Skip to content
Date and source

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

Oct 10Sat
  1. SunoMV · AI音乐视频制作32

    SunoMV hybrid MV money-saving guide: allocate text PVs, storyboard images, and AI video by segment

    The SunoMV editor breaks a 4-minute, 21-shot song into five tiers of visual cost: using only text PVs and existing storyboard images for motion effects both cost 0 credits, AI motion effects cost about 118 cr/shot, AI video costs about 700 cr for a 13.8-second shot, and shooting the whole film costs about 11,500 cr. Its "Balanced" option only shoots the 5-shot instrumental section and the 6-shot chorus, at about 5,600 cr, using full-screen text PVs for the verses and video overlaid with text PVs for the chorus.

  2. note · shio_batapopcorn · 图像视频创作分享18

    Does AI understand "rock paper scissors"? An English rock-paper-scissors hands-on test with Kling 3.0 and MiniMax H3 [Pollo.ai verification]

    On Pollo.ai, using the same alien character image, 8-second duration, no camera movement, and the same English prompt, the understanding of rock-paper-scissors by Kling 3.0 and MiniMax H3 was compared. When given "Please play rock, paper, scissors.", both mapped Rock, Paper, and Scissors to the rock, paper, and scissors hand gestures, but it was more like performing the words than an actual match.

  3. Curious Refuge 创作教程62

    How to save money with draft mode: a hands-on test of AI video generation costs with MiniMax H3 and Seedance 2.5

    Curious Refuge hands-on tested two AI video generation workflows of draft first, then render, to produce a 15-second three-shot sequence. Using MiniMax H3 for drafts and then switching to Seedance 2.5 for rendering cost about $27 in total generation and about 60 minutes of labor; while Seedance 2.5 draft mode generated the first sequence for about $2, and standard mode for about $8.

  4. note · 2tm · 图像视频创作分享22

    Fully automated, free production of AI animation and four-panel comics with MiniMax H3

    A creator used MiniMax H3 in a Linux environment to fully automatically generate AI animation and four-panel comics, with hardware consisting of an RTX Pro 2000 Blackwell with 16GB VRAM, an AMD Epyc CPU, and 48GB of memory. The work "Kurochi to Paka" tells a short story about a black cat and an alpaca, and a short animated version was also released on YouTube.

Oct 9Fri
  1. 伊藤P · 低显存图像生成实测31

    MiniMax H3 + Veda tested: a 5-second video takes about 5 minutes, but a 6-second video takes 31 minutes

    When running MiniMax H3 + Veda Sparse Attention on an RTX 3060 Ti (8GB VRAM) to generate 1376×768 video, a 5.2-second video completed in about 5 minutes, but a 6.6-second video took 31 minutes 30 seconds, a 7.3-second video rose further to 38 minutes 34 seconds, and a 10.1-second video terminated abnormally due to a CUDA memory allocation failure.

  2. VideoGen · 视频剪辑与镜头实战22

    What Features Does Image-to-Video AI Have? Breaking Down VideoGen's Core Capabilities

    Image-to-video AI turns static images into short clips; core features include start-frame control, motion prompts or automatic motion, duration and aspect ratio settings, native audio or silent output, quality tiers, batch conversion, lip-sync on static images, reference-image consistency, and post-generation editing.

  3. Comfy 官方博客74

    How I generate real-time video on a single GPU with MiniMax H3

    On a single RTX 5090 (32GB VRAM), the author used FastVideo's FastH3 V2 checkpoint with four-step sampling to produce 15-second 448×256 real-time video, aiming to bring down the cost of continuous generation from 6k per day.

    Why it matters: The author breaks down one by one the six optimizations for running real-time video on a single RTX 5090, which can be transferred to your own production workflow.

  4. note · knowhow_matome99 · 图像视频创作分享30

    The complete ComfyUI × MiniMax H3 guide: generate AI video locally on a PC for free and monetize it

    A guide explains how to use ComfyUI with MiniMax H3 to generate AI video locally on a PC for free and monetize it. It recommends the 8-Step V2 workflow of the distilled acceleration model FastH3; on an RTX 2070 Super with 16GB VRAM and an i7, the author takes about 15 minutes to generate a single video, and suggests using I2V to first produce a first-frame image and then generate the video. The author says that from posting AI videos on TikTok alone, monthly earnings once exceeded 550,000 yen.

