AI ToolsFOUR CATEGORIES CAN FIRST BE GROUPED ACCORDING TO USE: LARGE LANGUAGE MODEL, PHOTO GENERATION, VIDEO GENERATION, ASI AGGREGATION/WORKSTREAM PLATFORM。
Three concepts need to be identified:
White does not have to study model parameters at the outset, but it is enough to learn “where the portal is, what it is appropriate to do, how to ask questions, how it is to be repeated”。
Models are the capabilities behind them, such as GPTImage 2.0, Seedream, Veo;
The product entrance is where you actually opened up, like the chatGPT, Gemini, the bean bag, the dream;
The Syndication Platform is a platform where many models, materials, canvass and jobs are moved together, such as Loveart, Tapnow, FlowPix, LibTV。
I. LANGUAGE MATERIALS: LEARNING "QUESTIONS" AND "SAILAI TO WORK FOR YOU"
Each large-language model has its own characteristics and advantages and its main functions are similar (although the foreign model is clearly stronger than the bean bag)。
1. ChatGPT
web portal: (chatgpt.com)
Suitable for what:
WRITING, CHANGING ARTICLES, SUMMARIZING MATERIALS, TRANSLATING, MAKING FORMS, HEADSTORMS, LEARNING COACHING, CODE INTERPRETATION, PROGRAMMING, AND GENERATING AI CREATIONS (PRODUCTION OF PHOTO TIPS, VIDEO SCRIPTS, MIRRORS, CONTENT SELECTIONS, BUSINESS CREATIVE PROGRAMMES)
Basic usage:
Instead of just saying, "Book me an article," you're going to say identity, purpose, audience, format and style。
2. Gemini
Entry:
Gemini (gemini.google)
Suitable for what:
SEARCH FOR QUESTIONS AND ANSWERS, DOCUMENTATION, ENGLISH CONTENT PROCESSING, PHOTO UNDERSTANDING, IMAGE AND VIDEO GENERATION PORTALS, WHILE GENERATING AI, FOR EXAMPLE
Phrasing generation, video scripts and mirror design。
Gemini's membership has the advantage of enjoying most of Google's other AI platforms
For example:
2.1
Flow (https://labs.google/fx/tools/flow/)
Google's AI video and photo creation platform, with pro membership, has access to the unlimited use of the Nano banana model and 10 ve03.1 videos a month。
2.2
NotebookLM (https://notebook.google/)
NotebookLM is Google ' s AI notes and resource research tool for uploading documents, generating summaries, questions and answers, learning notes, audio overviews and video overviews。
2.3
Google Al Studio
This is a platform for developers to test Gemini models, make prototypes, call Gemini APLI, experience Nano Banana Pro and Gemini Pro。
Bean buns
access: https://www.doubao.com/
Free tools at the national level (less than CHTTPT, gemini, but free)
Suitable for what:
CHINESE WRITING, SHORT VIDEO SCRIPTS, SMALL RED BOOKS, LEARNING GUIDES, PHOTO VIDEO GENERATION, PPT/DOCUMENT TASKS, AND ALSO GENERATING AI
Doing, for example, bulk content selection, role-setting, story-planning, business-creative programmes。
Case of use of large language models:
1. Design video lenses
Before using video tools such as Clin, DreamSeedance, Veo, you can have a large language model to break you into multiple lenses。
Example:
Please break down the video theme into five spectroscopys: “How ordinary people use Al to improve their work efficiency”. Every spectroscopy includes a visual description, camera exercise, subtitles and bystanders。
2. Optimizing graphic and video alerts
Example:
If the first image or video generated is not working well, the large language model can help you to optimize the hint。
The problem with this is that the picture is too messy and the title is not prominent. Please help me fine-tune the hint so that the picture is more concise, the title area is more visible and suitable for the tutorial cover。
3. Batch generation content selection
Example:
When it comes to media, curriculum, and community content, it allows large models of languages to generate lots of choice questions。
PLEASE PRODUCE ME 20 SELECTED CONTENT QUESTIONS FOR AI WHITE, INCLUDING IN THE DIRECTION OF AI WRITING, AI PAINTING, AI VIDEO, AI TOOL ASSESSMENT, EACH HEADING FOR THE PUBLICATION OF A SMALL RED BOOK。
4. Generating role and story-setting
IF YOU WANT TO DO A SHORT-TIME, ANIMATED OR A SERIES OF CONTENT, YOU CAN START WITH A LARGE-FORMULA MODEL THAT SETS PEOPLE, WORLDVIEWS AND DRAMAS。
Example:
PLEASE HELP ME DESIGN AN AICTOP VIDEO, FEATURING AN AI WHITE AND AN AI ASSISTANT. PLEASE GIVE THE PERSON SETTING, THE STYLE OF THE CONTENT, THE NAME OF THE COLUMN AND THE 10 ISSUES。
II. Picture tool
The core of the graphic tool is not "will it paint?" It's whether you can tell the picture in your head. That's the "prompt."
