Matoto 🍅 (@matotoLab) header
Matoto 🍅 (@matotoLab) profile picture

Trying AI tools. Sharing what works, what doesn’t, and what I learn. AI 도구를 직접 써보고, 만들고, 배운 것을 나눕니다. 🇰🇷

Joined August 2021
Photo by Matoto 🍅 (@matotoLab)Photo by Matoto 🍅 (@matotoLab)Photo by Matoto 🍅 (@matotoLab)Photo by Matoto 🍅 (@matotoLab)Photo by Matoto 🍅 (@matotoLab)Photo by Matoto 🍅 (@matotoLab)Photo by Matoto 🍅 (@matotoLab)Photo by Matoto 🍅 (@matotoLab)Photo by Matoto 🍅 (@matotoLab)Photo by Matoto 🍅 (@matotoLab)
Honestly, GPT-6.2 Sol and Astra felt like trash to use on launch day. I gave them another shot a few days later, and they felt a bit better. If you bounced off them on day one, I think they’re worth another try.
18
Theo의 글을 읽어보셨으면 합니다. 저는 엔지니어가 아닙니다. GPT를 정말 좋아하고, 일상과 업무에서 잘 써오던 일반 사용자입니다. 그래서 이 글의 기술적인 내용을 전부 이해하지는 못합니다. 하지만 가장 크게 공감한 부분은 분명합니다. Opus 5.5를 더 선호하게 되는 이유가 단순한 코딩 성능이 아니라, 사용자의 의도를 이해하고 일을 끝까지 마무리하는 능력이라는 점입니다. 오늘 샘 알트먼도 모델이 똑똑한 것과 별개로, 함께 일할 때의 사용감은 더 개선해야 한다는 취지의 글을 올렸습니다. 저 같은 일반 사용자에게는 에이전트가 일을 잘하는 것만큼, 제 말을 이해하고 제가 원하는 방향으로 움직이는 것이 중요합니다. 무엇을 했는지 제가 이해할 수 있게 설명하고, 제가 판단할 수 없는 내부 문제를 다시 저에게 떠넘기지 않는 것까지요. 요즘 GPT를 쓰며 느끼는 가장 큰 아쉬움은 성능이 아닙니다. 성능이 뒤처진다고 생각하지도 않습니다. 소통의 문제입니다. 똑똑하다는 감탄보다, ‘내 의도를 알아듣고 함께 일을 끝냈다’는 경험을 더 자주 하고 싶습니다.
30
I shared my disappointment with OpenAI earlier, but there's still something I want to say. This moment reminds me of when the Codex app came out about nine months ago. Back then, Claude felt ahead to me, while GPT seemed to be falling behind. But with the arrival of Codex and improvements to the models, I watched OpenAI catch up with Claude quickly. Looking back, could today's competition and challenges also become the starting point for another leap forward? They say a crisis can be an opportunity. OpenAI certainly has the capability. I'm hoping they'll show us something soon that turns the tables again.
53
For the first time since I started using GPT, Tibo's reset and the banked reset haven't made me feel excited. I've been subscribing to the $200/month ChatGPT Pro plan for almost eight months. I know OpenAI is launching new features like Wallet and gradually improving existing ones. But I don't think those changes have reached the point where they dramatically improve people's satisfaction. I understand how difficult this work is and that it takes time. It's unfortunate that consumers don't seem to have the patience to wait that long. That includes me. Watching the competition makes the disappointment harder to ignore. For me, it's been a while since an OpenAI announcement made a really big impact. The 6.2 Sol I tried during testing was clearly better than 6.1 Sol, but compared with Claude 5.5, it didn't feel like a huge leap. Does OpenAI feel the same way, and is the model taking longer to launch because they're still improving it? Or are there other challenges we don't know about? I was hoping the 28-day reset journey would end with a major announcement. That was my expectation, not an official promise. But the longer the wait goes on, the more I wonder what's happening. Having cared about GPT for so long makes this stretch feel even more frustrating.
