实测 Lovart 2.0 设计 AI 进入个人助理时代Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era


AI free token/AI免费积分

AI 生图和改图已经很强了,却仍旧救不了双 11 期间疯狂加班的外观师。

AI 出图确实快,但曾经在广告公司做外观主管的我深知,出图之外的外观灵感、沟通协作和反复改稿,才是耗时的大头:

团队得先理解客户的品牌调性,才能沿着同一个方向做下去。换成 AI,我照样要反复交代要求,项目一长还怕风格跑偏。更麻烦的是,到了下一个类似项目,上次磨合好的经验又得从头教。

但老板和客户看不到这些隐形成本,他们只会问你,都有 AI 提速了,交付速度为什么没见成倍提高?

所以,现阶段的外观 AI 让我感觉本末倒置。过去是从指挥人,现在指挥 AI,该用的时间一点没少。

我真正期待的 AI,是在把需要交代清楚之后,它能顺着逻辑自己往下推进,而不是每走一步,都需要我像监工一样在旁边重新发号施令。

Lovart 刚刚发布的 2.0 版本,就把更新的重点放在了这些生成之外的工作上。新推出的个人外观 Agent「Arti」,会持续了解使用者的审美和工作习惯,并主动参与外观过程,让每次磨合出来的经验有机会在后续项目中继续发挥作用。

实测 Lovart 2.0 设计 AI 进入个人助理时代

这让我想把过去做外观主管时的需要,放到一个完整的外观委托里再试一次:在双 11 这样的赶工情形下,Arti 到底能让我少盯多少步?

Lovart 全新版本感受入口:https://www.lovart.art/

先记住我,再帮我往下做

此前 APPSO 在实测 Lovart 时就发现,它的审美表现已经很出色,但最大的痛点在于缺乏连贯记忆。每启动一个新项目,你都得像面对一位新同事一样,把它重新「带」一遍。

而全新的 Arti,正是为了应对 AI 的失忆难题而生。

首次进入时,它就会主动引导你建立个人记忆库,后续只需用自然语言就能随时增删或修正。你甚至可以直接导入作品集,或是让它从历史项目与对话中提炼出你的审美习惯与背景偏好,作为后续所有任务的底层参考。

实测 Lovart 2.0 设计 AI 进入个人助理时代

例如,把苹果即将发售的 HomePod mini 2 思路交给 Arti 时,它会先调取你既有的记忆库开展整合思考,再针对关键细节向你主动询问更多信息。

实测 Lovart 2.0 设计 AI 进入个人助理时代

不过,在一个外观项目中,团队要做的也从来不是单向地展现自我风格。它更像是一场将过往经验与当下素材相互融通的承上启下,需要结合不同品牌的视觉调性,精准判断某种流行趋势到底适不合适当下客户,并向他们清晰传递将来的外观方向。

做外观主管时,我对这个过程深有体会。单纯搜罗素材并不够,真正耗费心力的,是将海量碎片参考梳理成有逻辑、能支撑决策的深度调研与视觉提案。

而 Arti 的浮现,就像是在团队里安插了一位经验多元的外观研究员,可以主动为你开展深度的全网调研。

在之前音箱的海报生成后,Arti 就会自动在后台启动外观调研,到 Google、Pinterest、Behance 等数十个主流外观平台抓取和项目相关的高质量参考,最终交出一份涵盖溯源链接、深度趋势分析、专业外观建议以及情绪板的报告。

实测 Lovart 2.0 设计 AI 进入个人助理时代

报告生成后,你还可以一键将报告添加回对话框,情绪板里的图片素材还能直接拖入画布,毫无缝隙地接入你后面的创作工作流。

实测 Lovart 2.0 设计 AI 进入个人助理时代

这一切完全兼容后台异步运行。即使你切出页面去处理其他工作,它也会在幕后默默探索,绝不占用你的等待时间。

但当画布上堆叠的飞机稿和思路越来越多,另一个令人头疼的难题便随之浮现:经历几轮反复改稿后,AI 很容易漏掉最初确认的需要规范,前后产出的视觉风格也难免浮现断层,很难维持统一的外观语言。

