Apple has just released the update to bring macOS Tahoe to version 26.2, and security updates to Sequoia and Sonoma to bring them to 15.7.3 and 14.8.3 respectively. The latter two should also have associated Safari updates.

The update to 26.2 is about 3.78 GB to download to an Apple silicon Mac, and 2.5 GB for an Intel Mac. Some Macs may require larger downloads, though, with some in excess of 10 GB.

Tahoe 26.2 introduces Edge Light to light your face during low-light video calls, improves Podcasts with automatic chapter generation, adds filters to the Games library, adds AirDrop codes as an additional verification with unknown contacts, enhances Freeform tables, and more. Fuller release notes are available here, and are a significant improvement in themselves.

Security release notes for Tahoe report a total of 46 vulnerabilities addressed. Among them are multiple WebKit vulnerabilities, including two that Apple believes have been exploited already “in an extremely sophisticated attack against specific targeted individuals” in earlier versions of iOS. Notes for Sequoia list 25, and those for Sonoma 21. The Safari update for Sequoia and Sonoma does address those critical vulnerabilities.

Its macOS build number is 25C56, it updates iBoot firmware to version 13822.61.10 on Apple silicon Macs and Intel firmware to 2094.40.1.0.0 (iBridge 23.16.12048.0.0,0), and brings Safari to version 26.2 (21623.1.14.11.9).

Last updated at 23:10 GMT 12 December 2025.

Because most fish aim to spend their entire lives underwater, where few artists go to paint, fish are seldom seen in paintings. That contrasts with those who try to capture fish by going fishing, an activity I have previously covered in this series in this article and a second.

Most of the aquatic creatures seen in paintings of myths, including those accompanying the god Neptune, appear to be caricatures of marine mammals including dolphins, or sea-monsters bearing no resemblance to fish.

One exception to this is Joseph Stella’s The Birth of Venus from 1922. As might be expected, his treatment is completely novel and seems to have benefited from visits to an aquarium. Aphrodite is shown at sea, in the upper part of the painting her upper body above the waterline, and below morphing into an aquatic plant underneath, where it finally merges into a helical shell. Matching the birds and flowers above the water are brightly coloured fish below.

Another interesting exception is Joachim Beuckelaer’s depiction of water in his Four Elements cycle from 1569. This shows A Fish Market with the Miraculous Draught of Fishes in the Background, the one place even landlubbers would come across fish, combined with the Gospel story in the far distance.

In the summer of 1872, as a one-off, Gustave Courbet painted an allegorical still life of The Trout, that is “hooked and bleeding from the gills”, a powerful expression of his personal feelings after being imprisoned for damage to the Vendôme Column during the Paris Commune the previous year.

JMW Turner recruits a school of fish for effect in his Slave Ship from 1840. His threatening sky and violent sea put the ship in the middle distance, silhouetted against the blood-red sky. The foreground is filled with the ghastly evidence of the slaves who were cast overboard.

Seen in amongst a feeding frenzy of fish and scavenging seabirds are hands raised from the waves in their final plea for rescue, a gruesome manacled leg, and various shackles used to restrain the slaves when in transit. Further back on the left a vague white form could represent spirits, and on the right is the thrashing tail of a sea monster.

Fish make the occasional appearance alongside legendary mermaids, as in Hans Thoma’s Three Mermaids from 1879. These mermaids are remarkably human in form, lacking fishtails, and frolic with fish under the light of the moon.

Historically the most important fish in Europe has been the humble herring. In the Middle Ages herring fisheries prospered and were the foundation of Copenhagen and Great Yarmouth, and influential in early Amsterdam. They remain strongly associated with the Netherlands and Nordic countries, where they are commonly preserved in brine (soused) or pickled.

Gabriel Metsu’s Woman Selling Herring (c 1661-62) is going from door to door with her fish, here trying to convince an old woman standing with a stick at the door of her dilapidated cottage in the Dutch Republic.

In Hans Andersen Brendekilde’s Home for Dinner from 1917, a young girl holding some fresh fish stands talking to a man with a spade.

JMW Turner toured the West Country as far as Cornwall in 1811, and the Tate Gallery has his ninety-page sketchbook recording many views of the Cornish coast from that visit. He later developed several into fine oil paintings, although it’s unclear whether this watercolour of St. Mawes, Cornwall, from about 1823, had its origins in those sketches and studies.

As with his paintings of other coastal areas, Turner shows a fishing boat coming in to a beach to land its catch, and the great activity in the open air fish market in the foreground. Behind are typical Cornish cottages stepped up from the shore to the top of the coastal cliffs, and the castles of St Mawes (closer) and Pendennis, in Falmouth (more distant, on the other side of this estuary).

Richard Dadd’s Fish Market by the Sea, from about 1860, shows an impromptu open-air fish market, run by the fishermen’s wives, to sell their husband’s catch as soon as it had been landed.