1
203
Tibo is almost at 1M, and he just teased a possible banked reset on Thursday. My guess: a reset could land in the next day or two. Stay tuned, friends 👀 teapaw.org/thsottiaux/status/2120…
mildly interesting but @thsottiaux is just 800 followers away from a million followers 🤯
1
1
65
Ferroflora A 16-second procedural sculpture made in Codex (6.2 Sol, creator-reported). The gold sphere represents a magnet. As it moves, the spiked metal folds into petals, then closes again.
2
2
100
Last month, I shared a diagram imagining dot coordinating different tasks. Connecting multiple local computers was below the dashed line back then. Now it is a real feature. Being able to choose the computer for each task is a step in the direction I had hoped for. I’ve updated the diagram to reflect that. What I’m curious about now is whether it can accurately track progress across tasks, check the results, and keep the work moving through to completion.
🤖 Made with AI
You can now connect multiple local computers to the same dot and choose which one it uses for a task. It can build and test your app on a dedicated machine while you keep coding on your laptop.
2
57
6.2 Sol이다!!! 〈Tide Relay〉 새벽 바다의 무인 관측소가 깨어나는 18초짜리 3D 애니메이션. 파도와 반사, 카메라, 해무를 가르는 탐조등까지 코드로 만들었다.
1
1
185
프롬프트 (재현용) 새벽 바다의 무인 관측소가 깨어나는 18초짜리 3D 애니메이션을 만들어줘. 카메라가 수면 위로 다가가면 다리 조명이 차례로 켜지고, 마지막엔 탐조등이 해무를 가르게 해줘. 젖은 콘크리트, 일렁이는 물 반사, 바람과 파도 소리까지 코드로 구현해줘.
1
1
164
6.2sol 도착이야!!! 제가 Three.js로 만든 걸어 다니는 거대 성입니다. 여섯 다리로 안개 낀 골짜기를 건넙니다.
1
2
234
이 버전을 만들 때 참고한 원글 프롬프트입니다. teapaw.org/olezhavich/status/2113…
Replying to @olezhavich
Promt: Create an interactive cinematic 3D scene in Three.js called WALKING_CASTLE. Goal: Build a dramatic cinematic 3D scene of a giant castle walking through a landscape on enormous mechanical legs. The result should feel grand, magical, and mechanically impressive. Core concept: A large medieval-style stone castle has been transformed into a colossal walking machine. Instead of standing still, it moves across a mountain valley or open landscape on multiple huge mechanical legs. The scene should communicate both fantasy and engineering spectacle. Main requirements: Use Three.js. Start animating automatically on page load. No click required. No intro screen. No blank loading phase. The first frame must already be strong. The first 2–3 seconds must show interesting motion. The result should be immediately recordable after refresh. Visual direction: The castle should be large and visually rich. Include towers, walls, bridges, windows, banners, and layered architecture. The walking system should feel massive and mechanically believable. Use multiple large legs, joints, pistons, or articulated supports. The tone should be cinematic, epic, and premium. Avoid cartooniness unless it genuinely improves the scene. The landscape should support scale: mountains, valley, fog, forest, cliffs, or rolling terrain. Animation sequence: Start with a cinematic hero shot showing the full castle. The castle is already moving or begins moving almost immediately. Show a readable walking cycle:legs lift and plant with weight joints articulate structure subtly sways dust, mist, or debris reacts under each step Add secondary environmental movement:flags or banners move light atmospheric fog drifts maybe a drawbridge shifts or mechanical parts move Maintain a continuous walking rhythm so the scene always looks alive and recordable. What to animate: synchronized leg motion joints / pistons / structural motion castle body sway dust or impact response at each footfall moving atmosphere camera movement optional small ambient details like banners or torchlight Camera: Use smooth automatic cinematic camera movement. Show the castle from angles that emphasize scale. A low-angle shot, partial orbit, and side-tracking shot are encouraged. The viewer should clearly understand that this is a castle walking on giant legs. Effects: Use tasteful cinematic polish where useful: fog drifting dust atmospheric haze god rays if appropriate subtle bloom shadows that sell scale terrain interaction particles from steps Interaction: Optional interaction is welcome, but autoplay is the priority. Optional ideas: replay walking cycle alternate camera angle free orbit mode pause/play Technical quality: Use clean structure and readable code. Organize across multiple files if useful. You may use open-source libraries, shaders, and post-processing if they improve the result. The project must run locally in the browser. Optimize reasonably. Most important goals: strong first frame obvious giant walking castle concept clear multi-leg walking animation sense of weight and scale beautiful fantasy-industrial atmosphere great result for a 10–20 second social clip Important for recording: Assume I will refresh the page and start recording immediately. The walking motion and scale should be clear from the first few seconds.