Oct 8Thu
  1. Stuubzzz · 视频工作流实测71

    MiniMax H3 Four-Step Output: PDMD LoRA vs. NVIDIA SoL Refiner Benchmarked on RTX 5090

    The author benchmarked MiniMax H3's PDMD 4-step LoRA and NVIDIA SoL Refiner on an RTX 5090: PDMD averaged 101 seconds to output a 6.6-second clip at 2520×1440, while the 20-step Quality path took 753 seconds.

    Why it matters: The author benchmarked PDMD 4-step LoRA and NVIDIA SoL Refiner on an RTX 5090, providing itemized timing and image quality differences that can serve as a selection reference.

  2. Stuubzzz · 视频工作流实测62

    Stubelius Ultimate H3 0.11 Adds Intermediate Frames, Segment Continuation, and Timeline Pinning in ComfyUI

    Stubelius Ultimate H3 0.11 is a free open-source MiniMax H3 workflow by the author for ComfyUI, adding three ways to control segments by seconds in First/Last Frame mode: up to two intermediate frames, placing a video in the First Frame box to continue a segment, and pinning up to 8 audio clips or short segments to specified seconds.

Oct 7Wed
  1. QwenVL-Mod: Enhanced Vision-Language · ComfyUI创作项目发布28

    ComfyUI-QwenVL-Mod Update: Cleans Up GGUF/HF Model Directories and Fixes MiniMax Presets

    ComfyUI-QwenVL-Mod has cleaned up the GGUF and HF model directories, keeping three vision-language (VL)-capable DavidAU heretic models per backend (4B / 9B / 27B), all verified as image-text-to-text with working mmproj files, and removed text-only duplicates and the mislabeled "3.8-9B" entry.

Oct 6Tue
  1. Hedra 官方博客60

    Hedra Launches Developer API: One Key to Access 110+ Generative Models

    Hedra launches its Developer API, allowing access to 110+ generative models with a single API key and prepaid USD wallet, covering video, image, audio, and text, including Veo 3.1, Kling V3, Seedance 2.5, MiniMax H3, Nano Banana Pro, ElevenLabs V3, and more.

    Why it matters: Hedra consolidates 110+ video, image, audio, and text models into a single API and prepaid wallet, enabling developers to evaluate the integration costs of multi-model pipelines.

  2. あいきみ · 图像编辑与LoRA实测58

    RTX 3090 Hands-On with MiniMax H3's 360° Orbit LoRA: Generating Orbit Videos from a Single Image

    The author tested pablodawson's MiniMax H3 360° Orbit LoRA on an RTX 3090, using the same image as both the first and last frame of the video, generating an approximately 3-second, 73-frame orbit video. The back of the subject, not captured in the original image, appears midway through.

  3. 新智元 · 公开公众号 RSS62

    Utopai X Ranks Second Globally on Artificial Analysis Blind Evaluation; Utopai Studios Valued at $1 Billion

    Utopai Studios' in-house video model Utopai X ranks second globally and first in the US on the Artificial Analysis blind evaluation leaderboard with an Elo score of 1150 (±10), leading in both audio synchronization and physical effects, surpassing Gemini Omni Flash 1.1 and Dreamina Seedance 2.5.

  4. FastVideo 视频模型 · 模型发布62

    FastVideo Releases FastH3 Trim Lightweight Model and ComfyUI Repackaged Files

    FastVideo released the experimental lightweight model FastH3 Trim, reducing transformer blocks from 50 to 42, using rank-16 timestep conditioning, 8-step sampling, with a 4.2x smaller size than the base H3 and faster than FastH3 V2.

    Why it matters: FastH3 Trim offers a lighter video generation solution with 42 layers, 8-step sampling, and 4.2x size compression, along with ComfyUI files and parameters for various precisions.

  5. 阡陌的小站 · 本地创作实战70

    Tutorial: deploying Qwen-Image-2.1 and MiniMax H3 locally in ComfyUI with 8GB VRAM

    Qianmo published a tutorial on generating images and videos locally via ComfyUI on an RTX 4070 Laptop with about 8GB VRAM, using the Qwen-Image-2.1 jailbreak GGUF workflow for images and MiniMax H3's 04_i2v_fl2v_4step_sla_lowvram workflow for videos.

    Why it matters: A complete path to running Qwen-Image-2.1 and MiniMax H3 under 8GB VRAM, including model files, directory structure and parameter configuration, reproducible by comparison.