上周 Kimi K2 Thinking 发布，开源模型打败 OpenAI 和 Anthropic，让它社交媒体卷起不小的声浪，网友们都在说它厉害，我们也实测了一波，在智能体、代码和写作能力上确实进步明显。

刚刚 Kimi 团队，甚至创始人杨植麟也来了，他们在 Reddit 上举办了一场信息量爆炸的 AMA（有问必答）活动。

面对社区的犀利提问，Kimi 不仅透露了下一代模型 K3 的线索、核心技术 KDA 的细节，还毫不避讳地谈论了 460 万的成本，以及与 OpenAI 在训练成本、产品哲学上的巨大差异。

• 460 万美元这个数字不是官方的数字，具体的训练成本很难量化到多少钱
• K3 什么时候来，是看奥特曼的万亿美元数据中心什么时候建成
• K3 的技术将会继续沿用，当前效果显著的 KDA 注意力机制
• 视觉模型还需要我们去采集更多的数据，但目前已经在做了……

我们为你整理了这场 AMA 中最值得关注的几个核心焦点，来看看这家现在算是国产开源老大的 AI 实验室，是如何看待他们的模型，和未来 AI 的发展。

叫板 OpenAI，「我们有自己的节奏」

在这场 AMA 中，火药味最足的部分，大概就是 Kimi 团队对 OpenAI 的隔空回应。

最大的噱头之一：K3 什么时候来？Kimi 团队的回答非常巧妙：「在奥特曼的万亿美元数据中心建成之前。」

很明显这一方面是幽默，因为没有人知道 OpenAI 到底什么时候才能建成那个数据中心，另一方面似乎也在回应外界对于 Kimi 能用更少资源追赶 GPT-5 的赞叹。

当有网友贴脸开大，直接问 Kimi 怎么看 OpenAI 要花这么多钱在训练上时，Kimi 坦言：「我们也不知道，只有奥特曼自己才知道」，并强硬地补充道，「我们有自己的方式和节奏。」

这种自己的节奏，首先体现在产品哲学上。当被问到是否会像 OpenAI 一样发布 AI 浏览器时，团队直言 No：

我们不需要创建另一个 chromium 包装器（浏览器套壳），来构建更好的模型。

他们强调，目前的工作还是专注于模型训练，能力的体现会通过大模型助手来完成。

在训练成本和硬件上，Kimi 也展现了精打细算的一面。社区好奇 K2 的训练成本是否真的是传闻中的 460 万美元，Kimi 澄清了这个数字并不正确，但表示大部分的钱都是花在研究和实验上，很难具体量化。

至于硬件，Kimi 承认他们使用的是 H800 GPU 和 Infiniband，虽然「不如美国的顶级 GPU 好，而且数量上也不占优势」，但他们充分利用了每一张卡。

模型的个性与 AI 的垃圾味

一个好的模型，不仅要有智商，还要有个性。

很多用户喜欢 Kimi K2 Instruct 的风格，认为它「比较少的谄媚，同时又像散文一样，有洞察力且独特」。

Kimi 解释说，这是「预训练（提供知识）+ 后训练（增添风味）」共同作用的结果。不同的强化学习配方（即奖励模型的不同选择）会得到不同的风格，而他们也会有意的把模型设计为更不谄媚。

但与此同时，也有用户直言 Kimi K2 Thinking 的写作风格太「AI Slop 垃圾」，无论写什么话题，风格都太过于积极和正面，导致读起来 AI 味就是很重。他还举例子说，要 Kimi 写一些很暴力很对抗的内容，它还是把整体的风格往积极正面那边去靠近。

Kimi 团队的回答非常坦诚，他们承认这是大语言模型的常见问题，也提到现阶段的强化学习，就是会刻意地放大这种风格。

这种用户体感与测试数据的矛盾，也体现在对 Benchmark（跑分）的质疑上。有网友尖锐地提问，Kimi K2 Thinking 是不是专门针对 HLE 等跑分进行了训练，才会取得如此高分？毕竟这么高的分数，好像和他实际使用中的智能不太匹配。

对此，Kimi 团队解释说，他们在改进自主推理方面取得了一些微小的进展，这刚好让 K2 Thinking 在 HLE 上得分很高。但他们也坦诚了努力的方向，要进一步提升通用能力，以便在更多实际应用场景中和跑分一样聪明。

网友还说，你看马斯克的 Grok 因为做了很多 NSFW (非工作安全) 的工作，生成图片和视频；Kimi 完全可以利用自己的写作优势，让它完成一些 NSFW 的写作，一定能为 Kimi 带来很多用户的。

Kimi 只能笑而不语，说这是一个很好的建议。未来是否会支持 NSFW 内容，可能还需要找到一些年龄验证的方法，也需要进一步做好模型的对齐工作。

很明显，现阶段 Kimi 是不可能支持 NSFW。

核心技术揭秘：KDA、长推理与多模态

作为一家被称为「开源先锋实验室」的公司，而 Reddit 本身就是也是一个非常庞大和活跃的技术社区，Kimi 也在这次的 AMA 中，分享了大量的技术细节。

10 月底，Kimi 在《Kimi Linear: An Expressive, Efficient Attention Architecture》的论文，详细介绍了一种新型混合线性注意力架构 Kimi Linear，其核心正是 Kimi Delta Attention (KDA)。

通俗来说，注意力（Attention）就是 AI 在思考时，决定应该重点关注上下文哪些词语的机制。和常见的完全注意力和线性注意力不同，KDA (Kimi Delta Attention)，是一种更智能、更高效的注意力机制。