1. Nano Banana 2/Nano Banana Pro
official tool portal 1: (gemini.google)
Enter 2: Flow (https://labs.google/fx/tools/flow)
Nano Banana2 primary hit speed, mass and size generation
Nano Banana Pro is based on Gemini3 Pro Image, with a greater emphasis on reasoning, more realistic, real world knowledge, text rendering and high-precision visual expression。
Suitable for what: rectifications, illustrations, person-conformity pictures, product maps, clearer text designs, reference diagrams。
Strengths:
The ability to follow a strong hint, and the authenticity of the picture, is suitable for making Westerners
insufficient
The art style of "frightfulness" may be worse than Midjourney。
2. Image 2.0
official tool portal 1: (chatgpt.com)
advantage
First, it is appropriate for ordinary people to control in their own language。
YOU DON'T HAVE TO WRITE A VERY COMPLICATED PAINTING TIP, YOU CAN JUST SAY, "DO ME AN AI TOOL COVER WITH A HEADLINE AND A STYLE FOR SMALL
Second: currently the strongest image text generation and layout (details can be found in Image2.0)
insufficient
Midjourney might be better off if you're looking for a strong art style。
Sometimes a picture character is more greasy
3. Midjourney official tool portal:
https://www.midjourney.com/explore?tab=top
advantage:
First, Amigo。
Mildjourney is well placed to do a picture of what looks like high-quality, especially illustrations, concepts, film senses, fashion photography, fantasy style。
Second, style is strong。
If you want Cyberpunk, retro-film, European-American magazine, film posters, phantom worldviews, Midjourney tends to be very effective。
Third, it's for inspiration。
Brand visualization, IP image, poster orientation, scenery, etc. can all start with Midjourney。
Insufficient:
For White, the threshold is slightly higher。
Chinese text generation and complex layout are not its strong points。
Image or Nano Banana may be more appropriate if they need to be generated strictly by title, script, layout。
The dream of Seedream
official tool portal: https://jimeng.jianying.com/ai-tool/general? type=image&workspace=undefined
advantage
First, it is friendly to Chinese users。
The dream itself is one of the most popular entry points for Chinese creators, and it is appropriate to describe the image directly in Chinese. It's more like a Chinese aesthetic
Secondly, the ability to adapt is practical。
It is not just "from zero " , it is also suitable for background replacement, local modification, character maintenance, style conversion, scale-up, etc。
insufficient
The art aesthetics and styles are not as good as Midjourney。
The effect of text generation is also more general。
in general, it's "nanobanan, a youthful version of china."
III. Video tools
The currently recommended AI video generation model is: Violin 3.0, DreamSeedance 2.0 and the newly released MiniMax H3。
1. Kling AI
official entrance: https://klingai.com/app/omni/new?ac=1
Core features
Director-level control: Supports multi-censor/multi-story narratives, where a script can be produced in multiple lenses。
Strong Role Coherence: Through the Omni Reference System, assigns the appearance of the role/object and is consistent across the lens. Original audio sync: Generate video with sound, white, environmental sound, no later voice. Lens motion control: Multi-modular input (e.g., propulsion, rounding, pulling) can be specified in fine terms for camera motion and photo effects (e.g., propulsion, rounding, pulling, etc.): Supporting a dynamic transition from text or image generation and reading reference image generation. Length elasticity: can generate 3-15 seconds of video (or longer segments by connecting). Clarity burst (but expensive): The latest support is 4k straight out, which is invincible compared to AI。
It's a dream
Official entrance 1: Dreamhttps://jimeng.jianying.com/ai-tool/generate? type=video&workspace=undefined
Official entrance 2: Little Skylark
https://xyq.jianying.com/home?from_page=xiaoyunque_landing_page&tab_name=home
The comparison is more rational and real than that of Clin 3.0 Omni, Seedance 2.0. The hint is also followed to a greater extent. The only weakness is the price
Features:
2.1. Real world complexity generation
Physical authenticity: A significant improvement in the naturality, time-series consistency of human motor modelling, in strict compliance with the true world pattern。
2.2. Strong multimodular capability
Full input: Supports text, image, video, audio combination input (up to three videos, nine images, three segments of audio) director ' s reasoning: a basic director and photographic reasoning capability to plan camera sequences autonomously。
2.3. Sound video generation
Two-channel audio: Supports immersion of two-channel audio to generate multi-track output: both background sound, environment sound, role bystander, aligned with visual rhythm。
2.4. Productivity scenario applications
