Skip to content
Date and source

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

Oct 23Fri
  1. Reddit StableDiffusion · 创作者与工作流分享18

    Reddit user questions the quality of the Flux 3 model and asks whether the BFL team has responded

    A Reddit user says Flux 3 performs terribly, still producing "body horror"-style generation errors after updating nodes, questions whether the BFL team has responded, and asks whether this is toying with the open-source community. The user also mentions that about a month ago, when trying the free demo, generating a cola can with a vague prompt produced amazing results, but the image has since been deleted.

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

    How to correct the age portrayal of the Kari Kamiya character LoRA to make an SFW version

    The user made an SFW LoRA for the Digimon character Kari Kamiya's first-season appearance and plans to upload it to Civitai, but the generated results look closer to her age in the second season. Solo image material for this character is limited, and tags must be added manually to restore the correct hair color and outfit. The author hopes to correct the age portrayal directly in the LoRA so that other characters can later be added for group shots.

  3. note · anpontan_note · 图像视频创作分享8

    When using AI to generate title images, the copy was limited to "within 12 characters": after lining up 19 pieces, the last 5 all hit the upper limit

    A person who uses image-generation AI to make title images and set a character limit of "within 12 characters" for the copy lined up 19 pieces of copy from 10/4 to 10/22 by character count and found that 5 consecutive pieces from 10/17 to 10/21 all reached 12 characters or more, with 10/20 at 13 characters, exceeding the limit by 1 character, on the grounds of keeping "にも" to stay consistent with the wording in the body text.

  4. Reddit comfyui · 创作与工作流分享17

    Using low-quality small images to generate new images for a dataset—which model is best? A ComfyUI user asks for recommendations

    A user who had stopped using image generation for years started using ComfyUI again and tried Qwen Edit 2.1 and FLUX 2 Krea. Qwen's results were acceptable, but they don't know how to prompt to generate multiple different images of the same character. Their use case is to upscale small images of mythological creatures and pixelated small face images found online into a dataset. Their hardware is four RTX 3090s with 96 GB VRAM in total, 64 GB of RAM and a relatively slow Threadripper CPU; they mainly run LLMs day to day and have limited experience.

  5. 机器之心SOTA模型 · 公开公众号 RSS38

    Canva releases the Craft image generation model; Shanghai AI Lab open-sources the ShuSheng-A1.5 research agent

    Canva released the Craft image generation model on October 22. It uses a two-stage process—composition planning first, then pixel rendering—supports 2K output, more than 30 aspect ratios, and in-image text in 55 languages, and is available through Canva Dream Lab and a public API.

  6. note · maynoha · 图像视频创作分享38

    Writing prompts in Japanese to describe actions: drawing an autumn after-school scene on PixAI Tsubaki.3

    PixAI's Tsubaki.3 supports writing prompts as natural Japanese prose, making it easier to reproduce specified actions than with tag formats. The author uses 'a girl turning around and waving on a ginkgo-lined avenue' as an example, specifying in prose that she holds a bag in her right hand, raises her left hand to shoulder height to wave, and is shown full-body including the soles of her shoes; after generation, a Japanese instruction alone changed the bow on her chest to deep red.

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

    MOSS LoRA: one-click generation of realistic Instagram-influencer-style photos

    A LoRA called MOSS has been released that directly produces Instagram-influencer lifestyle photos with warm tones, soft colors, and natural lighting, requiring no extra filters or retouching. The LoRA is compatible with Moss base, SFT, and Kromos, suits everyday scenarios such as selfies, casual phone snapshots, street fashion, cafés, and travel, and is recommended at a weight of 0.6-1.0 with no trigger word needed.

  8. MindStudio Blog · 创作分享42

    Recraft V4.1 Flash Released: ~1.3-Second Image Generation with Shareable Custom Styles

    Recraft V4.1 Flash leads with fast image generation at approximately 1.3 seconds per image, at the cost of first generating at a lower resolution, then optionally upscaling and optimizing quality through the enhance feature. Custom styles support building from reference images, built-in example prompt previews, and newly added sharing via link for others to apply. Recraft still offers a free plan, allowing users to try V4.1 Flash and style tools.

  9. note · tasty_cougar8018 · 图像视频创作分享22

    Tagtelier Launches Anima Prompt Illustration Collection: Copy and Export by Category

    Tagtelier's website freely publishes a collection of prompt illustrations generated with Anima. Each image comes with its prompt and generation settings (model, sampler, CFG, steps, size, seed). Prompts are broken down by categories such as quality, number of people, face, expression, clothing, accessories, pose, background, and composition, allowing users to check and copy items individually or exclude specific words.