46
6.2sol 도착이야!!! 제가 Three.js로 만든 둔황 석굴의 밤입니다. 천장 벽화 속 비천이 깨어나 북두칠성으로 이어집니다.
1
1
171
이 버전을 만들 때 참고한 원글 프롬프트입니다. teapaw.org/gosailglobal/status/21…
Replying to @GoSailGlobal
用 three.js(0.160,importmap 从 jsdelivr 引入)写一个网页 3D 场景《藻井星河》:夜里的敦煌莫高窟,一座覆斗顶石窟。不用任何 3D 模型、图片和外部素材,全部用代码生成:形体用 three.js 拼,壁画用 Canvas 2D 现画成贴图。 先定坐标图再动手(单位米,y 朝上): - 崖面在 z = 0,朝东(+z),宽 200、高约 46,开五六层大小不一、上下错落的窟口;崖顶往西是鸣沙山,沙丘一道道尖脊,越远越高 - 要进的那个窟:甬道 z 0 → -5,宽 1.8、高 2.7;主室 x ±4、z -5 → -13,墙高 4.2;四披从墙顶斜收到 y = 7 的藻井框(±1.75),中间方井 ±1.05,再往上凹到 y = 7.35 - 西壁开龛(x ±1.6,y 0.8 → 3.9,深 1.6),龛里一尊坐佛,龛前一盏灯台 - 崖面右边一座九层楼:九重檐,屋角起翘,窗里透灯光 一块一块地做,每做完一块先渲染截图检查,再做下一块: 1. 纹样:千佛(小坐佛,按身光、头光、袈裟、衬里四种配色轮换,每行错一位,排出斜色带)、卷草、联珠、菱格、垂幔(三角幔加璎珞和铃)、宝相莲花、祥云、火焰纹、壸门。用莫高窟褪色后的颜色:土红、石绿、石青、白垩、赭黑。最后统一做旧:污渍、褪色、地仗剥落、细裂纹 2. 四壁和窟顶:墙面上面卷草、中间千佛、下面壸门墙裙;南北壁正中各一铺说法图;西壁龛楣画火焰纹;东壁门上画七佛。四披是梯形,满铺小千佛,上沿挂垂幔。藻井外框几圈纹样带,方井中心一朵大莲花 3. 人物:飞天用贝塞尔曲线线描平涂,高髻,长裙拖在身后,披帛,手里分别是花盘、莲花、横笛、撒花。供养人和说法图里的坐佛同样画法。龛里的佛像用车削体和球体拼,脸用球面贴图画出低垂的眼 4. 窟外:夜空渐变,西边留一点日落余晖,加银河和几千颗星。崖面网格挖出真正的门洞通进甬道,其余窟口画在贴图上;另做一张发光贴图(R 存亮度,G 存亮起的先后),着色器按时间把窟口一个个点亮 5. 光:月光从东边低空照进甬道,开阴影,再加一道加法混合的光柱;窟里一盏油灯,闪烁用几个正弦叠加,写成时间的函数;后面沿墙点亮八盏地灯。后期加泛光、暗角、暖调 6. 动的部分:四身飞天一开始是画在方井里的,醒来后离开藻井,各带两条飘带绕藻井盘旋(飘带沿飞过的路径往后拖,叠一层波动,再绕路径扭转),花瓣往下落。然后方井中心化开露出星空,飞天升上去缩成星点,正好落在北斗的斗魁四星上。北斗七星和三台用真实赤经赤纬做投影,仿敦煌星图一段段连线,旁边写朱红的「北斗」 镜头和字幕(共 60 秒,全部写成时间 t 的函数,方便逐帧出片): - 0–21 秒:从鸣沙山下远望崖面,慢慢推近;8.6 秒起窟口一个个亮起金光 - 21–37 秒:穿过甬道,灯亮,看佛龛,转身看千佛壁,抬头看四披 - 37–48 秒:正对藻井,画面慢慢旋转,飞天醒来 - 48–60 秒:穿过藻井升到星空,北斗连线,竖排片名《藻井星河》出现 - 字幕依次是:公元三六六年,僧人乐僔行至鸣沙山;忽见金光,状有千佛;遂架空凿岩,造窟一龛;此后一千年,画工在这里画了四千五百多身飞天;今夜,灯亮了,她们醒了;藏经洞里有一卷唐代星图,她们回到了北斗 页面: - 打开就播放这 60 秒,右上角能暂停、拖进度;另有「进洞看」模式,拖动转头、滚轮改视角,机位按钮有藻井、佛龛、千佛、回望甬道 - 声音用 WebAudio 现场合成:风、持续低音、五声音阶拨弦、磬,点按钮才开始 - 提供 window.renderAt(t):给一个时间就画出确定的一帧;网址带 ?render=1 时隐藏控件,用来让无头浏览器逐帧截图,出 1080p 视频 - 样式:暖黑底,片名用楷书竖排,朱砂方印做点缀,字幕像电影字幕 - 每次截图都检查:画面是不是太暗,纹样有没有变形,有没有穿模和大块黑屏
1
31
