Skip to content

Content Formats

Tutorials & Practice Latest

Ready-to-use practical content: workflow setup, tool usage, cost-reduction experiences, and lessons learned from pitfalls.

134 picksLast 30 days 76 itemsCollected 1,418 items

updated

Pick archive · page 2

Nov 7Sat21–40
  1. Reddit StableDiffusion · 创作者与工作流分享67

    ComfyUI-MiniMaxH3-TemporalWindow-SpatialTiles workflow generates 4K video without OOM on 16GB VRAM

    Reddit user Select_Bowler3099 shared a ComfyUI workflow that uses MiniMaxH3-TemporalWindow-SpatialTiles nodes to generate a 10-second 4K video on 16GB VRAM + 32GB RAM, with no OOM throughout, taking about 2 hours.

    Why it matters: A workflow for 4K video generation on 16GB VRAM; the author has published the node configuration and timing data, and consumer-grade GPU users can refer to this approach to avoid OOM.

Nov 6Fri
  1. AITuber · AI短视频与提示词71

    Can ChatGPT make videos? Yes, it can generate complete videos through the AITuber plugin.

    ChatGPT itself cannot generate videos, but through the AITuber plugin it can generate a complete MP4 with voiceover, visuals, and subtitles. The free version is available, and new accounts receive 100 credits. The plugin connects via MCP and can generate videos up to 20 minutes long per session, and supports automatic publishing to platforms such as YouTube and TikTok.

    Why it matters: ChatGPT no longer has a video model. This article provides a reusable workflow for complete video generation via MCP plugins, including free quotas, configuration steps, and automatic publishing.

  2. Stable Diffusion Art 教程65

    MiniMax H3 360 Orbit LoRA Tutorial: How to Turn a Static Photo into a 360-Degree Orbiting Infinite Loop Video

    Using MiniMax H3's 360 Orbit LoRA combined with the first-and-last-frame technique, you can generate a 360-degree orbiting infinite loop video from a single static photo in ComfyUI. The complete steps include model download, workflow loading, parameter settings (768x768, 73 frames, 28 steps), and the frozen-time prompt; rendering takes about 4-6 minutes on an RTX 4090.

    Why it matters: The tutorial provides the complete steps for generating a 360-degree orbiting infinite loop video using the 360 Orbit LoRA combined with the first-and-last-frame technique, which can be directly reproduced in ComfyUI.

  3. Kapwing 创作指南64

    7 video editing tasks that can be automated with ChatGPT or Claude

    Kapwing tested automating seven types of video editing tasks with ChatGPT or Claude, covering editing planning, subtitle generation, translation dubbing, and B-roll production. In the test, Claude Code with FFmpeg built and rendered a 70.7-second YouTube Short, while Kapwing MCP completed AI dubbing in five languages. The article notes that standard chat interfaces cannot directly edit videos; actual editing requires a coding agent or an editor connected via MCP.

    Why it matters: The article provides seven types of automatable video editing tasks along with corresponding prompts and tool choices, allowing creators to build their own AI editing workflows based on them.

  4. Kapwing 创作指南65

    How to Use ChatGPT to Edit Videos (2026 Guide)

    Kapwing MCP connects ChatGPT to a video editing workspace, allowing users to complete editing tasks such as uploading, subtitling, translating, dubbing, and generating B-roll using natural language. In testing, a single combined request generated six exported versions (the original English version and five dubbed versions).

    Why it matters: Kapwing MCP allows ChatGPT to directly operate a video editing workspace. The tutorial demonstrates the complete natural language editing workflow from uploading, subtitling, translating, dubbing, to generating B-roll, and provides reusable prompt examples and cost references.

Nov 5Thu
  1. Reddit comfyui · 创作与工作流分享77

    ComfyUI Tutorial: Freeze Vocal Injection into LTX 2.5 Lip-Sync, Wan 2.2 for Motion, Creating a 3:20 AI Music Video

    Reddit user keto_brain shares a ComfyUI workflow for creating a 3:20 AI music video: encode the isolated vocals as a latent and set the noise mask to 0, letting LTX 2.5 lip-sync to the provided vocals, while Wan 2.2 handles non-singing motion.

    Why it matters: The author has published the node chain for LTX 2.5 lip-syncing to provided vocals and the Wan 2.2 motion shot settings, which can be directly transferred to similar AI music video production.

