w33d | 地球上最后一个emo男孩.eth e/acc 🚀

把 Wuwa Tappo 搬到 Mac 了，现已开源！ 国服 / 国际服独立安装包，内置完整资源，不用 Wine 或虚拟机。支持 Apple Silicon（macOS 13+），两服均有 CI/CD 和官方更新检查。 非官方适配，欢迎试用反馈： github.com/Last-emo-boy/wuwa…

Everything orbits the light. My application is in. @meetmotes meetmotes.com/apply

真有这价格吧牛的

記得你是何時加入 X 的嗎？我記得！#我的X週年紀念日

Hey Tibo, I’ve been using GPT for a pretty long time. I started paying for Plus back in the GPT-3.5 days, mainly because I wanted early access to models like GPT-4 and 4o, and to new features such as the original GPTs as soon as they became available. It was incredibly powerful. I can genuinely say that, out of all the AI products I’ve used, ChatGPT has consistently given me the best overall experience — better than Claude and the others. The way I use it for programming has also changed a lot over time. I went from coding with a browser open and asking the model questions, to essentially programming through agents. My usage has kept increasing as well, and I eventually upgraded all the way to Pro 20x. But at the same time, it has been pretty disheartening to hear about claims of models being “nerfed” and usage limits being quietly reduced — even though I understand OpenAI may not agree with or acknowledge that characterization. It genuinely bothers me. I’ll probably continue paying for the service, and I may even upgrade to the $500 plan. But I really hope the models stop feeling like they are being downgraded over time. If usage limits need to be reduced, I can accept that. What I find much harder to accept is a noticeable decline in model intelligence or capability. I really miss those first few days after a new model launches — the moments when I use it and think, “Maybe AGI really is getting close.” And then, after a while, it often feels like the model gradually becomes less capable. It starts to feel almost like watching someone develop Alzheimer’s: its perception becomes duller, it misses more context, and it seems less sensitive to details that it used to understand immediately. There’s one more thing I really need: a model with genuine aesthetic judgment. I want a model that understands design, taste, visual hierarchy, creativity, and what actually looks good — so that I don’t have to switch over to Anthropic whenever I’m doing frontend design or creative work. I don’t particularly like that company, but their dominance in this area is strong enough that I still feel compelled to pay for their subscription. Hopefully, as GPT continues to improve, there will eventually come a point where I only need to subscribe to one AI service. @thsottiaux Love you guys. I really do — just want ChatGPT to keep getting better.

這算啥，Binance 認可 努力虧損中

若干年沒玩了 回來後買的第一個就是 $牛來 雖然高點80m才買的 讓好哥哥賺一點吧求求了 也不會做t我就純放著了 #牛來 #Binance

都说 Astra 建模能力挺强，于是拿了一张我特别喜欢的小卡插画，测试了一下。 然后前后折腾了大概两个半小时。看到最后这版的时候，还是有点眩晕瘫坐的 我的想法很简单，就是把这张图丢给它，看看能不能在 Blender 里把那种双人悬浮、长发被风吹开、巨剑居中、整个空间透着蓝色光的感觉做出来，按照往常的话就是后面我主要负责挑刺“太粗糙了”“剑再精致一点”“灯光还不够”“再看看原图”。它就继续改然后改渲染改灯光 人物用的是网上下载的官方模型，gpt 6 Astra 在 Blender 里接手了大剑场景搭建、姿态调整、头发和布料的轮廓、材质修正、布光、镜头，以及风流、晶体、碎片这些特效。近景建筑和剑上新增的装饰细节也是它搭出来的；远景是image 2生成的环境背景 我觉得最离谱的是那些很琐碎的部分，比如要处理模型重叠和材质问题，把原本偏硬的头发修顺一点，让人物、武器和背景之间有前后层次，再一点点把冷暖光、轮廓光、空气感和风的方向调到一起。 而且做完留下的是能继续打开、继续改的 .blend 工程。人物的骨骼、材质、灯光、镜头和场景对象都还在，哪一块不满意，后面可以接着改。 当然话又说回来，真和原图放在一起，表情、动作的自然程度、头发的飘逸感，还有插画那种特别漂亮的光影节奏，都仍然有优化空间。但我最开始给它的真的就只有一张图，后面主要是在描述我想要什么感觉。能一路推进到现在这个程度，我已经很满意了，之前我啥都不会，也算是让我掌握了新技能说是。 图 1 是原图参考，图 2 是最终渲染，图 3 是 Blender 工作过程截图。 #AI #GPT #Astra #WutheringWaves

I'm in the @Plasma One waitlist queue. Anyone have an access code to spare?

tf when u got gpt5.6 there is no need for claude max20 bye max20 tools developed with me myself I and Codex github.com/Last-emo-boy/riku…