在这次 AMA 活动中，Kimi 也多次提到，KDA 在长序列强化学习场景中展现了性能提升，并且 KDA 相关的想法很可能在 K3 中应用。

但 Kimi 也坦言，技术是有取舍的。目前混合注意力的主要目的是节省计算成本，并不是为了更好的推理，在长输入和长输出任务上，完全注意力的表现依然是更好的。

那么，Kimi K2 Thinking 是如何做到超长推理链的呢，最多 300 个工具的思考和调用，还有网友认为甚至比 GPT-5 Pro 还要好？

Kimi 认为这取决于训练方式，他们倾向于使用相对更多的思考 token 以获得最佳结果。此外，K2 Thinking 也原生支持 INT4，这也进一步加速了推理过程。

我们在之前的 Kimi K2 Thinking 文章中也分享了 INT4 的量化训练技术，这是一种高效的量化技术（INT4 QAT），Kimi 没有训练完再压缩，而是在训练过程中，就保持了低精度运算模型。

这能带来两个巨大的优势，一个是推理速度的提升，一个是长链条的推理，不会因为训练完再进行的压缩量化，而造成逻辑崩溃。

最后，关于外界期待的视觉语言能力，Kimi 明确表示：目前正在完成这项工作。

之所以先发布纯文本模型，是因为视觉语言模型的数据获取，还有训练，都需要非常多的时间，团队的资源有限，只能优先选择一个方向。

生态、成本与开放的未来

对于开发者和普通用户关心的问题，Kimi 团队也一一作答。

为什么之前能处理 1M 上下文的模型消失了？Kimi 的回答言简意赅：「成本太高了。」而对于 256K 上下文在处理大型代码库时依然不够用的问题，团队表示未来会计划增加上下文长度。

在 API 定价上，有开发者质疑为何按「调用次数」而非 token 收费。对使用 Claude Code 等其他智能体工具进行编程的用户来说，基于 API 请求次数的计费方式，是最不可控且最不透明的。

在发送提示之前，用户根本无法明确工具将发起多少次 API 调用，或者任务将持续多长时间。

Kimi 解释说，我们用 API 调用，是为了让用户更清楚的知道费用是怎么消耗的，同时符合他们团队的成本规划，但他们也松口表示会看看是否有更好的计算方法。

当有网友提到自己公司不允许使用其他聊天助手时，Kimi 借机表达了他们的核心理念：

我们拥抱开源，因为我们相信通用人工智能应该是一个带来团结而不是分裂的追求。

而对于那个终极问题——AGI 什么时候到来？Kimi 认为 AGI 很难定义，但人们已经开始感受到这种 AGI 的氛围，更强大的模型也即将到来。

和去年疯狂打广告营销的 Kimi 不同，在这场 AMA 力，杨植麟和团队成员的回答；确实能让人感受到在国产开源，逐渐占据全球大语言模型开源市场的背景下，Kimi 也更加有底气，更明确了自己的节奏。

而这个节奏很明显，就是在这场烧钱、甚至卷太空的 AI 竞赛中，继续走开源的路，才能推动技术往前走。

#欢迎关注爱范儿官方微信公众号：爱范儿（微信号：ifanr），更多精彩内容第一时间为您奉上。

爱范儿 | 原文链接 · 查看评论 · 新浪微博

Sirens are mythical woman-like creatures with alluring voices, best-known from their appearance in Homer’s Odyssey, but also featuring in other tales including that of Jason and the Argonauts. Typically their singing lures sailors to their death, and that reputation has led them to represent anything that’s dangerously attractive. Originally they weren’t described in any physical detail, but visual representations soon envisaged them as having the upper bodies of beautiful young women, and the lower bodies and legs of birds, and that has been incorporated and elaborated in later accounts and retelling.

At the end of the year that Odysseus and his crew stayed with the sorceress Circe, she helpfully advised him that he would have to sail past the sirens, two to five creatures who lured men to their death with their singing. In preparation, Odysseus got his sailors to plug their ears with beeswax before they reached the sirens, so they couldn’t hear their song, and to bind him to the mast. He gave them strict instructions that under no circumstances, no matter what he said at the time, were they to loosen his bonds, as he would be listening to the sirens’ song.

As the group reached the sirens, Odysseus told his men to release him, but instead they bound him even more closely to the mast. Once they had passed safely from earshot of the sirens, Odysseus used his facial expression to inform his men, who then released him, and they sailed on.

William Etty’s The Sirens and Ulysses from about 1837 is one of the pioneering accounts in paint of this story from the Odyssey. His three naked sirens are all woman, one playing a lyre, another holding double pipes aloft, all three doing their best to draw the sailors from Odysseus’ ship to a shore where there are the remains of earlier victims.

Edward Poynter’s The Siren from about 1864 has Aesthetic overtones in the lyre she is playing.

Arnold Böcklin takes an unusual approach of almost dereferencing Odysseus in his painting of Sirens from 1875, although there is an approaching vessel that could be his. The two sirens filling the canvas are very human down to the waist, below which they resemble birds. One sits facing us, clearly in full voice, and highly alluring in looks. The other, her back towards us, is playing an aulos and looks rather obese, to the point of almost being comical, her right breast laid upon a flat-topped rock. At their feet are three human skulls and other bones to indicate their graver intentions.

Gustave Moreau’s The Sirens (1882) shows them as beautiful figures in a static scene, with a saturnine setting sun. There is, though, a lone sail on the horizon that hasn’t yet attracted their attention. Their lower legs turn into the writhing coils of sea serpents.