  2. 可灵官方博客78

    TechHalla reviews Kling 4.0 and 4.0 Flash: control and speed

    Creator TechHalla evaluates Kling 4.0 and 4.0 Flash in a real project, saying 4.0 is for control and image quality, while 4.0 Flash is for speed and batch. He uses 3D animation as motion reference and images as identity reference, generating weapon animation in three steps with timing preserved; multi-shot hard cuts allow 16 generations to complete an entire music video, and 19 generated characters all pass consistency on the first try.

    Why it matters: Creator evaluates Kling 4.0 and 4.0 Flash in a real project, demonstrating transferable workflows such as multi-shot hard cuts, 3D motion reference, and character consistency.

  3. Reddit comfyui · 创作与工作流分享72

    MiniMax H3's RefMods Workflow: Maintaining Character and Voice Consistency on 6 GB VRAM

    The author built a RefMods workflow for MiniMax H3 in ComfyUI, which maintains character and voice consistency on 6 GB VRAM without LoRA training. Extract character references from 8 images or voice references from short audio, save them as reusable mods, then load them into any generation; set both visual and audio strength to 0.9. For two characters, it is recommended to save a separate mod file for each character and voice, and clearly describe the subject in the prompt.

    Why it matters: This RefMods workflow maintains character and voice consistency on 6 GB VRAM without LoRA training, allowing creators to reproduce the cross-shot consistency process.

Nov 4Wed
  1. Kapwing 创作指南71

    Kapwing Guide: When to Label Videos with AI Labels

    Kapwing has released a creator guide with a guiding principle: AI-generated content that viewers might mistake for real should be labeled with an AI label. The guide summarizes the differing rules on platforms such as TikTok, Instagram, Facebook, and YouTube, and cites research showing that AI labels can affect engagement, recommending that creators add their own wording to the disclosure.

    Why it matters: It provides a general principle to apply: AI-generated content that could be mistaken for real should be disclosed, and it compares platform rule differences to help creators decide when to label.

  2. note · kongo_jun · 图像视频创作分享65

    Hands-on test of MiniMax H3's open regenerate 2k refiner in ComfyUI

    kongo_jun tested MiniMax H3's open regenerate 2k refiner LoRA in ComfyUI and provided complete setup steps and generation results. Both LoRA ranks (384 and 16) took 270 seconds to generate, with little difference in results; the author recommends using rank_16. The author suggests specifying the first frame to control the image, otherwise the image will be altered; in actual testing, the resolution was set to FHD (1920×1080) due to memory considerations.

    Why it matters: The author tested two rank LoRAs of MiniMax H3 2K refiner, providing complete ComfyUI setup steps and generation results, which can be used to reproduce locally and choose rank_16.

  3. pixaroma · YouTube 创作视频70

    MiniMax H3 in ComfyUI hands-on: character replacement, lip-sync, singing, and timeline (Ep41)

    pixaroma tests MiniMax H3 in ComfyUI, releasing 25 free workflows plus 2 bonus workflows. H3 is a video model with audio; the workflows cover character replacement, lip-sync, singing, timeline long videos, and 2K remastering. Requires ComfyUI 0.42+ and Pixaroma 1.4.239+.

    Why it matters: The author tested MiniMax H3 and released 25+2 ComfyUI workflows covering character replacement, lip-sync, singing, and timeline long videos, ready for direct reuse.

Nov 3Tue
  1. AI Fruit · 中文图像视频制作指南68

    HyperFrames GitHub Repository Guide: Installation Tutorial and Comparison with Remotion

    HyperFrames is an open-source HTML-to-video framework by HeyGen. The author tested its installation and rendering workflow and compared it with Remotion. On a Mac, a 10-second 1920×1080 video rendered in under 20 seconds, producing deterministic MP4 output suitable for visual regression in CI. The framework is Apache-2.0 licensed with no per-render fees, making it suitable for agentic creation; Remotion has a more mature ecosystem and is better suited for React teams.

    Why it matters: The author tested HyperFrames' complete rendering workflow and compared it with Remotion in terms of licensing, ecosystem, and agent suitability, helping creators choose the right tool.

  2. Reddit comfyui · 创作与工作流分享69

    ComfyUI Tutorial: PIKA H3 + MiniMax H3 Singularity V2 + NVIDIA Regenerative 2K Super-Resolution KIJAI LORA (6GB VRAM)

    The author shared a ComfyUI workflow that uses MiniMax H3 Singularity V2 and Pika H3 LoRA to generate a 0.6MP video in 6 steps, then enhances it to Full HD via regenerative 2K super-resolution. Total generation plus super-resolution time is about 1 hour, final resolution 1920×1080, runs on 6GB VRAM. The author provided the required LoRA, custom nodes, and workflow links.