更别提在团队协作时,偶尔还会撞上同事随手拿粗糙的 AI 素材凑数交差、拉低总体质感的坑队友行为。

深谙外观师痛点的 Arti,对此拿出了一个相当懂行的解法:画布巡检。

与外观调研一样,画布巡检会在后台实时对齐最初的项目要求与外观规范,无感地做一致性核查。一旦发现偏离主题或细节遗漏的苗头,便会及时抛出预警。

实测 Lovart 2.0 设计 AI 进入个人助理时代

这就好比有一位极具品控意识的资深主美,在幕后为你随时兜底,你再也不必陷入翻来覆去给 AI 重复强调要求的循环中了。

Arti 的多人协作同样很打动我,很大程度应对了以往协作中同事反馈不及时,以及甲方「不说人话」的难题。(如果你也当过外观你肯定知道我在说什么)

团队可以在分享作品时打开「允许评论」,我可以接收修改,客户也能在外观稿的具体位置留意见。

实测 Lovart 2.0 设计 AI 进入个人助理时代

当这些外观被我或客户「打回返工」的时候,外观师也可以一键把这些评论交给 Arti 处理,不用再截图、圈位置,再向 AI 解释怎么改。

后续的修改还能在同一链接更新,随时分享客户。哪怕最后客户觉得「还是用第一版」,也能从历史版本里找回来。

实测 Lovart 2.0 设计 AI 进入个人助理时代

这种在在线文档里早已普及的打点批注与版本回溯,现在终于被 Arti 用 AI 的方式搬进了外观工作,彻底把外观师从沟通磨损中解脱了出来。

项目落地后,团队往往会借助复盘沉淀经验,让下一次的协作更进一步。但过去的 AI 却极度缺乏这种「成长性」——精心磨合过的外观资产用完就被扔在一旁,下次遇到同类需要,你仍旧得从某个赛博角落翻出旧案,重新写一堆 Prompt 把 AI 再教一遍。

Arti 的「Skill 自进化」也终于打破这种单点耗损。它能自动从重复劳动中提炼规律,将成熟的流程与交付标准直接封装成可复用的 Skill,让好不容易积累的外观经验不再用完即弃。

启动新项目时,你只需带入新素材,就能套用先前磨合出的最佳工作流,再也不必从零开始排兵布阵。

当「Skill 自进化」搭配上「记忆平台」,AI 便具备了真正意义上的自我演进能力。在先前项目里磨合出来的每一次经验、习惯甚至是手感,都会后续项目的数字资产,带来滚雪球般的协作复利。

往后,也许只需一句轻描淡写的需要,就能换来一套平台、高质感的视觉表达。

实测 Lovart 2.0 设计 AI 进入个人助理时代

前几天火出圈的「特斯拉会飞」,也能用简单的描述,快速落地成 KV + 延展视觉。

实测 Lovart 2.0 设计 AI 进入个人助理时代

数学家怀特海曾写道:「文明的进步,在于扩大笔者无需思索便能完成的关键操作的数量。」

Arti 展现出的正是这种潜质,把复杂的隐性经验,变成无需再反复干涉的本能。而且,这些总结出的技能完全透明可控,兼容随时查看、管理或关闭,不会带来额外的维护负担。

更有意思的是,Arti 的成长还带着温度。它引入了成长路径与角色设定,能在一次次协作中依据你的工作习惯解锁不同的形态与默契。毕竟,谁能拒绝在工作之余,顺便「养」一个越来越懂你的可爱 AI 宠物呢?

实测 Lovart 2.0 设计 AI 进入个人助理时代

你下班了,Arti 还能继续工作

如果说前面的能力应对的还只是外观流程里的「单点痛点」,那么 Lovart 2.0 最大的更新,是把这些能力整合起来,直接化身为一个 24 小时待命的自动化外观团队。

Lovart 把这个能力叫「外观委托」。

实测 Lovart 2.0 设计 AI 进入个人助理时代

就像给真实团队下发需要一样,你只需明确任务目标、期望的交付标准,或是附上基础素材,就能快速建立一个专属的委托任务。

实测 Lovart 2.0 设计 AI 进入个人助理时代

设定好后,Lovart 便能按照预设节奏,自动完成竞品动态监控、每日灵感收集、营销日历内容筹备等高频琐碎的工作。

为了让这种「自动化」既省心又真正安全可控,Lovart 在算力与流程管理上也下足了功夫。

你可以先试着单次试运行一次,再看成效挑选是否启动启动长期任务。而且,你也可以为任务设定积分上限,像是加装了一个「熔断器」,一旦触顶即自动停止,避免一觉醒来被天价账单背刺。

