Reddit user wildmonkeywrangler used MiniMax-H3 Image on RTX 5060 Ti 16GB to test generating hyper-realistic documentary-style photos. The prompt required five water bottles covered in gorilla fur, two of which had gorilla faces, and set negative prompts to exclude CGI, synthetic fur, etc.
Meituan released the audio-driven digital human model LongCat-Video-Avatar 1.6, which supports stable generation of over 10 minutes and uses an 8-step inference path. In 33,051 manual votes, it ranked second, leading by only 1.5, about 16 Elo. ComfyUI support remains at 1.5, and 1.6 has no node path yet.
A Reddit user is seeking suggestions for training an accent LoRA for LTX 2, considering using version 2.3 or 2.5, and currently has only SD 1.5 image LoRA experience. Their hardware is an RTX 3060 with 12GB VRAM and 64GB RAM, they have configured LoraTraining and AI Toolkit, and are using YouTube audio and auto-generated subtitles as training data.
The short drama industry has undergone three reshuffles, from paid traffic acquisition to the free model and now the AI era. Production is no longer scarce; distribution, copyright, and aesthetics have become the key factors determining whether a work can be seen and retain profit. In the first half of 2026, native new dramas on Douyin alone reached 221,900 titles, with only 0.48% breaking 100 million views. AI dramas crossing the 50 million view breakeven line accounted for just 1.3%. Hongguo's daily active users reached 203 million, exceeding the combined total of iQiyi, Youku, Tencent Video, and Mango TV. WeChat's Green Bubble short drama mini-program reached 143 million monthly active users.
Kling 4.0 and Seedance 2.5 simultaneously appear in Runway and Adobe model lists within weeks, but the same model name does not share quotas or rates. Runway and Adobe use their own billing units, and Adobe explicitly sends prompts and reference files to model developers.
Why it matters: The same model name may correspond to different outputs, prices, and terms on different platforms. When choosing a platform, platform terms should be included in model selection, rather than just looking at the model name.
On November 12, the Douyin comic drama daily chart Top 5 had a total of approximately 567 million new plays, up about 12.7% from the previous day, with a reshuffle rate of 40%. 'Nianji Laolu' topped the chart for consecutive days with 160 million plays. 'After Being Divorced, I Turned Things Around by Remarrying' rose to second place with 130 million new plays, an increase of about 31.9%; 'Mountain Wild Apricot Grove Helps Revitalize the Countryside' entered fourth place on its third day online, with 91.214 million new plays. Rural themes and family relationships have become important content angles.
On November 13, multiple micro-short dramas were released intensively on platforms such as Hongguo, iQiyi, and Tencent Video, with suspense, revenge, sweet romance, and comeback themes running in parallel. iQiyi's exclusive 'Return of the Mask' premiered, while 'Wild Dog and Bell' and 'For Your Peace' from Hongguo's lineup started airing. Hongguo's trending chart has been almost entirely taken over by AI dramas. Short dramas go from scheduling to release in just a few days, and the daily update pace continues to heat up.
China's first Xilankapu intangible cultural heritage micro-drama 'Jinxiu Zhixia' will air on the Hongguo short drama platform starting November 13. The drama uses the Tujia brocade technique of continuous warp and broken weft and Enshi folk songs as elements, and uses a time-travel plot that rewrites fate to push intangible cultural heritage and cultural tourism out of their niche. It is expected to drive a cultural tourism conversion chain of watching the drama, being inspired, traveling to Enshi, and experiencing intangible cultural heritage.
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.
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.
Higgsfield's official blog published a complete guide teaching creators how to create a faceless YouTube channel from scratch using AI, with the example channel being Statues With a Story. The guide clarifies YouTube's requirement to disclose realistic AI-generated content and lists the new thresholds for new applicants to enter full YPP starting February 1, 2027: 8,000 hours of watch time or 20 million Shorts views.
Why it matters: This guide provides a complete process from topic selection to monetization, and clarifies YouTube's disclosure boundaries for AI-generated content and the new YPP thresholds for 2027.