  10. note · myloose · 图像视频创作分享15

    AI Portrait Generation Study: Monolid Male Lead in the Short Drama 'Welcome to Thailand🇹🇭'

    When using AI portrait generation for monolid male characters, the illustrated style often turns them into double-lidded eyes, so the author switched to a photorealistic style to continue creating. The author set the monolid male lead as older, with slightly droopy eyes and curly hair, shaping him into a "good guy to dangerous" love-stealing character. After generating a strong-featured handsome rival for the love triangle, the images take on a strong Thai BL drama feel.

  11. note · tsugiito_design · 图像视频创作分享38

    Generating a 'Lying Down, Looking Back at Camera' Character with Animagine XL 4.0 Opt in ComfyUI: Tags and JSON for Supine Pose, Spread Long Hair, and Satin Sheen

    The author used Animagine XL 4.0 Opt in ComfyUI to generate an anime character in a 'supine, head on pillow, long hair spread out, looking back at camera' pose, at 1152×896 resolution, Seed 202610171, Steps 28, CFG 5, without using LoRA or ControlNet, relying solely on tag combinations for composition.

  12. Reddit StableDiffusion · 创作者与工作流分享30

    Is Building Images and Scenes with Prompt Replacement a Better Approach Than Generating from Scratch?

    A user shared that using prompt replacement on Gemini yields more consistent results than generating from scratch: when adding a lower body and legs to a female image, the AI would redraw the entire person and cause errors, but replacing an existing female into a full-body image usually transfers features correctly, after which the lower body clothing can be swapped. Complex scenes work the same way—finding a target scene image first and then replacing the character saves much more time than creating from scratch.

  13. lilting channel (English) · 创作分享22

    Testing Anima-Base v1.0 Character LoRA on Seven 2B Anima Derivative Models: Only RDBT v2.3 Preserves Kei's Red Bow Tie in Both Full-Body and Half-Body Images

    Testing a character LoRA trained on Anima-Base v1.0 across seven 2B Anima derivative models in ComfyUI on M1 Max 64GB, with the same prompt and seed 42, only RDBT v2.3 preserved Kei's red bow tie in both full-body and half-body images, and drew two characters as four. Hotaru Blend and CapAnima generated images in about 38 seconds at 8 steps, while Base required about 260 seconds at 30 steps.

  14. lilting channel (English) · 创作分享30

    Testing Four Anima 2.9B Derivative Models on M1 Max: Character LoRA Remapped to 40 Blocks Preserves Hairstyles, Taking About 1.4x the Baseline Time

    Testing Nova Anime AM, CottonAnima, AMIXA, and JANIMA—four Anima 2.9B derivative models—on M1 Max 64GB using ComfyUI, remapping a character LoRA trained on 28-block Anima-Base v1.0 to 40 blocks, with 840 out of 948 keys successfully remapped, and all four models preserving character hairstyles.

  15. Reddit StableDiffusion · 创作者与工作流分享71

    PixText: a lightweight model and ComfyUI node for fixing text in generated images

    The author has released PixText, a model with fewer than 1B parameters (about 0.95B) for fixing garbled text and Chinese characters in generated images. The model is trained directly in pixel space, requires no text encoder or VAE, and uses MeanFlow to complete text repair with only 4 sampling steps. In ComfyUI, users can select a region and enter the correct text, or first automatically detect text and then adjust the suggested boxes; a comparison node shows the before-and-after effect for each region.

    Why it matters: To address the problem of small text in generated images being blurred by VAE compression, the author provides a pixel-space repair model and a ComfyUI box-selection text-editing workflow.

  16. ArchiGen AI · 图像生成与ComfyUI实战20

    Can free AI rendering solutions be used for client projects? ArchiGen breaks down the licensing and delivery boundaries behind "free"

    ArchiGen points out that "free" in free AI rendering solutions only indicates price and does not mean the account type, project use and commercial output are licensed. Buyers need to check eligibility terms, commercial-use terms, output rights, quota reset and export deletion rules, and keep dated copies of the terms. Under free quotas, previews, failed tasks, variants and upscaling may share the same metering unit, and the resolution, watermarks and metadata of exported files also need to be confirmed before delivery.

  17. Reddit comfyui · 创作与工作流分享12

    How to composite different people into the same image in ComfyUI while preserving facial identity

    A user asks for help in the ComfyUI community: how to put different people into the same image without losing their respective facial identities, and have them pose and do other actions together. The user has tried krea 2 and flux Klein 9 distilled models with a basic image editing workflow, with hardware of an AMD 6570xt, 12GB VRAM and 32GB system memory, and hopes someone can recommend a feasible workflow.

  18. Promptsref.com · 创作分享25

    This week's three hottest Midjourney SREFs: retro Japanese anime, pink-gold dreamy fashion, and classical hand-drawn prints

    Promptsref has announced this week's three hottest Midjourney SREF styles: retro Japanese animation illustration, pink-gold dreamy surreal fashion, and classical engraving-style hand-colored prints. The three correspond respectively to late-night OVA nostalgia, commercial fashion editorial, and museum print texture, and can all be quickly tested via --sref and reused for series creation.

  19. Reddit comfyui · 创作与工作流分享38

    MOSS-Image news roundup: Edit LoRA merged into ComfyUI, CharRef v2, and a 500-prompt library

    Kijai's MOSS-Image Edit LoRA support has been merged into the latest ComfyUI Portable, giving MOSS-Image Edit model capabilities and ControlNet-like basic functions. Supporting releases also include the CharRef v2 character reference node, a 9.4Gb Dirty MOSS realistic fine-tune, an animation style LoRA, and a library of 500 usable prompts with example images.

  20. note · h_kitamura · 图像视频创作分享22

    Image/Video Generation AI News, October 23, 2026 issue: Midjourney has the highest copyright risk, Imizu City fire brigade AI ad, Runway integrates external tools, Helix World generates a walkable scene from a single image

    After comparing about 15 image generation AIs, VN.ai found that Midjourney had the highest proportion of generated images at risk of infringement, about 37%, and that it had improved little about a year and a half after being sued by two film companies. NTV HR Research Institute used generative AI to produce 3 recruitment commercials in different styles for the Imizu City fire brigade in Toyama Prefecture, including a 32-second horizontal live-action style, a 37-second vertical animation, and a 35-second horizontal live-action.

  21. note · palealloy · 图像视频创作分享30

    ComfyUI custom node: click to select lighting terms to strengthen the presence of illustrations

    To address the problem that ComfyUI users cannot remember the effects of lighting prompts, the author created a custom node where clicking lets you select lighting expression terms such as backlight, spotlight, God Rays, particles, low-key, and specular reflection, and hovering the mouse also displays text explanations. In the example, after overlaying these terms on an armored character, the image changes from flat to a heavy, solemn atmosphere. The author also notes that newer-generation models such as Krea2 already have fairly high inherent quality, but this node can be used to precisely hit the desired lighting effect.

  22. Reddit comfyui · 创作与工作流分享22

    Krea2 edit node loses detail: user seeks a way to preserve source image details 1:1

    A user reports that although Krea2 is not an editing model, operating through the edit node causes the output to lose a large amount of detail and exhibit pixel drift. They have tried various sampler/scheduler combinations, 8/10/12 steps, secondary upscaling, Krea2 enhance LoRA, raw/turbo models and the Wan2.1 VAE, all without noticeable improvement, and ask whether they can only wait for the local version of Krea3.

  23. Reddit StableDiffusion · 创作者与工作流分享20

    Seeking help with MOSS-style LoRA training: 150 images, 30 epochs, and loss stuck at 0.44

    A beginner trained a MOSS-style LoRA with FizGig's default settings, carefully selecting and strictly captioning 150 images and setting 30 epochs, but by epoch 18 the loss could not go below 0.44 and kept rising back between 0.48 and 0.56, so they stopped training and asked the community whether to continue or adjust parameters.

  24. ComfyUI Wiki · 工作流与开源创作分享72

    MOSS-Image's first instruction-editing LoRA released, with native support merged into ComfyUI the same day

    On October 22, 2026, MOSS-Image got its first instruction-editing LoRA, chinoll/Moss-Image-Edit, a rank-64 adapter that takes 1 to 4 reference images and a text instruction and covers changes to color, clothing, lighting, season, and visual style.

    Why it matters: The community adds an instruction-editing LoRA for MOSS-Image; the article gives two integration paths—ComfyUI native and a node pack—along with default parameters, which can be evaluated against existing editing workflows.

  25. Reddit StableDiffusion · 创作者与工作流分享58

    Creator shares Flux3 style transfer workflow with adjustable composition and style strength

    Reddit user Winter_unmuted shared a one-shot, adjustable image-to-image style transfer workflow based on Flux3, centered on a reference latent plus node. The author says composition strength (the later the start, the weaker and more creative) and the strength of the image style relative to the prompt (the earlier, the more image style; the later, the more prompt-dependent) can be adjusted separately, and includes gradient examples.

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

    User shares images generated with the MOSS Image 5B Base model without LoRA or fine-tuning

    User Unit2209 shared images generated with the MOSS Image 5B Base model, without LoRA or fine-tuning; most images used the default settings of 50 steps, CFG 3 and 1.5 megapixels, and the Base model rather than the SFT model. The image files include metadata extracted from the DeepBeepMeeps UI, but no detailed notes.

  27. Monster社区 · 中文ComfyUI实测与工作流20

    Comprehensive review of Fibo V2 image generation and editing: compared with Moss Image and Qwen Image 2.1

    A comprehensive review of the image generation and editing capabilities of Fibo V2, billed as "the world's number one," has been released, with a side-by-side comparison against Moss Image and Qwen Image 2.1. The review covers two types of tasks, image generation and image editing; the specific scores and conclusions require consulting the original hands-on results.