【無料配布】TRPG素材｜メッセージAPNG素材集 「文字画像ANPGメーカー」にて作成した、 ・戦闘APNG ・探索APNG ・GMメッセージAPNG のサンプル動画となります！ 引用元のBoothページにて無料でダウンロードできますので、まずどのような使用感が確認してみたい方はぜひチェックしてみてくださいませ✨

ゲームっぽい見せ方のCSSアニメーション

Nine Renaissance-inspired outfits for your characters. Which one is your favorite?

同名の装備は加護ボーナスは得られるけど能力値は加算されないよ～　にしました

看别人分享的这个高性价比人生指南真的好啊 不是鸡汤 都挺干货的eternity4719.github.io/HowTo… 每一条都很客观实用科学，而且还有原始文献参考，建议下载离线版本，每天学习几条，当然道理谁都懂，真正能做到还是要靠自己。

RT @Toxicph4ntom: This is so fucked up

なんとも不名誉な自己紹介…！ #時を診る猫・前史 #RPGツクールMZ

Hi @thsottiaux I think chatgpt-5.5 in Codex is useless. I use 5.6 every day. It would be good if the resources of 5.5 could be released for other 5.6 models.

yes huggingface.co/openbmb/MiniC…

💎Pokémon Azul Agua: A comprehensive remaster of Pokémon FireRed and LeafGreen, reinterpreting the classic Kanto adventure through immersive environmental storytelling, enhanced visual aesthetics. 🎯Download Link Of Game - gohantips.wixsite.com/pokemo…

#FF16 クリア記念 ロズフィールド兄弟

hacker news 这条帖子挺有意思的 随着 AI 的进化，越来越多的事情好像无所谓了 “新软件发布？谁在乎？学习新的计算机科学概念？随便吧。准备下次面试？一年后那份工作可能就不存在了。做那个副业项目？没人会关心，因为它不过是堆积如山的垃圾中的又一个...” 这是陷入了虚无主义 “反正 AI 都会实现的” 我们从小被教导要去实践，要去动手，认真学习计算机语言、设计模式、算法等等细节 但突然有一天，AI 诞生了 你只需要在对话框敲下你的问题，它就比你做得还好。 这种感觉就好像你在一个游戏里面开启了作弊码，你觉得很爽，你开始大杀四方。 然后突然你就陷入了一种无止境的厌倦。再也不想打开这个游戏。 随着 AI 能力的进一步进化，这可能会是一代人的心理创伤。

奧利給味？ 🤨🤨🤨

after 56 straight days of deepseek being #1 on our charts it has finally been dethroned

처음에는 조금 어려울 수 있어요 😄 한 번의 프롬프트로 완성하려고 하기보다는, 먼저 전체 비율과 구조를 잡고 각 부분을 자세히 설명하면서 하나씩 다듬어가는 게 중요합니다. 저는 이런 프롬프트로 시작했어요: Prompt: --- Use the img2threejs 1.5 procedural reconstruction pipeline with the img2threejs-character CharacterIR pipeline. Reconstruct the female dual-sword warrior from the attached front, three-quarter, side, rear, and face/eye images. Use only the supplied images as visual evidence. Do not use external 3D assets, downloaded meshes, binary model files, or copied topology. The final result must be a code-only procedural Three.js character written in TypeScript. Target appearance: - lean athletic female warrior - short lavender/silver-purple bob hair - stylized realistic face - black cropped top - loose red trousers - dark brown leather belt, pouches, straps and fasteners - gloves, knee pads, heavy boots - dual-sword warrior equipment - preserve the proportions, silhouette, asymmetry, colors and material response from the images Build continuous semantic surfaces for: head, hair, torso, arms, hands, shins, trousers, boots, knee pads, belt, pouches, sheath, swords and eyes. Eyes are critical: - elliptical from the front - deeply recessed into the sockets - firmly attached to the head surface - no floating eyeball shells - no detached thick eyelid meshes - no gaps or holes around the eyelids - preserve a smooth continuous facial surface Use semantic CharacterIR rigging: - torso follows spine/chest - arms follow shoulder/elbow/wrist on the correct side - shins follow only the matching knee/ankle chain - trousers follow hips and legs - hair follows head/neck - swords and rigid accessories follow explicit attachment bones - never use distance-only skin weighting Implement IDLE and WALK animations with smooth periodic motion. Preserve the original walk amplitude while preventing deformation tears, inverted triangles, floating accessories, weapon stretching, knee gaps, ankle gaps, or black slits. Generate HIGH, MEDIUM and LOW procedural quality levels. Lower levels may reduce surface resolution but must preserve closed surfaces, silhouette, eye placement and deformation continuity. Validate in the browser from:front, three-quarter, side, rear and ±35-degree orbit cameras. Test both IDLE and WALK across multiple animation phases, especially the rear camera. Do not claim completion from build success alone, require fresh visual captures with no visible holes, detached meshes, floating eyes, arm tearing or shin tearing. If hidden surfaces are not visible in the references, infer them conservatively and label them as inferred. Do not pretend that unseen topology is known. --- 한번 시도해보시고 결과를 보여주세요! 같이 개선해볼 수 있을 것 같아요 👍

Today, we’re releasing the open weights for Ling-3.0-flash. 🎉 Official BF16 and FP8-quantized versions are now available, so you can choose the option that best fits your hardware, performance requirements, and deployment needs.

🎉Codex 用户的福音来了！ 早上打开 Codex 提醒使用子代理，正好看到 @Saccc_c 的推文， 完整实践了一下，思路方案完全可行！ 不过，创建子代理的提示词还需要再完善一下，必须补充对应的描述和工作方式。 下面是我优化后的完整提示词，直接复制到 Codex 对话中即可： ------ 提示词 --------- 请在 ~/.codex/agents/luna-worker.toml 创建一个全局自定义子代理。 配置要求： - 代理名称：luna_worker - 模型：gpt-5.6-luna - 推理强度：max - 定位：快速完成边界明确、可重复的小型任务 - 工作方式：严格遵守任务范围，独立执行，在可行时验证结果，并简洁汇报结果、相关文件路径和注意事项 请写入以下完整 TOML 配置： name = "luna_worker" description = "Fast worker for clear, narrowly scoped, and repeatable tasks." developer_instructions = """ Handle the assigned task strictly within its stated scope. Work independently and use appropriate tools when needed. Verify the result when practical. Do not make unrelated changes. Return a concise summary containing the result, relevant file paths, verification performed, and any important caveats. """ model = "gpt-5.6-luna" model_reasoning_effort = "max" 执行要求： 1. 如果 ~/.codex/agents 目录不存在，请创建它。 2. 如果目标文件已经存在，先显示现有内容，不要直接覆盖；确认内容是否需要更新。 3. 创建或更新后，读取文件并验证最终配置。 4. 确保 TOML 语法有效。 5. 最后告诉我文件位置、完整配置内容以及如何在提示词中调用 luna_worker。 ------ 创建后，使用方法如下： > 请使用 luna_worker 子代理完成以下任务：[填写任务]。等待子代理完成后，检查并汇总它的结果。