Oct 5Mon
Oct 4Sun
  1. note · hideout927 · 图像视频创作分享40

    Generating character design sheets with MiniMax H3 in ComfyUI: no dedicated LoRA needed

    MiniMax H3 can leverage its own reference image and prompt comprehension capabilities to generate character design sheets in ComfyUI without a dedicated LoRA or additional training; generation uses Core nodes, and custom nodes only handle generating JSON and natural-language prompts. The node combines view instructions such as front, side, back, and facial close-up with the original image to keep the same character and clothing consistent, with H3 filling in parts not photographed. Japanese prompts perform worse, while English prompts pass more easily; if separate hand and foot images are not needed, this feature can be turned off.

Oct 3Sat
  1. FastVideo 视频模型 · 模型发布62

    FastVideo releases the FastH3 Trim 8-step audio-video generation model

    FastVideo released the experimental model FastH3 Trim, which removes 8 of the least impactful blocks from FastH3 V2's 50 MiniMax H3 transformer blocks and replaces each block's AdaLN timestep projection with a shared rank-16 basis, trained with 8-step DMD2 (checkpoint 300), enabling 8-step text-to-video generation with synchronized audio.

    Why it matters: FastH3 Trim uses pruning plus shared AdaLN to bring 8-step audio-video generation to consumer-grade hardware, and can be compared against V2 to weigh image quality against size.

  2. note · wise_azalea8300 · 图像视频创作分享65

    ComfyUI Video Perspective Change Workflow: Running Locally with MiniMax H3 and Angle Change LoRA

    The author released a ComfyUI workflow that uses the video generation model MiniMax H3 combined with an angle change LoRA to alter the camera perspective of existing videos, enabling wide-angle, side, low-angle, and 360° Orbit effects around a target. The workflow runs entirely locally and is free, with various techniques allowing execution in an 8GB low-VRAM environment. It supports both SFW and NSFW content, and the article includes download links and import instructions.

    Why it matters: The author published a ComfyUI workflow for changing perspectives using MiniMax H3 and an angle change LoRA, along with a local running solution for 8GB low-VRAM environments.

Oct 1Thu
  1. note · eeeemo_japan · 图像视频创作分享66

    MiniMax open-sources video generation model H3: runs locally, supports Omni Reference

    MiniMax freely released the open-source video generation model MiniMax H3 in August 2026; it can be installed and run on your own computer, reducing the cost of repeated regeneration to electricity. It supports Omni Reference, allowing up to 9 images, 3 videos, and 3 music tracks as references to maintain character and scene consistency; according to ASCII.jp, its cloud version's performance is considered comparable to Seedance 2.0.

    Why it matters: The open-source video model pushes the cost of a single generation down to electricity, and the article gives VRAM thresholds and licensing regional restrictions to help assess the feasibility of local production.

  2. Theoretically Media blog62

    Early Hands-On with Kling 4, Plus a Look at Gemini 4 Argon, Ideogram 4.5, and Eleven v4

    The author conducted hands-on testing of the Kling 4 early access version, which currently only offers the Flash model with 720p output, available to Ultra subscribers on annual plans; larger models are planned for October. Testing shows that Kling 4 appears to have resolved the lip-sync drift issue from 3.0, supports up to ten keyframe references, but performances are somewhat flat.

Sep 30Wed
  1. note · synth_brain · 图像视频创作分享65

    Three ways to make MiniMax H3 high-speed and high-resolution: Fused-Turbo-Int8, Spectrum, and Latent Upscaler

    The author shared a high-speed, high-resolution all-in-one ComfyUI workflow for MiniMax H3, the core of which is replacing the base model with the minimax-h3-fused-turbo-int8-convrot model, which integrates FL2V and Ref2VA, has a built-in 4STEP TurboLoRA and Mystic V2 LoRA that suppresses jitter and ghosting, and is about 21GB in size.

    Why it matters: The author tested MiniMax H3's model replacement and speed-up nodes, providing a comparison showing an 800×800 eight-second video dropping from 7 minutes 13 seconds to 3 minutes 37 seconds, which can be transferred to existing video production workflows.

Sep 26Sat
Sep 25Fri
  1. 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.

  2. 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 23Wed
  1. Wireflow · AI短剧与视频制作教程66

    How to Create AI Song Ads on a Single Wireflow Canvas Without CapCut

    Wireflow publishes a tutorial on how to complete an 80-second AI music ad on a single canvas — from script, song, and word-level timeline to stills, motion, subtitles, and end credits — without ever exporting to CapCut or Premiere.

    Why it matters: A complete breakdown of the eight steps to produce an 80-second AI music ad, from writing lyrics to rendering, along with the actual cost of approximately 6,100 credits and a record of pitfalls encountered.