6.2sol 도착이야!!! 제가 Three.js로 만든 버전입니다. 펠리컨이 일식 앞에서 자전거로 루프를 돕니다.
1
84
이 버전을 만들 때 참고한 원글 프롬프트입니다. teapaw.org/FiniYang/status/211685…
🔥 GPT-6.2 Astra 灰度测试第一弹 有请我们的老朋友 —— 鹈鹕!!! 让他们为我们表演一个鹈鹕360°立体骑车!! #codex #gpt
1
34
6.2 Sol has arrived! After all light has faded, only time remembers the last glimmer. The End of Time — an interactive Three.js artwork that lets you rewind time. I made this in 8 minutes and 30 seconds using the exact prompt I saw in another post. Prompt in the replies.
1
83
Please create an interactive 3D webpage themed “The End of Time” using Three.js. Bring your highest level of aesthetic judgment, artistic design, and 3D programming skill to the work. Push composition, lighting and shadows, materials, particles, animation, and spatial atmosphere to their limits. Make the visuals striking and let users freely interact with the scene using the mouse. Put no limits on your creativity or technical approach. Show me the full extent of what you can do. Develop, run, and test the work yourself, and deliver a fully playable experience.
27
Wait, is it today??
Replying to @flopsy42
I said something was launching yesterday, and we got GPT-6 Cyber... Not exactly what we wanted, but hey, at least we got something, LMAO And now we have a report from The Information, which gets things right 99% of the time, saying Astra and Sol are launching today... I'm feeling confident. I just hope OpenAI doesn't disappoint us and that things go as the journalists reported ahead of time.
45
This is exactly the improvement I wanted. Our RAM is back!!
we're giving you some RAM back. local threads could grow memory usage as they became long, even without starting other processes. @btraut declared war on the problem. enjoy in equally important news, @matteing made his handwriting into a font, which i use below
37