    Why it matters: This workflow demonstrates a path of low-resolution generation combined with regenerative super-resolution enhancement, allowing Full HD video to be produced with only 6GB of VRAM, and the method can be transferred to other models.

Nov 2Mon
  1. Stuubzzz · 视频工作流实测75

    MiniMax H3 Character LoRA Training Test: 29 Clips, 4000 Steps, Why Step 3000 Was Chosen

    The author trained a MiniMax H3 character LoRA using AI Toolkit on an RTX 5090, with 29 3-second clips, rank 16, 4000 steps, taking about 15 hours of GPU time, and ultimately chose the step 3000 checkpoint.

    Why it matters: The author tested training a MiniMax H3 character LoRA using AI Toolkit on an RTX 5090, and provides the specific settings of 29 clips and 4000 steps, along with 5 reusable rules for reference when training other video models.

  2. Stuubzzz · 视频工作流实测70

    MiniMax H3 RefMod and reference image testing: long-video facial consistency comparison

    The author's testing found that MiniMax H3's RefMod can improve long-video facial matching from 0.23 to 0.80; the 0.5 MP version scored 0.75 in the second segment compared to 0.25 without RefMod.

    Why it matters: Testing shows that MiniMax H3's RefMod can improve long-video facial matching from 0.23 to 0.80; the 0.5 MP version offers better value.

Nov 1Sun
  1. Reddit StableDiffusion · 创作者与工作流分享77

    Author creates 46-second sci-fi trailer entirely locally on MacBook M5 Max and reveals complete pipeline

    Author completes 46-second sci-fi trailer 'THE LAST SIGNAL' entirely locally on a MacBook Pro M5 Max 128GB, with no cloud rendering and no API calls, taking about 14.5 hours (including about 2 hours of waiting).

    Why it matters: The author reveals the complete local pipeline from storyboard to final film and the actual time taken, allowing creators to assess the feasibility of replicating a similar workflow on a MacBook.

  2. Reddit comfyui · 创作与工作流分享70

    The author used MiniMax H3 and the self-developed Video Builder in ComfyUI to create a Halloween music video.

    The author used the self-developed Video Builder to have Claude automatically complete lyrics, prompts, and the render queue in ComfyUI, producing a MiniMax H3 Halloween music video. The entire video has 19 scenes, 2:29, with 80 renders in total and about 12 GPU hours; the author manually handled story ideas, the Suno song, and Nano Banana scenes, while Claude automatically generated characters, planned shots, and checked lip sync.

    Why it matters: The author published the complete division of labor and 2-pass workflow for Claude-driven ComfyUI music video production, which can be reused in one's own video generation pipeline.

Oct 31Sat
  1. note · cocoredo_2509 · 图像视频创作分享68

    Pika H3 Flex: Import, Quality, and Cost Verification for Video Creators

    The author verifies Pika H3 Flex from a production-site perspective, pointing out that official performance cannot be directly equated with actual usability; character consistency, audio track separation, and 4K upscaling need to be evaluated separately. H3 Flex is based on MiniMax H3, supports up to 12 reference materials and 5 types of audio track separation, and after generating at 480p/768p can be upscaled to 1080p/2K/4K, but 4K is upscaled rather than natively generated.

    Why it matters: On-site verification of Pika H3 Flex shows that character consistency, audio track separation, and 4K upscaling need to be evaluated separately; the author's three short tests and recording method can be directly used for other model adoption decisions.

  2. Curious Refuge 创作教程69

    How to Use AI to Relight a Video (Step-by-Step Tutorial)

    Curious Refuge published a tutorial demonstrating how to use AI to relight videos, and tested the workflow's effectiveness on live-action footage and gray-box studio scenes. The workflow first exports still frames and relights them using an image model (Nano Banana 2.1 or GPT Image 2.5), then uses the relit images as references to apply to the original video. Testing found limitations such as exposure shifts, banding, and resolution loss, with output resolution lower than the 4K source.

    Why it matters: This workflow uses static-image relighting as a reference for the video model, allowing creators to evaluate the boundaries and limitations of AI relighting in post-production.

  3. Higgsfield 官方博客76

    Higgsfield's real-world test of 2026 AI video costs: a 10-second finished video ranges from $1.35 to $7.85

    Higgsfield's test shows that a single generation of a 10-second AI video costs about $7.85 (160.125 credits) on Seedance 2.5 at 1080p, and about $1.35 on Kling 4.0 at 720p. The actual cost must also account for retakes, image preparation, and post-production.

    Why it matters: Provides a method and budget formula for calculating the true cost of AI videos under per-generation pricing, allowing creators to plan retake counts and monthly output accordingly.