Skip to content
Date and source

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

Sep 29Tue
  1. drama-skills · 公开创作项目发布48

    v0.8.1 Released: Still-Frame Editing, Character Voice Binding, and Tested Prompting Methods

    The static comic drama tool v0.8.1 adds still-frame editing, allowing direct editing from keyframes and voiceover into a finished video, with support for mixed still-frame and video editing, push-pull and panning, continuous camera movement by speed, and voiceover and subtitles across cut points. The video prompt adds a reference audio field, reconciled with the character voice reference in the visual settings, compiled as reference_audio during production, and must be bound to the speaking character in the shot. This version also supplements, based on small-sample testing, prompting methods for reaction shots, two-person contact, keyframe art style, and TTS tone; the conclusions are limited to the samples and conditions in the records.

  2. Lightricks · LTX 模型71

    Lightricks Releases LTX-2.5-22B IC-LoRA Alpha Gen Matting Model

    Lightricks released IC-LoRA Alpha Gen based on LTX-2.5-22B, which takes ordinary RGB video as input and generates frame-aligned grayscale alpha mattes without green screens, masks, or prompts, and can handle semi-transparent elements such as hair, fur, smoke, fire, sheer fabric, glass, and water.

    Why it matters: It generates frame-aligned alpha mattes from ordinary video without green screens or prompts, offering a new alternative for matting and compositing workflows.

  3. 龙爪槐守望者 · 创作者专栏62

    Why Can a Text Model Like Opus Also Make Videos? Breaking Down the New AI Code-Generated Motion Graphics Trend

    Making videos with Claude's Opus 5.5 has recently become popular; the author explains that it is not on the same path as Seedance or Kling: video models directly generate the pixels of each frame, whereas text models like Opus write motion graphics in code, generating HTML, CSS, JavaScript, Three.js, and other code, defining text, logos, UI, 3D elements, camera, and animation timeline separately, then capturing frame-by-frame screenshots through a renderer and compositing them into an MP4.

Sep 28Mon
  1. 可灵 Kling68

    Kling 4.0 to Be Released in October; 4.0 Flash Now Available to Ultra Annual Subscribers

    Kling announced that Kling 4.0 will be released this October, and Kling 4.0 Flash is now available to Ultra annual subscribers.

    Why it matters: Kling 4.0 announced capability upgrades including audio-visual, reference, and multi-keyframe features, allowing creators to assess room for adjusting their production workflows accordingly.

  2. second-unit · 开源创作项目发布22

    Second Unit 2026.09.28 Released: Windows Version Updates Workflow Directory and Multiple Fixes

    Second Unit released version 2026.09.28, updating the Windows download to the main branch and incorporating the workflow directories and fixes added after the September 15 package. The update includes MiniMax H3, LTX 2.5, SAM 3.1, WAN 2.2, and Juggernaut workflow directories, as well as model-specific prompt extensions, environment presets, seed handling, and an H3 Pose Follow fix.

  3. 公众号:短剧内行人58

    Interview: "I'm a Ghost Courier in Modern Times" — How a Bilibili creator hand-built an AI drama with over 10 million views

    The AI-made drama "I'm a Ghost Courier in Modern Times" by Bilibili creator "Hongdou Xifan Zhong" surpassed 11 million views with its first three episodes and was selected for Bilibili's "Weekly Must-Watch" multiple times in a row; the first episode cost over 6,000 yuan, and the third rose to 33,800 yuan. He alone serves as screenwriter, director, and editor, uses a script ratio of 7 minutes of comedy to 1 minute of sentiment within 8 minutes, and uses a stopwatch to generate visuals according to the actual duration of the dialogue. He signed with Bilibili's UP Animation Theater to receive computing subsidies and promotional resources, and believes AI dramas make it easier to build a creator's personal brand assets than the kichiku (remix/meme) section.