运行结束后,Lovart 会交付一份详尽的执行报告,从历史交付物到每一步的积分消耗都清晰透明。你还可以挑选是否将报告自动归档至关联项目,基于报告展开二次创作。

实测 Lovart 2.0 设计 AI 进入个人助理时代

对于经常需要应对电商大促、按固定节奏搜集参考和筹备内容的人来说,Lovart 的外观委托,直接把那些繁重冗余的重复性劳动,重构成了「流水线化」的高质工作流,让外观师得以从杂务中抽身,把更多的精力投入到创作中。

少一点重复,才能多许多创意

1960 年,J. C. R. 利克莱德在《Man-Computer Symbiosis(人机共生)》中记录了一个深刻的观察:

所谓的「思考时间」,绝大部分都被查资料、计算和绘图等前置准备所消耗,真正做出决策,往往只需短短几秒。

他预言,将来的人机协作,应该是由人类负责设定目标与评估成果,而计算机则去处理那些为判断铺路的例行琐事。

还在外观行业时,我就曾无数次想象过这样的将来,但现实却让我无法从杂务中抽身。直到 Lovart 的 Arti 到来,我仿佛看到了曙光:它不仅能主动寻求信息、听懂需要并顺畅落地,平时还能按计划默默工作,只在关键节点交由我来做最终裁决。

现在,Arti 真正将外观工作带入了 Personal Agent 时代。

实测 Lovart 2.0 设计 AI 进入个人助理时代

Personal Agent 最让人期待的,是 AI 能长期熟悉一个人的工作。放到外观里,「懂你」得落到很多重复的小细节:为什么这一页要留白,为什么那张参考只借构图,为什么一幅足够漂亮的画面,仍然会被否掉。

对于珍重自己作品的外观创作者,必然都希望自己能多花一点精力在外观本身,少花一点力气在重复的工作上。

Arti 给我留下的,是一种能持续磨合的合作关系。每一次合作的积累、人与 AI 磨合出的默契,都能沉淀为下一次高质协作的复用资产,而我作为创作者仍然有空间尝试、反悔、推翻自己。

即便等 AI 学会越来越多的外观方法,笔者仍能在作品里辨认出那个做挑选的人。

Lovart 2.0 全新版本已经向所有使用者开放,国内使用者可以访问 https://www.lovart.art/ 感受。10 月 19 日至 30 日,专业版与旗舰版订阅低至 5 折,旗舰版年包最高加赠 3 个月会员。

本文来源:https://www.ifanr.com/1684840
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。

The AI life map and the change map are already very strong, but they still can't save the cosmetician who worked crazy overtime during the Double 11 period.

AI mapping is indeed fast, but I once worked as an appearance supervisor at an advertising company. I know that appearance inspiration, communication and collaboration outside the mapping, and repeated revisions, are the time-consuming bulk:

The team must understand the brand tone of the customer before they can proceed in the same direction. In exchange for AI, I still have to repeatedly explain the requirements. The length of the project is still afraid of style deviation. Even more troublesome is that when it comes to the next similar project, the experience of the last run-in has to be taught from the beginning.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

But the boss and the customer can't see these hidden costs, they will only ask you, both have AI speed up, why hasn't the delivery speed increased exponentially?

Therefore, the appearance of AI at this stage makes me feel that the end is inverted. In the past, it was from the conductor, now it is commanding the AI, and the time it should take is not less.

The AI I really look forward to is that after explaining the need clearly, it can follow the logic to push itself down, rather than each step, I need to re-direct the command next to the supervisor like a supervisor.

Lovart has just released version 2.0, and the update focuses on work outside of these builds. The new personal look Agent “Arti” will continue to understand the user's aesthetic and work habits, and actively participate in the appearance process, so that each experience will have the opportunity to continue to play a role in subsequent projects.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

This made me want to put my past needs as a cosmetic supervisor into a complete cosmetic commission and try again: How many steps can Arti save me in a rush situation like Double 11?

