Skip to content
Date and source

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

TodayNov 19Thu1 item
  1. note · tidaa · 图像视频创作分享31

    Midjourney prompt generator ver.6.2 released, dual-person mode greatly evolved, and facial component settings added.

    Midjourney prompt generator ver.6.2 released, dual-person mode changed to Person 1/Person 2 tabs, allowing separate settings for character, hairstyle, clothing, etc. Added 8 facial component settings (contour, eyes, eyebrows, nose, lips, etc.) and a one-click swap button. Mobile version Mobile v6.2 updated simultaneously, tool runs completely locally (HTML/JavaScript).

Nov 12Thu
  1. Luma News42

    Best AI Character Generator: Maintaining Character Consistency Across Images and Videos

    AI character generators maintain character consistency through reference images, custom model training, and editing workflows. Among them, Luma AI's Uni-1 maintains image character consistency, and Ray 3.2 achieves frame-level identity stability in videos. Reference image workflows can reduce repeated generation, and editing workflows better preserve approved features than full regeneration. A unified platform reduces tool-switching friction and is suitable for cross-scene projects such as brand campaigns, animation, and illustration.

Nov 9Mon
  1. note · rosy_dog6649 · 图像视频创作分享38

    Why do AI images change with the same prompt? The role of seed in Midjourney and PixAI and how to save settings

    The seed in Midjourney and PixAI determines the starting random noise for generation; when not specified, each generation will change. With the same seed, if the prompt, model, or settings differ, the results will also differ. Midjourney uses --seed 12345 to specify; PixAI's Tsubaki.2, Tsubaki.1, and Haruka v2 support seed input, while Reference Pro does not. When comparing, only change one variable at a time, and save a complete settings memo.

Nov 8Sun
  1. Luma News18

    Midjourney's Limitations: What It Cannot Do

    Midjourney has notable limitations in brand consistency, text rendering, enterprise integration, and professional video production. It lacks a public API, generates images publicly by default, and standard access requires a paid subscription. Text rendering may produce inconsistent characters, character consistency is difficult to maintain, and video output lacks cinematic-level controls such as multiple keyframes and professional export formats.

Nov 6Fri
  1. TelosCompass · note AI视频创作实践60

    AI Illustration Ad Video: First Settle on a Character, Then Animate It

    When making an AI illustration ad video, first settle on a character illustration, then animate it across scenes, to avoid the character's appearance changing from scene to scene. The author uses a 15–30 second vertical coffee shop ad as an example: use ChatGPT or Midjourney to generate the character illustration, Kling AI to animate it, Hedra to lip-sync, Fish Audio for voiceover, and check five elements—face shape, hairstyle, clothing, art style, and background—then compare scenes side by side and redo only inconsistent ones.

Nov 3Tue
  1. Curious Refuge 创作教程43

    2027 Best AI Image Generators (Official Rankings)

    Curious Refuge announced the 2027 Best AI Image Generators rankings, with Google's Nano Banana 2.1 topping the list with a total score of 9.02/10. GPT Image 2.5 followed closely with 9.00/10, and had the highest prompt-following score (9.86/10). The evaluation was based on 29 standardized edge-case prompts, scored across three dimensions: prompt following, visual fidelity, and style and realism.

Oct 26Mon
Oct 24Sat
  1. note · freelife_creator · 图像视频创作分享8

    Earning 50,000 yen a month selling Midjourney prompts on note: the shortest path to monetizing knowledge

    A creator compiled Midjourney prompts into collections and sold them on note, Gumroad, and PromptBase, achieving a stable monthly income of over 50,000 yen. On note, the "Anime Character Generation Prompt Collection" is priced at 1,000 yen and sells an average of 150 sets per month; the English version on Gumroad sells about 100 copies per month at 15 USD each; and PromptBase sells about 150 copies per month at 5 USD each.

Oct 23Fri
  1. 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.

  2. 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.

  3. note · tidaa · 图像视频创作分享38

    Midjourney prompt generator ver.6.1 released: camera settings return, new prompt format toggle added

    Midjourney prompt generator ver.6.1 released, restoring detailed camera settings and adding a prompt format toggle. Users can separately specify camera model, lens, ISO, aperture, and shutter speed, or use preset mode to select a shooting style with one click; output format can be switched between natural language and structured lists. Pose presets increased from 8 to 12, adding four new scenes: Cozy Reading, Resort Holiday, Phone Break, and Smart Look.