  4. 沃垠AI · 公众号公开订阅66

    Jimeng launches sample film mode: 480P drafts upgraded to clear 1080P, measured video cost reduced by about 38%

    Jimeng AI's web version launched a sample mode: first generate a draft at 480P, and after you are satisfied, natively upscale it to a 1080P final cut; the cost of 480P is about 1/8 of 1080P. The author tested a 15s video: two 480P generation attempts plus upscaling totaled 99 yuan, while generating directly at 1080P cost 156 yuan, saving about 38%; the final cut basically reproduces the sample, with no obvious difference to the naked eye in positioning, dialogue, core props, depth of field, or sound effects. Sample mode is also available in the canvas and retains Seedance 2.5 capabilities such as all-purpose reference, first and last frames, and smart editing, but does not yet support 3-minute ultra-long videos.

    Why it matters: Jimeng launches sample mode, replacing direct high-definition generation attempts with 480P drafts plus native upscaling, which creators can use to recalculate storyboard preview and trial-and-error costs.

  5. Lightricks · LTX 模型62

    Lightricks releases the LTX-2.5-22b IC-LoRA SDR-to-HDR model

    Lightricks releases LTX-2.5-22b-IC-LoRA-SDR-To-HDR, which converts 8-bit SDR video to 16-bit HDR and outputs EXR frames in addition to an HLG master.

    Why it matters: Lightricks turns SDR-to-HDR into an IC-LoRA based on LTX-2.5 and provides a ComfyUI workflow and EXR output path, so it can be integrated directly into post-production pipelines.

  6. VibeClips · AI视频创作教程22

    How VIBE uses models such as Kling, Sora, and Veo to generate AI music videos

    VIBE is an AI video generation app that can call models such as Kling, Sora, and Veo to turn a song, lyrics, or a single reference image into a music video ready to publish with a soundtrack. Its approach is to split the song into sections such as intro, verse, and hook, write visual prompts for each section separately and generate 4–15 second clips, then edit and stitch them together on a phone in song order.

Sep 27Sun
  1. Lightricks · LTX 模型71

    Lightricks releases the LTX-2.5-22b IC-LoRA old-film restoration model

    Lightricks releases LTX-2.5-22b-IC-LoRA-Restore on Hugging Face, an In-Context LoRA based on a reference video that renders old footage — low-resolution, low-bitrate, tape, or black-and-white/sepia scans — into clean, color, high-definition footage, removing compression damage, flicker, dirt, and scratches.

    Why it matters: Lightricks open-sources an old-film restoration LoRA, providing a two-stage workflow from 480-line scans to 4K and measured gains, which can be compared with existing restoration pipelines.

  2. Ciaro · 中文AI电影制作教程22

    How to maintain character consistency in AI video

    Character consistency in AI video is fundamentally a process problem rather than a prompt problem; character drift, prop inconsistency, and scene changes are the three core risks. A workable approach is to first break down the script, build a character bible plus prop and scene reference libraries, generate hero reference frames first and then derive shot variants, and review in batches and correct on the timeline. Ciaro Pro's asset management and character workflow can keep the script, reference images, generated frames, and edits linked together.

  3. note · genial_crocus627 · 图像视频创作分享22

    What to decide before choosing a video generation AI: 5 criteria to validate with free credits

    Before choosing a video generation AI, first write a one-sentence purpose stating "for whom, where, and conveying what," then validate it with free credits against 5 criteria: whether it can produce a video that fits the purpose, how many retries the free credits allow, whether it is easy to modify, whether the work hours required are acceptable, and whether it can be used continuously after paying.

  4. 袋鼠帝AI客栈 · 公众号公开订阅62

    Jimeng launches Sample Mode: 480P draft pulls, then native upscale to 1080P, measured savings of about 66% credits

    Jimeng AI has launched "Sample Mode": first use 480P for low-cost draft pulls, then once a satisfactory take is selected, natively upscale it to a 1080P final cut. The author measured that a 30s video at 480P costs 270 credits per run, while direct 1080P output requires 1920 credits; 5 draft pulls plus upscaling totals 3270 credits, saving nearly 66% versus 9600 credits for all-direct 1080P output.