multiple scenes such as commercial advertising, visual effects, game animations, talk videos, etc. support 480p, 720p, 1080p resolution, with a video length of 4-15 seconds。
Mini Max H3
Official entrance: Conch AI: https://hailuoai.com/video/create
Developer API: https://platform.minimaxi.com/docs/api-reference/video-general-v2-create
MiniMaxH3 is the latest full-modular video generation model released by MiniMax. It does not generate video only by text or single picture, but by understanding the relationship between text, image, video and audio, it produces video with original sound。
Features:
3.1. Full-state reference capability
Supports text, picture, video, audio as reference input。
For example, you can refer it to the mirror of a video, to the image of the person in a picture, to the sound or atmosphere of a sound, and to a new complete video。
There are advantages to complex reference relationships, video editing, movement migration。
3.2. Original video generation
Supports original two-track audio output。
They can generate images, person voices, environmental sound and music at the same time, without the need for first-time images and then separate sound。
3.3. Strong command compliance and branding
MiniMaxH3 is more capable of following the text, brand information, electrical product displays, UI interfaces, etc。
IT'S GOOD FOR BRAND ADS, ELECTRICIANS' SHORTS, DYNAMIC POSTERS, GAME UI PRESENTATIONS, PRODUCT PROMOTIONS, ETC。
3.4. Video motion migration and multi-photo capability
Supports video-to-video action migration: a new character and scene can be migrated by reference to the person action, lens rhythm or mirror mode in a video。
The model is home-grown with a multi-photographic modelling capability and suitable for video that requires a lens narrative and more complex movement。
3.5 UP TO 2K, 15 SECONDS VIDEO
OFFICIALLY SUPPORTED VIDEO OUTPUT WITH A MAXIMUM OF 15 SECONDS, 2K RESOLUTION, AND PROVIDED 2K GENERATION BY DEFAULT。
The quality and detail are very strong and are particularly suitable for images that need to be scaled up to see, display product details or contain brand text。
3.6. High value for money
Official Mini Max gives a positioning: at 2K resolution, the single-second cost is less than one third of the mainstream model; at 768P, the cost is about half of the mainstream 720P model。
So if you want high resolution, multi-modular reference and proto-audio, MiniMaxH3 deserves to be tested。
NOTE: H3 IS A NEW MODEL THAT HAS JUST BEEN RELEASED, AND THERE IS ROOM FOR IMPROVEMENT IN THE VERY SMALL PICTURE DETAILS OF A COMPLEX SCENE. THE FUNCTIONS, PRICES AND ENTRANCES THAT ARE ACTUALLY AVAILABLE ARE BASED ON THE CURRENT PAGE OF CONCH AI。
IV. AI SYNDICATION/WORKSTREAM PLATFORM
This type of tool does not simply generate a picture or video, but rather integrates models, canvass, material, scripts, spectroscopes, batch generation, editing processes. They are suitable for brand visualization, advertising, media matrix, and course packages。
Advantage statement
ONE UPLOAD, MULTIPLE PLATFORMS: EACH TOOL IS NOT REQUIRED TO HAVE A MEETING PERSON, NOR IS THERE ANY NEED TO UPLOAD THE SAME MATERIAL REPEATEDLY. SYNDICATION PLATFORMS CAN MOBILIZE DIFFERENT AI MODELS TO GENERATE PICTURES, VIDEOS, TEXT IN A WORKFLOW。
Work-flow reuse: The community template or existing process can be used directly to quickly generate content and replace text, picture or product material。
Suitable to complete projects: branding, social media series, course packages, etc., can complete the entire process from the conceptual to the final product at one platform。
1. FlowPix
Access: FlowPix official network. (Flowpix.club)
SUITABLE FOR WHAT: RE-USE OF WORK, BRAND CONTENT, SOCIAL MEDIA IMAGES, ELECTRICIAN VISION, POSTERS, AI PHOTO/VIDEO WORKFLOW LEARNING
2. Loveart
Source: Loveart Network. (lovart.ai)
Suitable for what: design type, brand design, Logo, packaging, social media, marketing
3. LibTV
Access: LibTV official network. (liblib.tv)
Suitable for what: video creation process, script to film, professional video collaboration
4. Tapnow (like Libtv, all focused on film production)
Entry: Tapnow Network/App. (tapnow.com)
Suitable for what: visual workstream, short film creation, lens, brand vision
Operational steps and logic are consistent with Libtv
Summarize
THE MOST IMPORTANT THING FOR AI WHITE IS NOT TO LEARN ALL THE TOOLS AT ONCE, BUT TO CREATE A SIMPLE PERCEPTION:
The Big Language Model is responsible for thinking and writing, the Picture Model for visual design, the Video Model for dynamic expression
Synthetic platforms can directly bring all the above platforms together for one implementation。
The entry route can be simple:
Ideas were written in ChatGPT/Gemini/peaset, images were produced in Nano Banana/GPT Image/demersion/Midjourney, and short videos were made in Clint/Seedance/Veo. Or make a complete work directly with FlowPix/LibTV/Lovart/Tapnow。