Moreau’s slightly later group portrait of The Sirens from about 1885 is more complete, with Odysseus sailing past, but its three figures are clearly all woman and no bird.

Eight Dancing Women with Bird Bodies (1886) is one of Hans Thoma’s unusual mythological paintings. The best-known women with bird bodies were the sirens, who range in number from two to five. In another painting showing the sirens trying to lure a passing ship, Thoma paints similar figures, suggesting these are intended to be sirens.

John William Waterhouse’s Ulysses and the Sirens (1891) is closer to the Homeric account, although he provides a total of seven sirens, shown as large eagle-like birds of prey with only the head and neck of beautiful women. He has added bandage wrappings around the head of each sailor to make it clear that their ears are stopped from hearing sound, a visual device that links neatly with the text. His sirens are clearly singing, particularly the one closest to the viewer, who is challenging the hearing protection of one of the sailors. Another sailor, at the stern of the ship (left of the painting), is seen clutching his ears.

Almost a decade later, Waterhouse painted this non-narrative portrait of The Siren (1900).

The Sirens (1903) marked Henrietta Rae’s return to painting narrative works featuring classical nudes. Odysseus’ ship is in the distance, as three beautiful sirens use their aulos and lyre to lure its occupants.

Late mythology suggests an unpleasant end for these sirens: Hera challenged them to a singing contest against the Muses. When the latter won, the penalty they exacted of the sirens was to have all their feathers plucked out to turn into crowns. As a result of that disgrace, the sirens turned white, fell into the sea, and formed the islands including modern Souda, on the north-west coast of Crete in the Mediterranean.

Sirens have steadily spread their presence into other paintings, particularly during the twentieth century.

Georg Janny’s fantasy painting of Sirens Bathing by the Sea from 1922 is throughly other-worldly, and there’s no trace of their bird legs.

More cryptic is Paul Nash’s Surrealist Nest of the Siren (1930), which brings together the incongruous, and hardly refers to Homer’s story. The painting is framed by brightly-painted walls with pillared decorations, perhaps ornate wainscot panelling. In the middle of these is what might be a painting, but also seems to be a three-dimensional plant trough containing sinuous shrubs. In the middle of those is a small nest, like an acorn cup.

Standing in front of this is a structure resembling a weather-vane, mounted on a turned wooden shaft. At the weather end of the vane is the faceless figure of a siren; the leeward end appears purely decorative. Three red rods appear to have detached themselves from the walling, two protruding from the plant trough, the third resting on the floor.

They even manage to sneak symbolically into other classical stories.

In Girodet’s ink and chalk drawing of The Meeting of Orestes and Hermione (c 1800), Hermione is seen at the right, her arms folded, looking coy as Orestes approaches her. The second woman, with Orestes, is presumably Hermione’s maid. This is one of a series of illustrations made by Girodet to accompany Racine’s play, and has subtleties you might expect from a great narrative artist. Visible in the gap between the figures is a table-leg in the form not of a Fury foretelling Orestes’ fate, but of a siren, implying that Hermione is luring Orestes to her. Hermione, for all her apparent coyness, has let the right shoulder-strap of her robe slip, in her enticement of Orestes. She has assumed the role of femme fatale, as portrayed by Euripides and Racine.

In more recent literature, sirens appear in the less-known second part of Goethe’s play Faust.

Margret Hofheinz-Döring is one of the few artists who has painted from this second part. With the Sirens from 1962 is her pastel painting showing the sirens among rocky inlets of the Aegean Sea, a sub-scene concluding the second act.

Apple has just released macOS 26.1 Tahoe, together with security updates for Sequoia to bring it to 15.7.2, and for Sonoma to 14.8.2. There are also separate updates for Safari in 15.7.2 and 14.8.2.

The Tahoe update has at last appeared here in Europe, and is a hefty 4.7 GB for Apple silicon Macs.

Security release notes for 26.1 report around 90 vulnerabilities have been fixed, none of which have been reported as being exploited in the wild. Listings for Sequoia give about 54, and for Sonoma about 46.

The only other release notes available so far are for enterprise here.

Details provided by Apple beyond the general ‘bug fixes and updates’ include a new tinted option for Liquid Glass, that has already been widely discussed among beta-testers, Apple Music AutoMix support over AirPlay, better FaceTime audio in low bandwidth connections, and Communication Safety and Web content filters enabled by default for existing child accounts. That seems surprisingly little to squeeze into a mere 4.7 GB, and I suspect there will have been more extensive changes.

The build number for macOS 26.1 is 25B78. Firmware in Apple silicon Macs is updated to iBoot version 13822.41.1, and Safari is version 26.1 (21622.2.11.11.9).

I will post a detailed analysis of changes tomorrow, 4 November.

Important note for those intending to update to 15.7.2 or 14.8.2 rather than Tahoe:
To be certain the correct updates will be installed, in the Also Available section of Software Update, click on the ⓘ button to the right of the Update Now button for Other Updates and select the appropriate macOS update and Safari, deselecting the Tahoe update there. That should ensure you don’t inadvertently upgrade to Tahoe.

[Last updated 06:42 4 November 2025]

By 1884, Lawrence Alma-Tadema (1836–1912) and his wife Laura (1852–1909) were exhibiting successfully, as well as raising their daughters Anna (1864-1940) and Laurense (1865-1940). Although the younger Laurense went on to become a prolific novelist and poet, Anna was taught to paint by her parents.

Anna was a precocious and brilliant painter in watercolours, and her earliest surviving works, made when she was only seventeen or eighteen, document the interior of the family home near Regent’s Park, London. Sir Lawrence Alma-Tadema’s Library in Townshend House, London (1884) is a meticulously detailed account of that room, even down to the details of its stained glass. That cannot have been this painting’s original title, though, as Anna’s father wasn’t knighted for a further fifteen years.

Anna’s small watercolour of The Drawing Room, Townshend House (1885), painted the following year, shows her improved skills at depicting surface light and texture. This painting was exhibited at the World’s Columbian Exposition in Chicago in 1893, a remarkable achievement for someone who was only eighteen at the time that it was painted.

Lawrence’s A Foregone Conclusion (1885) was commissioned as a wedding gift to the second wife of Sir Henry Tate, founder of the London gallery, Amy Hislop, by Tate himself. It’s one a series on the theme of courtship and marriage: this shows a suitor (left) bringing an engagement ring to his girlfriend ready for his proposal of marriage to her. She hides behind a marble wall with her attendant, their attitude suggesting that his proposal will be successful.

Anna progressed rapidly, and this Self-portrait was one of her early oil paintings, probably dating from around 1887, or slightly later. Its puzzling background is another part of the family home, by then in Saint John’s Wood, to which they moved in 1886. It’s believed that this too was exhibited at the World’s Columbian Exposition in Chicago, in 1893.

Lawrence’s Priestess of Apollo from about 1888 shows one of the priestesses devoted to the god of the sun, who rode the chariot of the sun across the sky each day. He signed this painting in the inscription under the arch, dedicating it to Sir Charles Hallé (1819-1895), the famous conductor who founded the Hallé Orchestra in 1858, and his second wife, the widowed violinist Wilma Neruda; they married in 1888, and this appears to have been another wedding gift.

Spring (1894) is one of Lawrence’s largest and most complex paintings, which took him four years to complete. It shows the spring celebrations in honour of the goddess Flora, the April Florialia.

Hidden in its extravaganza of flowers, marble, pretty girls and young women, are some rather ruder references. The banner at the back of the procession, hung from a balcony, contains verse addressed to Priapus, the god of fertility. This was amplified by attached lines from the contemporary poet Swinburne, which continued by referring to the budding of sexual love in the spring, although those lines remained unquoted here. Finer details on some of the instruments and other objects in the painting show couples frolicking in a state of undress, and there are herms concealed in other parts. The painting was purchased by Robert Mendelssohn-Bartholdy.

A Knock at the Door (1897, Opus 90) is Laura’s most explicit painting in terms of dates. It’s set in 1684, during the period of peace between the Second Treaty of Westminster (1674) and Treaty of Nijmegen (1678), and the crisis in relations with England arising in 1688. We see an attractive young woman, apparently checking that she is looking at her best in a mirror, presumably just before she receives a visitor. On either side there are brief glimpses of open windows.

In 1899, Lawrence was knighted, and in 1900 he was awarded the Grand Prix at the Exposition Universelle in Paris. His second wife Laura died in 1909, and the following year a memorial exhibition of her paintings was held at the Fine Art Society.

Lawrence Alma-Tadema died in 1912, and was buried in Saint Paul’s Cathedral in London, a great mark of honour for any British artist.

Anna continued to exhibit at the Royal Academy until 1928. Following her father’s death, the value of his paintings collapsed, plunging both daughters into poverty. Neither ever married, but Anna continued to make a frugal livelihood from her paintings, which all seem to have vanished.

Reference

Barrow RJ (2001) Lawrence Alma-Tadema, Phaidon Press. ISBN 978 0 7148 4358 2.

Painting often runs in the family, as so well illustrated in the Brueghels of Flanders and Brabant. This weekend I look at an artistic family you’re unlikely to have heard of, the Alma-Tademas of London, husband Lawrence, wife Laura, and daughter Anna.

Born as Lourens Alma Tadema in Dronrijp, in the north of the Netherlands, Lawrence (1836–1912) was left to draw and paint following crises in his physical and mental health when he was only 15. By the following year his skills were sufficient for him to enter the Royal Academy of Antwerp, Belgium, where he spent four years. During the later years of that training he worked as a studio assistant to one of the professors at the academy, and became fascinated by Merovingian history.

In 1858-9 he moved to work for Baron Jan August Hendrik Leys, who also acted as his mentor. Leys was an avid painter of mediaeval scenes, who was also a major influence on the early work of the French painter James Tissot.

The Education of the Children of Clovis (also known as The School of Vengeance, or The Training of Clotilde’s Sons) (1861) returns to tales from Merovingian history, showing Saint Clotilde watching her young sons being taught the royal art of axe-throwing. It’s no wonder that later one of them was to murder two of her grandchildren.

In 1862, Alma-Tadema set up his own studio, and the following year he married, honeymooning in Florence, Rome, Naples, and in the excavated ruins of nearby Pompeii.

The year 1869 was a disaster for the Alma-Tadema family. Although Lawrence’s career as a painter had been starting to flourish, his wife Pauline had not been well for a long time, and died suddenly of smallpox. His own health had faltered, but his doctors in Brussels couldn’t agree on a diagnosis. This left him with one option: to go to London to get a second opinion, which he did in December. Once there he was quickly invited to Ford Madox Brown’s house, where he met the two younger Epps girls, then aged nineteen and seventeen.

The oldest of them, Emily (c 1842-1912), was taught by the leading Pre-Raphaelite landscape painter John Brett, and married to become Mrs Emily Williams. The middle daughter, Ellen (1850-1929), was taught by Ford Madox Brown, and married the poet Sir Edmund William Gosse (1849-1928). After their husbands died, Emily and Ellen shared a studio, but neither really achieved fame in their painting.

It was the youngest daughter, born Laura Theresa Epps (1852–1909), with whom Lawrence fell in love at first sight, although he was nearly thirty-four himself. He returned to his family in Brussels, still infatuated with young Laura. Then in July 1870, the Franco-Prussian War broke out, and the whole Alma-Tadema family – Lawrence, his two young daughters, and his sister Atje, who acted as housekeeper – fled to London, where he settled in the September, marrying Laura in July of the following year.

At the age of only nineteen Laura found herself an artist’s wife, step-mother to Laurense and Anna (then seven and four years old), and trying to start her own career as a painter. He quickly became friends with the artists of the Pre-Raphaelite Movement, and his palette brightened. He enjoyed rapid commercial success, and in 1873 Queen Victoria made him an official ‘Denizen’, giving him most of the rights of citizenship.

This is his portrait of The Artist’s Wife when they married.

The Vintage Festival (1871) is one of two similar versions of the same scene, painted using the much brighter and lighter palette that Lawrence developed as he settled in England. It marks the start of his mature style, using classical Greek and Roman themes depicted in intricate detail. These details were researched extensively, sometimes taking as long as two years for a single painting. Following early criticism of his initial attempts to depict the surface properties of marble, he devoted a great deal of effort to painting marble better than any previous artist, so becoming the master of marble.

The meticulous detail in his paintings often gives the impression that these are large canvases and panels, although many are quite small. He was obsessive with what he saw as accuracy, and repeatedly repainted passages until he felt they were perfect.

This painting shows his vision of a Bacchanal being celebrated by musical Bacchantes in a private villa. Its tame nature may surprise those accustomed to thinking of such festivals as rampage, but gives scope for fine details such as the leather straps on the faces of those playing wind instruments. It also has a strong sense of Aestheticism, with its evocation of sound and music, as well as the heady aromas from burning incense.

Despite her new and demanding roles, Laura was quick to achieve recognition in her painting. Her first work to be exhibited at the Salon in Paris was in 1873, and that same year she started exhibiting at the Royal Academy in London. I don’t know if her early painting The Tea Party (date not known, Opus 7) was one of those, but I believe it shows her step-daughter Laurense playing with her dolls.

Lawrence was also a skilled painter of portraits; his most touching, though, must be This is Our Corner (1873), a double portrait of his daughters Anna (1864-1940) in front, and Laurense (1865-1940) behind. They both remained unmarried through adult life: Anna became an accomplished artist, and Laurense a prolific novelist and poet.

Sunny Days (1874) may come as a surprise to those who know Lawrence for his Salon-style works, but during the 1870s he started painting small landscapes, typically of his family when they were on holiday. This shows his daughter Anna in what could be seen as the closest that he came to Impressionism, and was perhaps modelled after one of Monet’s paintings of the time. On this occasion, the landscape is only in Surrey, but he also painted these when in mainland Europe.

Laura’s World of Dreams from 1876 introduces a theme that she was to develop in her mature works: the reflection and transmission of images in mirrors and windows, adding depth to what would otherwise be a shallow view. Here a nurse/nanny (or possibly mother) has fallen asleep, exhausted, on a large illustrated family Bible, which is open at the start of the book of Amos.

Laura was one of only two British women artists to have work accepted for the Exposition Universelle in Paris in 1878.

A Looking out o’Window, Sunshine (?1881, Opus 81) is by far the most outstanding of Laura’s paintings that I have found, with its captivating portrayal of childhood, radiant lightness, and its play of reflections on the glass of the window. It’s also composed to reverse the normal balance in brightness between indoors and out: in reality, of course, outdoor light levels would have been much greater than those indoors, but Laura has managed to ‘cheat’ that by her choice of light colours in the interior wood, clothing, and chair.

Reference

Barrow RJ (2001) Lawrence Alma-Tadema, Phaidon Press. ISBN 978 0 7148 4358 2.

Apple has just released macOS 26.0.1 Tahoe, which fixes the problem upgrading to 26.0 on Mac Studio M3 Ultra models, and apparently fixes other urgent bugs.

For Apple silicon, the update is a 1.76 GB download.

Tahoe 26.0.1 fixes a single vulnerability, although Apple doesn’t report that it’s already being exploited. The same is also fixed in Sequoia 15.7.1, and in Sonoma 14.8.1.

macOS 26.0.1 has build number of 25A362, Safari version 26.0.1 (21622.1.22.11.15), and a Darwin Kernel version of 25.0.0. There has been no change in iBoot firmware, which remains at 13822.1.2.

As Apple hasn’t been forthcoming about what else has changed, here’s my list:

• Passwords app has gone from version 2.0 to 2.0.1, suggesting it has at least one significant bug fixed.
• AppKit framework has had an increment in build number, also suggesting bug fixes.
• CoreText framework likewise, with bug fixes for a higher build number, possibly related to the fixed vulnerability in font handling.
• Security framework has a substantial increase in build number, implying bug fixes there as well.

Otherwise, remarkably little has changed.

Updated 1910 29 September 2025.

这是 3DFiti 这个艺术创作项目的第二个作品，它包含了四套模型。

分别是巴索罗缪 · 大熊在《海贼王》中的四个关键场景，场景和人物各有四种颜色。

观看地址：

https://youtu.be/DZ-Ho09J0U4

https://www.bilibili.com/video/BV1fe3TzaEqz/

这次的想法来源于，我想把多个模型藏在一片区域内的不同地方，在实地勘景的时候，我意识到这些地点可以借由《海贼王》中大熊这个人物的故事串联起来，于是将两者结合，做出了这件作品。

我参与 3DFiti 这个艺术项目，旨在通过 3D 扫描和 3D 打印技术，使我们生活的周边环境增加一些趣味性和艺术性。在人类学家项飙（英国牛津大学社会人类学教授，德国马克斯普朗克社会人类学研究所所长）所描述的那个「附近」的概念之中延展出来，一切我可以到达的地方，只要有我感兴趣的公共区域内「空缺」的空间，就有可能成为这个艺术项目的创作场所。

这次的四个模型分别对应四个「奖品」：

• 可以来参观我的工作室 x1
• 可以在线下设计咨询（免费）2小时 x1
• 可以在线上设计咨询（免费）1小时 x2

只有找到模型并按照二维码所示联系我的人，才可以获得「奖品」。

这是一次 3D 打印的创作，也是一次 Citywalk 路线的开发，还是一次与人连接的尝试。谁会找到这些模型呢？

Sharing: 3D modeling & rendering flow in Figma

Background

Current situation

I have been working for the current company for half a year. I have a clear picture of the working pattern: We are a small design team, but each member has to be responsible for both the UI and UX design for many internal systems.

Some of the design requirements are not complex at all, but the stakeholders want to make the interface aesthetically beautiful and different from other internal systems. In contrast, other demand sides hope designers can provide a clear and understandable solution to tackle the complex interaction flow.

The previous one was much more difficult for me since I was interviewed for the senior UX designer position, and I had not expected to address these design requirements, which may have occurred considerable times.

So it is time-consuming to design a portal website or refine a running internal system since I have to try multiple design solutions to ensure it will be aesthetically beautiful and outstanding.

Ideation

I wonder if 3D pictures could be a great approach to suit this type of design requirements. So I researched external products from large firms, trying to know how they use 3D pictures in their SaaS products.

After collecting and investing in multiple products, I found the patterns:

1. 3D pictures are most likely used in landing pages, login pages, portal pages, and the like;
2. Admin panels and dashboards do not use 3D pictures frequently, but they would appear in entries, feedback, and background decoration.

Conclusion

Based on the pattern above, real working conditions, and small-sized tries, I summarized the first conclusion:

1. 3D pictures are not frequently used on dashboard pages and system admin. However, using 3D pictures could improve the variety of pages. So it is not worth it to spend so much time on it.
2. 3D pictures can effectively meet the expectations of demand sides when they want to make their dashboard and admin pages distinguish from other similar sites.

Difficulties

Since I don’t have skills in 3D design, and my working strength is too high to spend much time learning, modeling, and rendering delicate 3D pictures, I just simply searched 3D visual elements on the Internet and then applied them to my deliverable.

Initially, this approach was workable and can help me complete my work quickly. However, drawbacks are uncovered after multiple uses:

1. Difficulties of matching business contexts. 3D pictures on the Internet support common scenarios well but can’t always precisely describe certain business scenarios.
2. Lack of series of an element. Although we finally found a perfect 3D picture, it is hard to reach other similar elements from the same series, resulting in poor scalability.
3. Risks of infringement. Using 3D pictures on the Internet might cause infringement, so applying it directly consists of potential risks.
4. Inconsistent quality. Free elements are not guaranteed quality and sometimes fail to meet the standards of our design team.

Here, I draw the second conclusion:

Searching for 3D pictures on the Internet has limited effectiveness, so we can’t fully rely on this approach.

After drawing this conclusion, I feel horrible: does it mean that I should strive to learn skills in 3D design? My work strength doesn’t allow me to learn professional 3D software like Cinema 4D or Blender.

After thorough consideration, I think the core demand for me is not to learn 3D software, rather, I should have the ability to gain 3D pictures. Apart from that, the whole process should consist of these features that suit my workflow well:

1. Quick render. Since it is not worth it to spend much time on it during my work time.
2. Consistent style. Ensuring our design has great scalability.
3. Great customizability. It means that I can fully control elements on 3D pictures, ensuring outputs highly represent the business requirements and contexts.

By chance, I know there is a plugin in Figma called “Vector to 3D”. By learning outputs published in the community, it seems that this plugin could meet the features above well, improving my work effectiveness and quality. So I asked my team leader if I could spend some time discovering this plugin and sharing my experiences with team members, and she approved.

Therefore, this post is to record my internal sharing: My experience of how to build 3D modeling & rendering flow in Figma that improves our design effectiveness and quality.

Work with Vector to 3D

Introducing the plugin

This is a paid plugin, we can search for and install it in the Figma community.

The control panel is comprised of three parts: Global, Object, and Animation settings. Here I will simply introduce the interface of the first two parts, ensuring all readers learn what this plugin can make.

For the Global setting, we can adjust the cameras, and background on the “General” panel; apart from that, we can set the render quality and details on the following panel called “Render”; and then we can control lights on the next panel named “Light and Shadow”, this panel supports varies functions of lights, like the number, size, position, strength, color and so forth.

For the Object setting, we can make a 3D model by extruding, inflating, or revolving, and then increase the detail of the model by setting “Mesh Quality”. Moreover, this plugin also supports material settings, like roughness, metallic, transmission, and the like.

These features are more than sufficient to handle nearly all of my design tasks.

Make a template

In this part, I will introduce two functions of this plugin: Save Preset and Template Preset.

Saving preset allows us to memorize the current view, modeling, and material settings of the frame. We can reload all the settings by the next open, even sharing with friends.

Templating Preset is not only the masterpiece of connecting the plugin with Figma but also a key factor in effectively generating 3D pictures.

Once we built a template, the plugin would memorize all the raw object preset, next step, we can replace the raw object with anything else, and succeed the raw object preset to it.

This approach is highly convenient for us when producing 3D elements. We just have to create a template, and then we replace various shapes, finally, we can produce endless 3D elements that fit our actual needs well.

How it works: First, we need to set the frame as a component and make the replaceable element that we want to replace with a new one a child component. (In the following example, I set the square bottom as a replaceable element.)

Second, we duplicate the whole component and tune settings in the plugin, then save it. (In the following example, I make a glass ball and a square bottom base with a metallic surface)

Third, we need to create a bunch of shapes that are to replace the replaceable element, making them an individual component and ensuring their layer number, order, and name are the same as the replaceable element.

Last, we can hold the keys “cmd” + “option”, drag a new shape component to replace the replaceable element, and then reload the model again. Now we can see the new shape is applied to the same presets.

Now we successfully made a template, we can produce endless 3D elements by creating and inserting new shapes.

Actual practice

Here, I will share a case that is already used multiple times in my daily workflow.

In this template, the shape of icons and background are replaceable, generating endless pictures and fitting nearly all of my design tasks. So it can be a resource library, empowering other team members.

Meanwhile, I created a color pattern, which not only can be used to represent states like success, fail, warning, and the like but also can be applied to the system with different color themes, ensuring our outputs are highly scalability.

Benefits

In the post above, we learned the interface of Vector to 3D, how to make a template, and how I use this flow in actual workflow.

We are familiar with the 3D design workflow in Figma. To conclude this post, I’d like to summarize the key advantages of this process.

1. Low learning costs

We can simply recognize that the plugin “Vector to 3D” is a lite version of 3D software, simplifying complex functions from traditional 3D software and keeping essential functions.

2. Full design process within Figma

This workflow allows us to produce great 3D elements that don’t have to leave Figma. Apart from that, we can simply tune colors in Figma, which doesn’t like the traditional workflow that always switches from software to software.

3. High scalability

One of the core features of Vector to 3D is that makes up models from vector shapes, by taking advantage of this, we can efficiently produce 3D pictures that effectively meet our needs and align well with business contexts.

4. Consistent styles

The plugin Vector to 3D can coordinate with the components of Figma, not only remembering global presets like lights and cameras; but also allowing replacements to succeed presets like materials and position from previous elements. By templating these presets, we can endlessly produce 3D pictures with the same style.

This workflow is not only successfully used in my daily work but also influences my colleagues. After this sharing, 50% of my team members purchased and used this plugin in their design tasks.

Solution for non-Figma users

If you or your team are not using Figma for work currently, or you are not a designer but just simply want to try the flow from SVG shapes to 3D models, you can visit the online version of Vector to 3D: https://www.meimu.design/vector-to-3d/

Sharing: 3D modeling & rendering flow on Figma was originally published in Bootcamp on Medium, where people are continuing the conversation by highlighting and responding to this story.

其实有点出乎我意料，除了 Apple Vision Pro 那期，最受欢迎的居然是聊《九龙城寨》和《暗恋桃花源》的这两期。而且，刚发的《谈判专家》这期的收听量也在稳定上涨。聊 AI 那期尽管内容很多，但收听量比这些都少得多。

从博客后台数据能看到，最近一周的主要收听量中，三分之二都来自这三期聊戏聊剧的节目。

我原本以为，收听我节目的主要人群，是过去在知乎和 B站看我讲设计的读者和观众。

这么看下来，我有一个猜测：收听我播客的人群当中，有很大一部分比例，可能是此前并不认识我的路人，他们对科技类话题的兴趣，没有对娱乐类型的话题高。

挺好的，这也挺符合我最初对播客的预期，这样我就可以不用老聊设计和产品了！

荒野楼阁 WildloG：https://suithink.me/zlink/podcast/

小宇宙： https://suithink.podcast.xyz

Spotify：荒野楼阁 WildloG

YouTube：荒野楼阁 WildloG

Apple Podcast 在中国大陆地区目前只支持通过 URL 订阅：

https://suithink.me/category/podcast/feed/