  5. note · zou_aidouga · 图像视频创作分享34

    Don't take up an AI video side hustle if you've never made videos: the reasons edits get stuck and what to do before starting

    An independent video creator advises that people without video production experience should not rush into treating AI video as a side hustle right now; the core reason is that AI-generated video cannot be changed in just one place—every revision requires regeneration, causing visuals to change and costs and schedules to spiral out of control. In his September tests, rough-quality 1-minute video cost about 100 yen, but two videos generated from the same prompt and same image diverge visually from a certain second onward. He suggests first practicing with free editing apps to change only one thing at a time, cultivating judgment, and recording the number of redos and the cost.

Sep 26Sat
  1. note · aun_585 · 图像视频创作分享32

    How Dreamina Seedance 2.5 uses repeated prompts to prevent people and props from multiplying in the frame

    To address the problem of people or props multiplying on their own when Dreamina Seedance 2.5 generates video, the official guide only lists avoidance items such as no duplicate people, but in practice entering such negative terms often still fails. The author's approach is to repeat the description from three angles: explicitly only one woman, the number of people stays unchanged, no other people—and likewise for props—which is more effective than a single negative sentence. This method may still be broken when the number of shots increases or the prompt is complex, and the reason is not yet clear.

  2. note · ai_freak · 图像视频创作分享52

    How to unify characters and character voices in AI animation

    AI FREAK shares two approaches to unifying character voices in AI animation: one is to first generate the voice with Ondoku-san or ElevenLabs and then use Kling or HeyGen for lip sync; the other is to write the dialogue into the prompt when generating video with Seedance 2.5 so it is generated together. The author uses the second approach in the self-made 3DCG animation "BRAIN BUSTERS," but the same character still shows timbre changes across different shots, so the character features along with the voice timbre are written out to about 800 characters to fix the voice.

  3. AI Fruit · 中文图像视频制作指南38

    How to make AI Fruit fruit story videos: a full scene-by-scene tutorial

    AI Fruit's Story workflow can directly generate a multi-scene fruit short drama from a single plot synopsis; planning the scenes is completely free, and credits are only deducted when generating scenes. At the default Grok Imagine Video 720p rate of 3 credits per second, a 3-act short drama costs about 54 credits and 4 scenes about 72 credits, and an exact estimate is shown before clicking. Projects support 3, 4 or 6 scenes sharing the same model, resolution and aspect ratio, and export as 9:16 vertical video ready to post on TikTok.

  4. Pika18

    Pika API Club member spotlight: AI-Bridge

    Introducing the Pika API Club member spotlight—showcasing the developers and small teams building genuinely exciting work with the Pika API. AI-Bridge @doerstokyo342, an AI studio based in Tokyo, offers video production, corporate training and custom AI tools. After joining the Pika API Club, their generative media costs were cut in half. Read the full story 👇

  5. The New Protagonists22

    Future Film Fridays 09.25.26: Avid's agentic media production framework, YouTube vertical short dramas and Meta Muse

    Avid sets out four requirements for agentic media production: understand project context, operate within explicit permissions, know when human approval is needed, and leave an audit trail. Luma Agents, Runway Agent and Adobe Creative Cloud have each launched their own agent solutions, and Netflix acquired Ben Affleck's InterPositive for $587 million.

Sep 25Fri
  1. Ciaro · 中文AI电影制作教程22

    AI-generated shot handoff: how to preserve editing context when replacing shots

    Video editing for AI-generated content needs to separate media replacement from creative revision: if a new clip changes timing, eyeline, performance or composition, it should trigger an editing decision rather than a silent overwrite. The article proposes managing replacements with six types of records—shot identity, story function, approved intent, continuity constraints, replacement lineage and decision history—and uses the train station platform close-up of Scene 8 / Shot 14 as an example to show that even after V3 corrects the costume, it may still fail in the edit due to changes in eyeline or duration.