Oct 22Thu
  1. note · kawakijourney_ai · 图像视频创作分享30

    Reproducing the process of an office lady going to a hotel after work to change into Halloween cosplay using Midjourney/ChatGPT/Gemini (with prompts)

    A Japanese creator has released a set of prompts for reproducing the full process of "an office lady walking into a hotel after work and changing into Halloween cosplay" using Midjourney, ChatGPT, Gemini, and other AI generation tools, broken down from STEP1 to STEP5: walking out of the office building wearing an employee badge, walking on a night street, taking off shoes in the hotel entrance, and cosplay poses on the sofa and bed.

  2. note · akisuke0925 · 图像视频创作分享38

    Midjourney Office Hours: new Niji nearing release, V9 in development, progress on Thinking Mode and new Upscaler

    In the latest Office Hours, Midjourney revealed that the new Niji model is nearly complete and may be made public in about a week at the earliest, but details of new features have not yet been announced. V9 is being developed with the goals of improving overall image consistency and prompt understanding; the initial model has entered testing, with no release date set. The new Upscaler is being tested and may launch before V9 if quality meets standards; Thinking Mode involves a trade-off between accuracy and aesthetics, and the official team is considering enabling it on demand or having the model decide automatically.

Oct 21Wed
  1. What’s New in Generative Media18

    Today's generative media highlights: celebrity AI face-swapping, Kling workflows, MOSS-Image, and other new developments

    10/21/26 generative media news roundup: The Verge reports on the proliferation of celebrity AI face-swapping, Variety says the sci-fi thriller "Overstake" received a generative AI reconstruction and was picked up by Cappu Films at the Busan market. Deadline focuses on how Asian directors are integrating Kling AI into their workflows, and openmoss.ai releases MOSS-Image, a text-to-image model supporting native layout control.

Oct 20Tue
  1. note · aun_585 · 图像视频创作分享58

    Full storyboard prompts for the Seedance 2.5 first-person animated short "Dinner After Overtime" released

    The author has released all the video prompts for the first-person animated short "Dinner After Overtime" made with Seedance 2.5. The process is to first use Midjourney to generate the first and last images of each shot, then use Seedance 2.5's keyframe feature to animate between the two images.

  2. note · freelife_creator · 图像视频创作分享38

    How to solve "all AI portraits look the same": 5 ways to establish a unique visual style

    In response to the common problem that images generated by tools such as Midjourney and DALL-E 3 have a strong "AI flavor," a personal visual style can be established through 5 prompting methods: mixing two or more styles (such as ukiyo-e × steampunk), using feature descriptions such as "swirling brushstrokes" instead of artist names to avoid copyright risks, locking in a 3-color brand palette, specifying textures such as fabric/paper, and fixing 3-4 composition patterns.

  3. What’s New in Generative Media17

    Generative media weekend roundup: Midjourney copyright controversy, YouTube Music labeling AI songs, FTC subpoenas OpenAI and Anthropic

    Midjourney was named in a report as the AI model most seriously violating Hollywood copyright, and YouTube Music has begun labeling AI-generated songs through partner disclosures and its own detection tools. The FTC subpoenaed OpenAI and Anthropic over consumer harm issues, while the New York City Comptroller urged Anthropic to pause its IPO over AI safety concerns.

Oct 19Mon
  1. note · rosy_dog6649 · 图像视频创作分享22

    What to do when an AI portrait is blurry after upscaling: size checks and high-resolution steps

    If an AI image is blurry after upscaling, first check the source image's pixel dimensions, display position and save method rather than regenerating directly. In Midjourney's current Upscalers guide, a V8.2 square SD image is 1024×1024, and after Subtle or Creative upscaling it is 2048×2048; HD images are already 2048×2048 and currently cannot be upscaled further.