Lovart new version feel portal: https://www.lovart.art/

Previously, when testing Lovart, APPSO found that its aesthetic performance was already excellent, but the biggest pain point was the lack of coherent memory. Every time you start a new project, you have to “bring” it back as you would a new colleague.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

And the new Arti was born to cope with the memory loss problem of AI.

When you first enter, it will actively guide you to build a personal memory bank, and you can add, delete or modify it at any time in natural language. You can even import a portfolio directly, or have it distill your aesthetic habits and background preferences from historical projects and conversations as the underlying reference for all subsequent tasks.

For example, when you hand over Apple's upcoming HomePod mini 2 idea to Arti, it will first pull your existing memory for integration thinking, and then proactively ask you for more information on key details.

However, in an exterior project, the team never has to show their style in one direction. It is more like a commitment to integrate past experience with current materials. It needs to combine the visual tone of different brands to accurately judge whether a certain fashion trend is suitable for customers and clearly convey the future appearance direction to them.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

As an Appearance Supervisor, I have a deep understanding of the process. Simply searching for materials is not enough. What is really laborious is to comb the massive fragment reference into an in-depth investigation and visual proposal that is logical and can support decision-making.

The emergence of Arti is like installing a multi-experienced appearance researcher in the team, who can take the initiative to conduct in-depth network-wide research for you.

After the poster of the previous speaker is generated, Arti will automatically launch the appearance survey in the background, crawl to dozens of mainstream appearance platforms such as Google, Pinterest, Behance and other high-quality references related to the project, and finally hand over a report covering traceability links, in-depth trend analysis, professional appearance suggestions, and mood boards.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

After the report is generated, you can also add the report back to the dialog box with one click, and the image elements in the mood board can be dragged directly into the canvas to seamlessly access the creative workflow behind you.

All of this is fully compatible with running asynchronously in the background. Even if you cut out the page to handle other work, it will be quietly explored behind the scenes, never taking up your waiting time.

However, as more and more airplane manuscripts and ideas are stacked on the canvas, another headache problem emerges: after several rounds of repeated revisions, it is easy for AI to miss the initial recognition of the need for specifications, and the visual style produced before and after will inevitably emerge faults, making it difficult to maintain a unified appearance language.

Not to mention that when working in a team, they occasionally run into colleagues who take rough AI materials at their fingertips to juggle and pull down the overall texture of the pit teammates.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

Arti, who was familiar with the pain points of the exteriorist, came up with a fairly proficient solution: canvas inspection.

Like the appearance survey, the canvas inspection will align the initial project requirements and appearance specifications in the background in real time, and check the consistency without feeling. Early warnings are thrown as soon as there are signs of deviations from the subject or omission of details.

This is like having a senior master with a strong sense of quality control. Behind the scenes, you can always pocket the bottom for you, and you no longer have to fall into the cycle of repeatedly emphasizing the requirements to the AI.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

Arti's multiplayer collaboration was also very impressive to me, and to a large extent responded to the lack of timely feedback from colleagues in the past collaboration, as well as the difficult problem of Party A's "not speaking". (If you've ever had an appearance, you know what I'm talking about.)

Teams can open “Allow comments” when sharing work, I can receive changes, and customers can leave comments in the specific location of the sketch.

When these appearances are "called back to work" by me or the customer, the designer can also send these comments to Arti for processing with one click, without screenshots, circle positions, and then explain to the AI how to change them.

Subsequent changes can also be updated at the same link to share the customer at any time. Even if the final customer thinks "still use the first version", they can get it back from the historical version.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

This kind of dotted annotation and version backtracking, which has long been popular in online documents, is now finally moved into the appearance work by Arti using AI, completely freeing the designer from communication wear and tear.

After the project has landed, the team will often use the review to precipitate the experience and let the next collaboration go further. However, in the past, AI was extremely lacking in this kind of "growth" - when the carefully polished appearance assets were used up, they were thrown aside. The next time you encounter a similar need, you still have to turn out the old case from a certain cyber corner and rewrite a bunch of Prompts to teach AI again.

Arti's “Skill Self-Evolution” finally breaks this single point of attrition. It can automatically refine the rules from repetitive work, and directly encapsulate mature processes and delivery standards into reusable Skills, so that the hard-won appearance experience can no longer be used up and discarded.