Synthesia has been selected for the inaugural AI 100 list, becoming one of the top 100 private AI companies globally. In 2026, the company completed a $200 million Series E funding round at a $4 billion valuation, and launched Syren, an AI video agent that can turn prompts into brand videos; it also brought video creation into ChatGPT and Claude through MCP integration. The company has over 65,000 customers, including 90% of the Fortune 100.
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.
Creator Mohannad Khamra's The Pink Hotel is a three-minute surreal short film, released in June 2026, generated by LTX 2.0. Upon rewatching, the creator noted that some shots have softening, temporal inconsistencies, and loss of detail, but believes the story itself still holds. The creative goal was not technical perfection, but to use generative video to build a short film with consistent style, deliberate camera work, and consistent narrative pacing and humor.
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.
JVT (Just Video Transformer) proposes a simple large-patch DiT-style Transformer for pixel-space video diffusion, without pretrained autoencoders or U-Net-style skip connections. Its generation quality is comparable to latent diffusion baselines in both class-conditional and unconditional settings, and it supports joint generation of pixel-aligned modalities. Through camera-conditioned RGB-disparity generation, JVT improves camera following and obtains explicit scene geometry without requiring an additional reconstruction model.
VennifyAI releases ComfyUI node pack Happy Latent, using LTX 2.5 through a timeline editor, supporting multi-scene video, 4K upscaling, reference frames, and custom audio. GitHub provides Happy Latent and free workflows, including a starter workflow and an upscale workflow.
Adobe and Microsoft announce expanded collaboration, with Photoshop, Premiere, After Effects and other apps receiving AI processing, editing, color and effects optimizations on Windows high-performance PCs.
Luma's AI cartoon generator guide introduces three types of tools: photo-to-cartoon, text-to-cartoon, and animation, emphasizing cross-scene character consistency. Luma's Ray 3.2 supports multi-keyframe sequencing and HDR workflows, and Layers driven by Uni-1 allow individual element editing. Cartoonize.ai offers over 20 art styles, suitable for quick photo-to-cartoon conversion.
Micro short drama 'Young Madam from Northeast 3' released a trailer, with 8.034 million reservations across all platforms, expected to launch on November 14. The drama stars Liang Wenjing and Yewen Kevin, continuing the Northeast comedy and family comeback setting, and ranks among the top of the short drama reservation charts. Reservations exceeding 8 million provide a traffic base for the launch, and the platform will give weighting in distribution; whether it converts to views and payments remains to be seen.
Hongguo Short Drama's 11.11 chart shows 'The Heiress's Scheme' topped the daily chart with over 100 million heat value, and 'The Legend of the Ten Thousand Demon Records Season 14' was far ahead at No.1 with 94.37 million comprehensive heat. In the AI drama hot broadcast chart, 'The Naive Son-in-law of the Great Tang: My Sister is the Crown Princess Season 5' topped with 68.07 million. The chart metric is comprehensive heat, not actual recharge data, and should not be directly equated with play counts.
The women's football-themed micro short drama 'Chasing Wind Girl' has launched on the Hongguo platform, reaching a peak popularity of 25.82 million. The drama is produced by Maiya Cultural Tourism, based on the real story of the Sanhe Primary School women's football team, using women's football competition as an entry point to tell a story of women's growth. Industry insiders believe that vertical realistic themes still have room for growth in the Hongguo ecosystem, and diversification of themes will become key to platform differentiation competition.
AI micro-drama "山河对弈帝王棋" launches on 11月12日, with members able to watch the full series. The AI production path is moving from novelty to regular supply, but issues of visual stability and portrait authorization are also surfacing. Platforms need to front-load AI synthesis labeling, portrait authorization, and content review to let the track escape wild growth.
AI short drama '要送我联姻那我就换个继承人' premiered on November 5 and rose from No. 89 to No. 23 on Hongguo's daily chart within four days, with a popularity index of 57.34 million. The drama's page has already labeled its format as an AI short drama. The article argues that its upward trajectory is a signal that AI short dramas are shifting from competing on volume to competing on quality. Against the backdrop of over 220,000 AI dramas being released in the first half of the year, with only 0.48% surpassing 100 million views, this drama's independent run driven by character design and pacing is particularly worth watching.
