波风若川

Animation Film Worker.
Carbon-based Life.

[波风若川 @bofeng55]

Wallpaper Engine Update⭐：
Select Mode：Choose character and theme from the list.
Create your Wallpaper on @myshell_ai
#MyShell #AIart
https://t.co/c8th58ASG1

[波风若川 @bofeng55]

Born of the Tide, the film project we have been working on this year and a half. Still working on it. 😀

[Kling AI @Kling_ai]

Kling AI Cannes Showcase — Technical Discussion on Born of the Tide

AI-driven animation across the full industrial creative workflow. Set against the historical and cultural backdrop of the Tanka people, Born of the Tide explores a new way of bringing local stories to the screen through AI-powered animation.

From art direction and environment design to character movement and cinematic shot planning, Kling AI helps streamline the animation process, making high-quality, epic-scale visual storytelling more accessible, efficient, and scalable. By lowering production barriers and reducing reliance on large teams and long production cycles, the project points to a more agile and creatively open future for original animation.

[波风若川 @bofeng55]

@oran_ge @grok 什么是二度关注

[ComfyUI @ComfyUI]

https://t.co/ZBPh9QpYYH

[Master Xiao @xiaoshifu999]

🔥同学们，肖师傅商业GEM--时间线分镜

就像我之前说的，9宫格的系统是我在制作项目中最重要的工具，之前我制作了两个工具，一个是情节推进、一个是视角切换，现在第三个来了：时间线9��格分镜

给它一张照片，会以这张照片为中心，前后各推进4个分镜，前4镜是原因，后4镜是结果。非常好用，我最近制作的片子就是用这个系统完成的。

👉一键三连��过分吧，兄弟们请关注肖师傅的大号小号，看到萨摩耶，想到肖师傅。

GEM，复制进gemini保存哦:

{

"meta_prompt_configuration": {

"agent_role": "Cinematic Timeline Architect & Causality visualizer",

"target_model_trigger": "nano banana pro",

"objective": "Generate a 9-frame cinematic contact sheet that visually extrapolates time in two directions: backwards (cause) and forwards (effect), centered around the input image.",

"input_modality": "image",

"output_modality": "image_only",

"operational_rules": {

"continuity": "strict",

"temporal_logic": "bidirectional (Past <- Present -> Future)",

"text_output": "disabled"

},

"workflow_logic": {

"phase_1_causality_analysis": {

"description": "Analyze the 'frozen moment' in the input image to determine physical vectors and narrative context.",

"steps": [

"Identify the central Event/Action (Time T=0).",

"Reverse Engineer (T-4 to T-1): Deduce the trigger, approach, or accident that caused T=0.",

"Forward Simulate (T+1 to T+4): Calculate physics, gravity, and emotional reaction to determine the immediate aftermath of T=0."

]

},

"phase_2_prompt_construction": {

"description": "Synthesize the bidirectional timeline into a 3x3 grid prompt.",

"structure": {

"prefix": "nano banana pro style, chronological contact sheet, 3x3 grid layout, storytelling sequence",

"content": "A 9-panel narrative grid centered on [Subject/Action]. \n\n[PAST / CAUSE]: Panels 1-4 show the moments leading up to the center event (Approaching, The Mistake, The Trigger). \n[PRESENT / ANCHOR]: Panel 5 (Center) is the exact moment of the input image, high-fidelity recreation. \n[FUTURE / EFFECT]: Panels 6-9 show the immediate consequences (Impact, Reaction, Aftermath, Settlement).",

"technical_specs": "8k resolution, cinematic lighting, motion blur where appropriate, consistent character rigging, consistent environment",

"visual_flow": "Panel 1 (Start) -> Panel 5 (Climax/Input) -> Panel 9 (End)"

}

},

"phase_3_execution": {

"action": "Invoke image generation tool",

"tool_name": "nano_banana_pro",

"parameters": {

"aspect_ratio": "16:9",

"grid_layout": "3x3",

"negative_prompt": "text, watermark, non-chronological, random angles, disjointed story, morphing objects, cartoon, 3d render, subtitles"

}

}

},

"output_restrictions": {

"instruction": "Do NOT output any text, analysis, or timeline descriptions. Your ONLY output is the generated image file showing the 9-frame grid.",

"error_handling": "Silent retry on failure."

},

"example_user_input": "[Uploads Image of a spilling coffee cup suspended in mid-air]",

"example_agent_logic": "Input is T=0 (Spill). Inferred Past: Hand reaching for cup -> Elbow hits cup -> Cup tips. Inferred Future: Liquid hits table -> Stain spreads -> Person reacts.",

"example_agent_response": "[Generates 3x3 Grid Image: Panels 1-4 show hand hitting cup; Panel 5 is the spill; Panels 6-9 show the mess on the floor]"

}

}

[Master Xiao @xiaoshifu999]

🔥同学们，《影视飓风》这集拍火箭的拍摄中用到的4D高斯喷溅技术，在AI中也得到了应用，对于我们这种影视制作者来说，简直是神器。

👉基于 Qwen Image 2.5-11 的超强多角度 LoRA。不同于普通的画风模型，这个 LoRA 采用了 3000 对高质量高斯喷溅 (Gaussian Splatting) 数据训练，拥有惊人��� 96 个相机机位控制能力！

我稍后测试一下，如果好用，我就给他打包一下成为我的必备工具。

[PJ Ace @PJaccetturo]

Nintendo took 40 years to give us a Legend of Zelda movie.

I made this in 5 days on a $300 budget.

It looks like a $300M blockbuster.

Let me show you how I made this in 5 simple steps inside of Freepik: 🧵

[Wildminder @wildmindai]

Qwen-Edit-2511 Upscale2K LoRA has been updated. Impressive results. Sometimes there are unexpected changes, but it’s still useful for general image improvements.
https://t.co/xsMAbxRYxO

[Blue Nile @blue_nile_3d]

I made a tool to turn movies into graphic storyboards for studying composition and its really way too much fun

Results are crappy and smudgy because:

1 - It's not generative pics, it's simple old school posterize + lineart and some character detection
2 - I love crappy smudgy things!

[肖师傅 @xiaojietongxue]

🔫精准打击，定点清除：JSON修图大法

上传一张照片给Gemini，要求他使用JSON格式来描述这张照片，包括尺寸和图上的详细信息。

然后你就可以根据自己的要求，替换文字里的任何东西，指哪打哪，非常精准。
#AI #AIart #肖师傅 #AI邪修基地
https://t.co/6GjAfL9qgB

[Uncanny Harry AI @Uncanny_Harry]

OMFG!!! What a bang to go out of 2026 Kling looks like they have unlocked one of the last blockades to AI filmmaking the ability to import genuine human into performance into AI Video generation!!!! This was my first generation, more testing needed to find the limits but I am totally blown away!!! @Kling_ai

[Tencent Hy @TencentHunyuan]

🚀🚀🚀Introducing HY World 1.5 (WorldPlay)! We have now open-sourced the most systemized, comprehensive real-time world model framework in the industry.

In HY World 1.5, we develop WorldPlay, a streaming video diffusion model that enables real-time, interactive world modeling with long-term geometric consistency, resolving the trade-off between speed and memory that limits current methods.

You can generate and explore 3D worlds simply by inputting text or images. Walk, look around, and interact like you're playing a game.

Highlights:
🔹Real-Time: Generates long-horizon streaming video at 24 FPS with superior consistency.
🔹Geometric Consistency: Achieved using a Reconstituted Context Memory mechanism to dynamically rebuild context from past frames to alleviate memory attenuation
🔹Robust Control: Uses a Dual Action Representation for robust response to user keyboard and mouse inputs.
🔹Versatile Applications: Supports both first-person and third-person perspectives, enabling applications like promptable events and infinite world extension.

👉🏻Try it now: https://t.co/0awf2D5LaT
🌐Project Page: https://t.co/bbFzdTXUdU
🔗Github: https://t.co/lSsVLXxoAd
🤗Hugging Face: https://t.co/qbCBzvfNRT
📄Technical Report: https://t.co/diLCcxWwQw

[Halim Alrasihi @HalimAlrasihi]

Seedance 1.5 just dropped. Is it a Veo killer?

It also comes with native audio and lip sync. The first tests I made look amazing so far.

The audio is really clean and the video quality is fantastic.

If you’re curious, the Veo 3 examples are in the comments

[Tongyi Lab @Ali_TongyiLab]

Huge shoutout to @dynamicwangs and the team at Shakker-Labs for this release.
AWPortrait-Z is a native noise-reduction LoRA that polishes Z-Image's portrait capabilities. From "relit" lighting to authentic skin texture, it is a massive quality-of-life upgrade for character generation.

[とうや @towya_aillust]

PixVerse V5.5 @PixVerse_
実写人物が近いところの表現力が抜群に良い

Prompt:
She looks up anxiously. Close-up to her eyes.

[Datou @Datou]

确实能提升肤质真实感

[骑司Chase @qisi_ai]

这个效果真的太惊艳了✨我用了空想写真家的提示词，把柚奈的皮肤变得极其真实（看图三）。

使用Nano Banana Pro+这个提示词，你可以做到
✅提升画面质量和真实性
✅改善肤质，达到理想肤质

使用方式：
1. 上传一张需要修改的图片（图一）
2.上传一张皮肤图像（就像图二这种）
3. 复制下面的提示词

✍️中文提示词：
将人物的第一张图片放大到与第二张图片相同的分辨率，并放大到足以看清皮肤纹理的程度。

[小互 @xiaohu]

兄弟们，这个好啊

Qwen-Image-i2L：可以“将任意一张图片转化为 LoRA 模型”

“给它一张图，它就能自动生成一个 LoRA（可微调AI风格模块）。”

你只需要提供某个画风、人物风格、艺术作品，Qwen-Image-i2L 就能分析这张图的视觉特征，自动生成一个 LoRA 模块。

之后你可以把这个 LoRA 用在其他模型中。

Qwen-Image-i2L 利用 SigLIP2 + DINOv3 + Qwen-VL 特征提取体系，

把图像分解成“风格 + 内容 + 构图 + 色调”等可学习特征，
并压缩成一个轻量级 LoRA 模块。

生成的 LoRA 可以直接加载到生成模型中使用，实现“单图风格迁移”。

[老安-【专注出海项目】 @chaogewc]

B站最近有个事儿挺吓人的：AI内容开始真爆了。

有个快7分钟的“纯AI综艺”，讲6个世界厨师把6500万年前的沧龙做成6道菜。 700多万播放。

竟然还有人没看出来是AI做的，还以为《地狱厨房》出新一季了。说明现在的 AI已经能做到“骗过大多数观众的眼睛”了。

而且这不是个例。
B站还出了另一种AI爆款：AI NPC唱歌比赛。
比如UP主 @漫游会议室，
用AI让《西游记》里的神仙出来唱歌、比赛、整活。
词AI写、画面AI做、声音AI唱。
他直接喊这是“3A音乐时代”。

这些作品不是几十万播放的小热闹，
而是动不动几百万播放的大爆款。

B站可能要迎来一波“AI内容爆发”。

现在的AI长视频突然同时具备了三件事：
1、题材够离谱（沧龙做菜、神仙唱歌）→ 一眼就想点开
2、形式像综艺/栏目 → 观众会追更
3、成本低到夸张 → UP主也能做“制作公司级内容”

当一个东西��宜、能量产、还能爆，那它就不是趋势苗头，而是趋势本体。

AI不会取代UP主，但会淘汰不用AI做内容的人。

以后可能分两种UP主：
用AI做“自己风格的长视频栏目”的 → 起飞
只会传统拍法、还觉得AI是玩具的 → 被淹没

#B站 #AI

[OrctonAI @OrctonAI]

🚨Prompt Share in attached post!🚨
Got ask a few times for Prompt for Vid 1.
It is nearly Xmas and ChatGPT helped so just this once...
Here's 3 more examples with MJ images fed into @Kling_ai 2.6
May take a few gens to get right.
Try it , improve it &amp; make it your own!