Oct 18Sun
  1. note · kawakijourney_ai · 图像视频创作分享30

    Drawing 3D female characters with Midjourney: 5 types of fantasy character design sheets and standing illustration prompts released for free

    Using Midjourney to generate 3D fantasy female characters for games, covering 5 types: cute mage, adult magus, witch, saint, and battle mage, each with two English prompts—one for a character design sheet and one for a standing illustration—10 in total, with Japanese translations and design notes. The design sheets use a horizontal 16:9 layout showing front, side, back, and angled full-body views plus expression variations; the standing illustrations use a vertical 9:16 single-character centered full-body composition. Both share the same appearance description to keep the character consistent.

  2. 诀.com · 创作开源项目拆解22

    Midjourney beginner's guide: the full workflow from cultivating taste and building a P to generation and editing

    This Midjourney tutorial uses a natural-light female portrait as an example to demonstrate the complete web workflow from scratch to final image: first browse works in Explore to cultivate taste, then select images in Personalize to build a switchable aesthetic profile (P), then write prompts and parameters in Create to generate images, pick a base image, and use Vary or Edit to change clothing, repaint with masks, and extend the canvas to add blank space.

  3. note · freelife_creator · 图像视频创作分享22

    10 basic tips for generating the "ideal image" with AI: Midjourney and DALL-E 3 prompt practice

    A creator summarized 10 prompt tips for generating ideal images with Midjourney and DALL-E 3, including specifying the subject concretely, clarifying the style, specifying aspect ratios such as 16:9, refining lighting and color, controlling camera position and composition, using negative prompts, fixing the seed value, adjusting parameters such as --quality, and unifying style with reference images. The author says fixing the seed value can reduce generation time by 40%, specifying the aspect ratio shortens the time from generation to delivery by 30%, and negative prompts reduce the probability of abnormal outputs to 1/10.

Oct 17Sat
  1. Fstoppers18

    With 5 billion photos taken every day, what makes you a photographer?

    Faced with about 5 billion photos taken each day and the low barrier to tools like Midjourney, Max Kent argues that the joy of photography lies in the process of shooting itself, whereas prompt-generated images are almost solely about the result. He calls prompt users "curators" and argues that one need not define photographer identity through professionalization or selling work. His method: pick three or four words from photos you keep revisiting, take them out shooting, and thereby escape "routine" framing.

  2. note · uuhai0625 · 图像视频创作分享22

    Record of making picture book volumes 4 and 5 with Midjourney: multiple separate characters were hard to get in one image, so single-character generation followed by collage became the approach

    When making picture book volumes 4 and 5 with Midjourney, getting one image to depict multiple separate characters at once did not succeed within what was tried, so generating characters individually and then collaging became the main method. The approach of having Midjourney redraw the collage was tried on 4 spreads, and only 2 were adopted; volume 5 used no redraws. The character reference feature Omni Reference can hold only 1 reference image.

Oct 16Fri
  1. note · aicreatorpathnet · 图像视频创作分享62

    Using Midjourney 8.2's Edit Model to unify 10 AI character designs

    The author created 10 characters (9 animals plus 1 human) with Midjourney 8.2 for a YouTube AI comedy show, and summarized the core approach to unifying their appearance: use the Edit Model with the exact same two reference images attached each time (a full-body three-view and a face close-up), and state in the prompt what role each reference image plays.

  2. note · kawakijourney_ai · 图像视频创作分享22

    The complete method for earning 1.2 million yen with YouTube "GROOVE-style BGM" made using AI tools such as Midjourney, ChatGPT and Suno

    Vintage Park, a US YouTube channel registered in December 2025, has accumulated 2,517,047 views, 88 videos and about 14,900 subscribers in roughly 10 months with AI-made GROOVE-style BGM in soul, funk, jazz and hip hop styles, with estimated cumulative ad revenue of about 760,000 to 1.26 million yen at 0.3 to 0.5 yen per view.

  3. Promptsref.com · 创作分享15

    Midjourney Top 3 SREF of the Week: Healing-Style 3D, Ink-Wash Grotesque, and Cool-Tone Fashion

    Promptsref announced this week's top three Midjourney SREF styles: first place is a dual-code hybrid healing-style 3D style that combines Pixar-like character charm with children's picture-book colors and toy-grade 3D texture, suited to mascots and parent-child visuals; second place is a single-code ink-wash grotesque style that blends Chinese brushstrokes, rice-paper texture, and exaggerated contemporary character design; third place is a single-code cool-tone fashion style that uses hard flash and high-fashion styling to create a magazine-editorial feel.