When launching a new project, you can simply bring in new elements and apply the best workflows from the previous ground up, instead of having to line up from scratch.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

When “Skill Self-Evolution” is paired with “Memory Platform”, AI has the ability to truly self-evolve. Every experience, habit, and even hand feeling that is grounded in a previous project will result in a follow-up project's digital assets, bringing snowball-like collaborative compounding.

In the future, it may be possible to exchange a set of visual expressions with a platform and high texture for an understated need.

The "Tesla can fly" that flashed out of the circle the other day can also be described briefly and quickly landed into KV + extended vision.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

The mathematician Whitehead once wrote: "The progress of civilization is to expand the number of key operations that can be completed without thinking."

It is this potential that Arti exhibits, transforming complex hidden experiences into instincts that no longer need to interfere repeatedly. Moreover, these summarized skills are completely transparent and controllable, compatible to view, manage or close at any time, and do not impose additional maintenance burdens.

What's more interesting is that Arti grows with temperature. It introduces growth paths and role settings to unlock different shapes and tacit understandings based on your work habits, one collaboration at a time. After all, who can refuse to “raise” a cute AI pet who understands you more and more outside of work?

If the previous capabilities only deal with the "single point of pain" in the appearance process, the biggest update of Lovart 2.0 is to integrate these capabilities into a 24-hour standby automated appearance team.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

Lovart calls this ability “Appearance Commissioning.”

Just like sending a need to a real team, you can quickly create a delegated task by clarifying mission objectives, expected delivery criteria, or attaching basic materials.

Once set up, Lovart can automatically complete high-frequency trivial tasks such as dynamic monitoring of competitors, daily inspiration collection, and preparation of marketing calendar content at a preset pace.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

In order to make this “automation” worry-free and truly safe and controllable, Lovart has also worked hard on computing power and process management.

You can try a single trial run first, and then see if the performance selection starts to launch long-term tasks. In addition, you can also set a point limit for tasks, such as adding a "fuse", which will automatically stop once it reaches the top, so as to avoid being pricked by the sky-high bill upon awakening.

At the end of the run, Lovart delivers a detailed execution report with clear and transparent points consumption from historical deliverables to each step. You can also choose whether to automatically archive the report to the associated project for secondary authoring based on the report.

For those who often need to cope with e-commerce promotion, collect reference and preparation content at a fixed pace, Lovart's appearance commission directly restructures those heavy and redundant repetitive tasks into a "pipelined" high-quality workflow, allowing the designer to withdraw from chores and put more energy into creation.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

1960, J. C. R. Liclyde recorded a profound observation in Man-Computer Symbiosis:

Most of the so-called "thinking time" is consumed by front-end preparations such as checking data, calculations, and drawing, and it often takes only a few seconds to really make a decision.

He predicted that in the future, human-computer collaboration should be responsible for setting goals and evaluating results, while computers will deal with routine trivia that paves the way for judgment.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

When I was in the appearance industry, I had imagined such a future countless times, but the reality prevented me from withdrawing from chores. Until the arrival of Lovart's Arti, I seemed to see the dawn: it can not only actively seek information, understand needs and land smoothly, but also work silently as planned, and only leave it to me to make the final decision at key nodes.

Now, Arti is really bringing appearance work into the era of Personal Agent.

The most exciting thing about Personal Agent is that AI can be familiar with a person's work for a long time. Put it in the exterior, "know you" has to fall into many repetitive small details: why this page should be left blank, why that reference only borrows compositions, why a sufficiently beautiful picture will still be rejected.

For creators who value the appearance of their work, they must want to spend more energy on the appearance itself and less effort on repetitive work.

Hands-on with Lovart 2.0: Design AI Enters the Personal Assistant Era

What Arti has left me is a partnership that can continue to run. The accumulation of cooperation and the tacit understanding between people and AI can be precipitated as a reusable asset for the next high-quality collaboration, and as a creator, I still have room to try, regret, and overthrow myself.

Even if the AI learns more and more methods of appearance, I can still identify the person who made the selection in the work.

The new version of Lovart 2.0 is now available to all users and can be enjoyed by domestic users at https://www.lovart.art/. From October 19 to 30, Professional and Ultimate subscriptions are discounted by up to 50% and the Ultimate Annual Pack comes with up to 3 additional months of membership.

This article is from: https://www.ifanr.com/1684840 For copyright issues, please write to: 13951431913#139.com or submit a review on the right.

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry