@aschapmann · X

@thebreadmakerx · X

congrats $THREE on shipping something genuinely new in the AI x crypto space with volume slowly rotating back into SOL, projects like this are due for a serious repricing. here are the key things you need to know about - take 3 selfies and get a fully rigged animatable 3D AI agent in under 60 seconds. no plugins, no Blender, no installs. - every avatar has a brain. an LLM agent with memory, emotions, tool use and personality lives inside it. it holds conversations, remembers context across sessions, executes tasks autonomously and learns from interactions. - embed your agent anywhere in one line of code. the agent-3d web component works in WordPress, Webflow, Shopify, React, anywhere. and the real unlock is x402, the HTTP payment protocol baked into every agent endpoint. when someone interacts with your embedded agent, calls one of its skills or requests data from it, they pay per use automatically. - every capability on the platform is a payable x402 endpoint.

@kephen20936 · X

I've been playing around idly with ChatGPT since 5.1 (last year), they're now on 5.6(Sol) and have introduced Work and kept Chat as it is, I've done most of my projects on @titanforge3ds with 5.6, I started off with Work, I also use for 3D assets. It's about 20 bucks a month plus tax. Which brings my GPT Business and AI expenses to about 75USD/Month plus another $60.00 USD in Web Hosting expenses per month and $72.00 USD per year in DNS Renewals. If your PC can handle it, cook up some projects with Unreal (@UnrealEngine), Infinity Ward (@InfinityWard) or Unity (@unity), I also just saw a guy use Claude or something with Roblox AI Vibe-Coded game he made for his daughter. You have a lot options coming into it this time of the Season of 2026, but everyone other than OpenAI/ChatGPT seem to want to nickle and dime you for instant-projects (10-30minute 1-passes) rather than doing multi-stage sub-phased projects. I'm not particularly good, and don't expect to g

zxhywork · V2EX

MiniMax H3 Max 最近 MiniMax H3 Max 出来之后，搜索结果里一半在讲「 5 秒 768p 、大约 3 秒出片」，另一半还在把它和 Meta 的图模型搅在一起。我自己复盘了一圈：它是 fal 在开源 MiniMax H3 上做的速度档，不是 H3 的一个开关，也不是什么 H3 Max Image 。 手头本来就有一套生成工作流，周末就单独拆了个站出来，把 H3 Max 放在默认引擎上，同一条 prompt 还能切到 Veo 3.1 / Kling / Wan 对比。文生视频、图生视频都在一个框里，音频是跟画面一起出的，不是后配。 入口在这：[MiniMax H3 Max 在线视频生成]( https://h3maxvideo.org/) 实际用下来比较在意的几点： 1. 分辨率就是 480p / 768p ，片长 5–15 秒。要 2K 或完整参考包，还是走标准 MiniMax H3 ，别指望 Max 这一档。 2. 声音是同一次推理里出来的，提示词里把对白、环境声、拟音写清楚，比先出无声再配轨省事。 3. 积分可以白嫖一点再决定要不

FollowCloud · V2EX

Minimax h3 开源太强了，现在有很多特别实用的 lora，搞了个在线的 h3 lora 集合 我感觉 minimax h3 也算是下半年一个现象级的模型了，主要是开源，现在除了 fal 推出了超级快速的 h3 max ，开源社区还有很多 lora 因为我自己没机器，没办法本地部署，所以就整了个在线的 lora 测试网站，集合了很多现在比较不错的 lora ，比如自然人物肖像 Realism People ，运镜 Camera Motion ，趣味变形 Claymation Transformation ，复古录像质感的 VHS Tape 大家感兴趣也可以来看看，现在注册送积分，应该够体验一次，如果有好的建议让我们改进或者想继续体验，可以发邮箱沟通，可以多送点积分哈~ 首页可以用 h3 max ，https://www.maxh3.com/ lora 页面就可以选择不同的 lora 模版进行测试，也可以点击 lora 看每个 lora 的具体介绍啥的： https://www.maxh3.com/minimax-h3-loras

@otterdeckhq · X

Otterdeck's AI turns a project brief into three client-ready proposals in minutes, in any currency — then tracks opens, captures sign-off, and invoices accepted ones, so you win more work and get paid.

Hevzo — 一站式的 AI 视频和图片生成平台，集中了主流的视频和图片模型

zaczuo · Product Hunt

V2Fun Generate 3D character with 8K textures and AI motion capture

Veo3 AI Video Generator — 支持文本转视频和图像转视频，配备原生音频生成和精准唇形同步，2-5秒完成专业级视频创作。