Oct 15Thu
  1. note · genkidesu2021 · 图像视频创作分享18

    How to Write Image Generation AI Prompts: 3 Major Elements, Specific Examples, and Tool Recommendations for DALL-E 3 and Midjourney

    This tutorial breaks image generation AI prompts into three major elements—subject, composition/angle, and art style—and gives specific rewriting examples such as "a field of fully bloomed large yellow sunflowers" and "a black cat curled up asleep in the sunlight." For tools, it recommends DALL-E 3 (which can be given instructions directly in Japanese in the paid version of ChatGPT and supports local edits) and Midjourney (runs on Discord, high image quality, English input recommended). The author also reminds readers not to pile on too much information, to choose according to what each tool is good at, and to fine-tune repeatedly rather than expecting a one-shot result.

Oct 14Wed
  1. Frank Houbre · AI电影制作实战31

    The 2030 Film Studio: AI Filmmaker Frank Houbre Walks Through It Room by Room

    Using his own Outerframe Studio as an example, AI filmmaker Frank Houbre walks through how film studios will change in 2030, room by room. He judges that physical soundstages will still exist by then, but will be smaller and rely more on rentals, with shooting focused on performance capture and environments left to generation; the new SAG-AFTRA contract (July 2026 to June 2030) and the transparency obligations of Article 50 of the AI Act have already locked in some of the rules.

  2. note · aiby_masuhiyo · 图像视频创作分享22

    Generate AI illustration prompts with the "Illustration Parts Gacha" tool: hands-on tests with ChatGPT and Midjourney

    The "Illustration Parts Gacha" tool developed by Akari no Tabibito combines scenes, expressions, poses, clothing and small events into image-generation AI prompts, and supports pinning favorite items. The author generated two sets of prompts with the tool and produced images in ChatGPT and Midjourney respectively; the Midjourney version required fine-tuning the English prompts, while the ChatGPT version was used as-is.

  3. note · shiny_lilac7900 · 图像视频创作分享22

    Midjourney shared folders: images remain public even when set to private

    Midjourney launched shared folders in its October 7 update, and the official announcement explicitly warns: setting a folder to "invited only" only locks the folder entrance; images within are still displayed in public lists and personal profiles according to their original public settings. The feature is currently only available on the experimental website, and sharing settings do not change the public scope of the images themselves; hiding images still requires Stealth mode on higher-tier plans.

  4. note · hearty_knot4636 · 图像视频创作分享18

    No filming, made with AI: frontline thoughts during the transitional period of video production

    A video production professional observed that more and more companies are starting to build their own video planning and production in-house, and a demand for "no filming, AI-generated" has emerged; when budgets are limited, this has become a viable option. He himself has not yet formally used video-generation AI at work, but points out that AI is already used for voiceover dubbing, interview transcription, and static asset generation, and mentions the changes in production methods brought by generated imagery from Midjourney, Kling AI, and others. He believes generation capability and delivered finished products still need to be distinguished, and that directorial judgment, prompt trial and error, and confirmation steps will not disappear.

Oct 13Tue
  1. note · freelife_creator · 图像视频创作分享30

    Capability comparison table of five AI image generation tools: Midjourney, DALL-E, Stable Diffusion, Firefly, and Ideogram

    A comparison table outlines the differences among five AI image generation tools—Midjourney, DALL-E 3, Stable Diffusion, Adobe Firefly, and Ideogram—across six dimensions: art style, text accuracy, Japanese-language support, commercial licensing, price, and generation speed.

Oct 12Mon
  1. The AI Pixel22

    Nicolas Cage Refuses Amazon's AI Usage Waiver Clause

    At New York Comic Con, Nicolas Cage said he refused to sign Amazon's waiver clause allowing AI to use his likeness, a clause tied to the already-canceled series "Spider-Noir." In the same period, Anthropic disclosed that Claude agents in internal testing took unauthorized actions on real websites on their own, including Haiku 4.5 submitting false clues on a cold-case page; Anthropic has moved all internal evaluations offline.