  2. Hedra 官方博客62

    Hedra Compares API Pricing Across 11 Video Models Including Seedance, Kling, Veo, and MiniMax: 10-Second 1080p with Audio Ranges from $0.60 to $7.91

    Hedra compared API pricing across 11 video models for the same task (10 seconds, 1080p, native audio, one starting frame plus a text prompt), with the lowest at $0.60 for LTX-2.3 Fast and the highest at $7.91 for Seedance 2.5—a 13x difference. Nine models meet the spec; Veo 3.1 maxes out at 8 seconds and MiniMax H3 has no 1080p setting, so both are listed separately as close substitutes.

    Why it matters: For the same 10-second 1080p with audio task, API unit prices across 11 video models differ by 13x, along with a reusable method for model selection and cost verification.

  3. AI Fruit · 中文图像视频制作指南52

    How to Choose Prompt-to-Video: LaunchVideo's Code Rendering vs. Generative Models—Two Routes

    LaunchVideo, released on September 24, 2026, uses a coding agent to write video as HTML, then renders it frame by frame with headless Chromium and encodes it to MP4 at 1920×1080, 30fps with ffmpeg. A single run takes about four minutes and consumes roughly 100,000 tokens, producing a 20- to 40-second finished video, without using any video model at all.

  4. VibeClips · AI视频创作教程15

    AI Video Creation Apps vs. Desktop Software: Why Mobile Became the Mainstream in 2026

    In 2026, AI video creation apps outperform desktop software in most scenarios because rendering runs on cloud GPU clusters rather than local chips—a $200 phone and a $3,000 workstation submitting the same task can get identical results. Apps like VIBE support models such as Kling, Sora, and Veo, generating 4- to 30-second videos at 480p to 1080p, and offer a free tier including Seedance Pro Fast.

  5. Perfect Video Resolution · ComfyUI创作项目发布18

    ComfyUI-PerfectVideoResolution Released: Workflow Rewritten in Pure Shell to Pass Security Scan

    ComfyUI-PerfectVideoResolution released a workflow update, rewriting the changelog extraction logic in pure shell. Previously, the embedded Python (pathlib/subprocess) in the workflow was flagged by the registry security scanner as python_sensitive_file_access, causing version 0.1.1 to be blocked.

  6. note · howrass_note · 图像视频创作分享40

    Generating Intense Combat Videos with MiniMax H3: Second-Level Prompting and Ref2V Reference Image in Practice

    The author attempted to generate an approximately 30-second gun-sword combat kill formation video with MiniMax H3, finding that when only writing vague prompts like "intense battle," the model would strike once and then stop and wait. Only after specifying actions segment by segment—approach, collision, dodge, counter, knockback—by time intervals such as 0–3 seconds, 3–7 seconds, etc., did the combat actually unfold.

Sep 24Thu
  1. note · genial_crocus627 · 图像视频创作分享30

    Design Methods for Repurposing AI Videos into Still Images, Note Articles, and SNS Copy

    Once an AI video is completed, it can be broken down into three types of content—still images, Note articles, and SNS copy—using a "material set" approach: first, consolidate the finished video, narration script, images used, title options, and key points, then define the video's core theme in a single sentence as the axis for repurposing.

  2. note · mii_chai_ai · 图像视频创作分享30

    Google Flow Generates a "My Pet's Potato Chips" Video from a Single Pet Photo, with 16 Prompts Included

    Using Google Flow's free tier, simply upload a single pet photo to directly generate a 9:16 vertical video of the pet lying in a bag of chips sneaking bites, without needing to generate an image first. The author published 16 prompts covering themes such as "My Pet's Potato Chips," "Jaga Chai," "Chai's March," "Chai's Mountain," and "Popcorn," including Google Flow video prompts, image generation prompts, and Gemini video prompts.