A comparison of MOSS-Image and MiniMax H3-Image under 15 identical prompts shows that the quality gap between the two is smaller than expected. Both perform comparably on English posters, Chinese signage, portraits, and product photography, but both fail on UI mockups, and charts generate meaningless labels. The differences come more from workflow and control than from raw output quality.
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.
The author tested the community GGUF quantized version (Q4_K_M, 11.57GB) of MiniMax-H3-Image on M1 Max 64GB, generating a 4MP anime illustration in about 5 minutes, with 6 sampling steps at about 47 seconds per step.
Why it matters: Community GGUF quantization makes MiniMax-H3-Image runnable on a 64GB Mac, and the author's testing provides a complete reference for speed, image quality, and editing workflow.
MOSS Image's 4-step LoRA has been run on iPhone, with DiT and TE using 4-bit QAD quantization and running on GPU, and VAE as fp16 running on Neural Engine.
MiniMax H3-Image (open-source weights) and Seedream 5.5 generate high-end fashion editorial photos under the same compact powder reference and 16:9 fisheye prompt. The author prefers H3-Image's composition, lighting, and silver reflections. Seedream 5.5 is stronger in realism, especially facial details, but H3-Image has better framing and visual balance. As an open-source weights model, H3-Image offers good value in image quality.
Baidu unveiled ERNIE 5.5 Flash Preview at Baidu World 2026, featuring a sparse MoE architecture with 280 billion total parameters and 14 billion activated, supporting a million-token long context.
Runway now supports 3D. Users can use Tripo H3.1 to generate detailed 3D models from text or images, or upload their own models. Generated models can be used as references for image or video generation from any angle, and can also be exported for use in other 3D tools.
娱乐资本论 survey found that the AI short drama 'The War Hero’s Bloodline Strikes Back', which was hotly discussed on Chinese social media as 'breaking into Hollywood', actually launched on DramaWave in July this year, with a cumulative total of approximately 63,700 favorites and 492 comments, far below the platform's daily chart TOP1 with 850,000+ favorites.
NEM Zagreb 2026 announces first batch of speakers. The event will be held on December 7-8 at the Esplanade Hotel in Zagreb. The conference focuses on three main themes: content, video business, and connected TV. It will discuss topics such as micro-drama and audience behavior, and will host the NEM Awards ceremony. Timothy Oh, CEO and CMO of COL Group International, will serve as the keynote speaker.
free-short-video v8.1.1 released. Docker images support multi-architecture linux/amd64 and linux/arm64, and have been pushed to GHCR and Docker Hub. The npm package can be run via npx or global installation, requiring Python 3.10+ and ffmpeg. This release fixes issues with keyframe mode upload, end frames, and style plates, with no API breaking changes.
Zebracat alternatives should be chosen by input type: Revid AI is the closest, Visla suits real footage, and Elai.io suits on-camera explanations. Revid AI Hobby is $39/month and supports text, links, or trending video mashups; Visla Pro is $22/month and focuses on real footage; Elai.io Creator is $29/month.
The AITuber guide categorizes Kaiber music video alternatives by use case: Neural Frames is closest to its artistic style and supports whole-song, track-by-track, and frame-by-frame adjustments; Runway uses Gen-4.5 to provide shot-by-shot control but requires you to edit it yourself. Plazmapunk offers AI clips for €9.99/month, Rotor Videos edits real footage to the track, and AITuber supports word-by-word captions.
Seedance 2.5 Flash was released on November 12. A 5-second 720p no-input video example is priced at approximately 0.99 yuan/second, about 34% cheaper than Seedance 2.5 at the same specifications, but 1080p and 4K are not yet supported. The vendor claims generation speed is about twice that of Seedance 2.5; the author notes this data comes from the vendor and no independent benchmark has been seen. The API is expected to launch on Volcano Engine on November 15, and will gradually become available in applications such as Jimeng.
Why it matters: Seedance 2.5 Flash is priced about 34% lower than Seedance 2.5, and speed is claimed to be about 2 times faster, but the author reminds that these data come from the vendor and need to be verified independently.