the 4 images generated in a batch, one completely ignored the tag prompt and was an accidental product; the author also mentioned that 2D hair strand gloss is relatively difficult to render, and the presence or absence of gloss affects the sense of three-dimensionality.

  2. Reddit StableDiffusion · 创作者与工作流分享12

    Can you get started with AI image generation on a low-spec laptop? An Acer Swift 5 and Nitro 5 user asks for help

    A user who has been following image generation since 2023 asks whether it is possible to run customizable image generation models locally on low-spec laptops from before 2022 (Acer Swift 5 SF514-55TA and Acer Nitro 5 AN515-53). The user says they cannot afford large-capacity storage or high-performance graphics cards, and is also not inclined to use online or credit-based platforms.

  3. Reddit StableDiffusion · 创作者与工作流分享10

    Which AI Model Should I Choose for Local Video Generation with an RTX 5070? A Beginner Seeking Recommendations

    A beginner on Reddit asked which AI model is suitable for local video generation, hoping to avoid paying for AI tools, produce videos with a self-built local environment, and publish them to social platforms to earn extra income. Their hardware is an RTX 5070, 12GB VRAM, and 32GB RAM; they prioritize beginner-friendly models and are also willing to spend time learning.

  4. IT之家18

    IT Morning Brief 1026: Apple iPhone Duo's inner-screen gaming black bars draw complaints; SparkLink 3.0 standard results released; Zhang Xue sets Guinness record for motorcycle water skimming; Xiaomi Wuhan employee housing Guanggu Mengxiangjia opens for sale

    Running Honor of Kings and Genshin Impact on the Apple iPhone Duo's inner screen is not yet supported in full screen, and the large black bars have drawn complaints from netizens; two days after going on sale, the phone's price fell below its launch price in some regions, with Pinduoduo's 10-billion-yuan subsidy making it about 200-300 yuan cheaper than the launch price. The SparkLink 3.0 standard results were officially released at CEIC2026 and the International SparkLink Alliance Industry Summit, targeting all-scenario intelligent IoT and aiming at directions such as AI, passive, and millimeter wave. Zhang Xue rode a modified motorcycle, the MX450, to complete a 40.74-kilometer water ride in Mayang, Hunan, setting a Guinness World Record.

  5. Reddit StableDiffusion · 创作者与工作流分享12

    Recommendations Requested: AI Super-Resolution Models or ComfyUI Workflows for Realistic Portraits Without Artifacts

    A user on Reddit is asking for recommendations of AI super-resolution models or ComfyUI workflows suitable for realistic portraits, requiring preservation of original facial features and identity, enhancement of details such as skin texture and hair strands without producing an artificial look, and compatibility with both low-resolution and higher-quality photos. The user notes that some super-resolution tools hallucinate details, over-sharpen skin, or slightly alter facial features, and hopes for clean, realistic results with minimal artifacts. Both traditional and diffusion model methods are acceptable, with quality prioritized over speed.

Oct 25Sun
  1. Reddit StableDiffusion · 创作者与工作流分享10

    Reddit user asks: what model is used to generate this kind of character image?

    Reddit user Vosquad97 posted asking what model the creator used to generate this kind of character image, speculating it is an editing-type model, possibly edited from game screenshots or official images. The user believes the character details are quite accurate, judges it unlikely to be a closed-source model, and explicitly rules out GPT because of its poor performance in multi-angle character generation.

  2. Reddit comfyui · 创作与工作流分享15

    Looking for recommendations on SOTA speech-to-speech translation models that preserve the speaker's voice characteristics

    A user posted in the ComfyUI community on Reddit, looking for the current SOTA models capable of speech-to-speech translation (e.g., Spanish to English or German) while preserving the speaker's voice characteristics, preferably open-source. The use case is translating Spanish TV series into other languages to share with family and friends, since professional dubbing versions are not always available. They mentioned that some tools they've found significantly alter personal voice characteristics such as speech rate and pitch, and they hope the translation process can maintain the original voice, even if it comes with a slight accent.

  3. Reddit StableDiffusion · 创作者与工作流分享22

    How can you get sharper, more detailed results with MOSS-Image on Mage.space?

    A user reports that images generated with MOSS-Image on Mage.space look soft, blurry, and lacking in detail, and that the overall image quality reminds them of older models such as Lustify SDXL. They ask whether a specific prompt structure, settings, or workflow is needed, and whether the sharper, more realistic results seen in the community come directly from MOSS-Image or rely on additional tools or post-processing.

  4. Reddit comfyui · 创作与工作流分享18

    How to precisely control facial expressions when using One Piece LoRA in ComfyUI

    A user asked for help in the ComfyUI community: when using One Piece LoRA to generate images, the facial expression is always wrong; they want a simple smile with raised lower eyelids, but the result has an overly large open mouth and uncontrolled eyelids. The user asked whether there are specific prompts or other methods to control facial expressions more precisely.

  5. Reddit StableDiffusion · 创作者与工作流分享22

    Which applications are worth learning for producing longer AI video content? A ComfyUI user seeks a workflow management solution

    A user who manually chains together image generation, multiple video versions, upscaling, and then editing in ComfyUI already finds the process cumbersome when producing a 30-second video, and asks whether any video project tool can manage relationships between assets and support revising already-generated content without reloading each one in ComfyUI. The user currently stores audio, first frames, raw generations, and upscaled files only by directory, has previously used only basic editing tools such as ffmpeg and AviDemux, and is unfamiliar with how video generation applications integrate with ComfyUI.

  6. Reddit comfyui · 创作与工作流分享20

    How can ordinary GenAI users get started with ComfyUI? A discussion on the necessity of moving from cloud tools to local workflows

    A practitioner working in AI film and advertising currently uses Nanobanana and GPT to generate storyboard frames and uses Omni, Kling, and others to generate video as needed, and is considering moving to ComfyUI for local image/video generation. They find the ComfyUI interface rather complex but more professional, and ask whether it can truly help their work, whether this shift is necessary, and where to start learning.

Oct 24Sat
  1. note · dreamy_quail9288 · 图像视频创作分享12

    AI animated short "Fallen Angel" Episode 5 "Tracker" released: a Halloween 2026 work, the Black Angel held back for Episode 6

    The AI animated short series "Fallen Angel" updates with Episode 5 "Tracker," a Halloween 2026-themed work that the author says leans warm and healing in style. The author states that the Black Angel does not appear in this episode and will formally appear in Episode 6, and plans to finish the remaining production before Halloween. A related article on the production workflow adjustments will be published the next day.