Adobe正在测试AI agents如何能像Photoshop专业那样使用编辑工具
关键点
- Adobe Research and UC San Diego 构建了 EditAgent，一种将图像编辑任务分出分步骤的AI系统，并配对一个通用编辑器，并配有用于磨削或缩放等事务的专用工具。
- 一个标准语言模型评价平行生成的图像变体，使用计量统计和特写收成。 有了这些数据，它的业绩就超过了为工作专门培训的评价模式。
- EditAgent产生比之前的AI agents更准确的结果，但需要大约十倍的计算。 然后可以使用已核实的输出来训练更快的编辑模型。
Adobe研究人员为图像编辑建造了AI agent，将通用编辑器与专用工具相配。 一个标准语言模型使用所测量的数据比较结果，并选择最好的一个。
用于图像编辑的AI agents将复杂的请求分解为步骤并一般用同一种工具重试失败步骤，而单独的模型则通过眼球来判断输出。
Adobe Research和UC San Diego的研究人员想用EditAgent来改变这一点，根据新论文，这个系统的工作更像是Photoshop的亲信。 它为每项任务选择正确的工具，并将结果与使用硬数字的原作进行比较。
通用编辑器失败过多的任务
在经过600个图像编辑，5个专业人士盲目评分的测试中，通用编辑器Flux-2-Klein-9B未能产生可用结果42.7%的时间。 专门工具大约解决了其中三分之二的失败，特别是磨损、旋转和放大。 更多与同名编辑的重复几乎没有帮助。
研究人员还测试了agent是通过一步一步地提炼结果还是通过平行生成多个变体来做得更好的。 迭代的完善堆积出一些小错误，因为每轮都建立在已经被改变的图像上。 在78.5%的比较中，专业人员选择了从原作中产生的平行变体。 agent只需要通过一半的评价。 EditAgent基于这两个发现。
硬数字将标准模型转化为更好的判断
Editate Agent按步骤规划一个请求，对于每个步骤，将通用编辑器与匹配的专业工具相匹配。 然后，一个标准语言模型在一个单行道中比较所有变体。
对于每个变体，模型都获得一个验证记录，其中有编辑区域的特写，变化区域的变化地图，并测量出亮度和颜色的统计数据。 例如，照片应该轻轻地亮出。 数字显示通用编辑器过度曝光并转出颜色取暖。 专门工具仍然在范围之内。
研究人员在他们600版的基准测试上测试了这个。 法官模型必须从几个候选人中挑选最佳结果，他们计算出它选择一个同样被评为可用的形象的频率。 为了比较起见，他们运行了"EditScore"等模型，这些模型被精细地调整了人类对图像编辑的收视率。
没有核查记录，所测试的7个语言模型都无法匹配最佳的微调模型。 有了它，所有七个击败每个精细的模型。 Qwen3.8-27B制取时间为81.3%，而EditScore-72B制取时间为74.8%. 对于研究人员来说，这说明法官模型所看到的可能比如何培训更重要。
每次回合后，EditAgent还检查后期编辑是否意外地未完成更早的。 以报纸为例，"放入更光天化日下"的指令使之前被霜冻的窗户再次清晰。 agent抓住了冲突并修复了冲突。
EditAgent改进了开源编辑器Qwen-Image-Edit-2511和Flux-2-Klein-9B，跨越了三个基准测试并超越了另外七个AI agents. 在EdiVal 基准测试上，Qwen变体在它的第三部编辑上经过了商业模型Seedream 4.0和GPT-Image-1.5. 但agent需要的计算量大约是单一编辑器的十倍，其性能取决于哪些工具被插入。
agent 为更快的编辑生成培训数据
研究人员还使用EditAgent作为数据源。 由于agent只保留了已核实的结果，它产生了约26000个培训实例。 编辑器对这些数据进行了微调，在所有三个基准测试都表现较好。 而由FLUX直接生成的训练数据则实际上使FLUX更加糟糕。 这种权衡可以使agent的高计算成本随着时间推移而值。
Adobe在6月将AI agents添加到Photoshop,Premiere，以及其他Creative Cloud的app上，从一个公开beta开始。 用户描述自己想要什么，而agent处理单个步骤。 这些agents选取正确的工具并检查自己的出品程度，是一个直接适用于Adobe的航运产品的问题。 黑森林实验室从此取代了论文所测试的通用模型：新的Flux 3 Image Klein有56亿个参数，小于Flux 2 Klein，然而它比所有其他开放模型甚至商业Flux 2 Pro在黑森林实验室内部测试中的得分都高。
AI News Without the Hype – 由人类治愈
订阅THE DECODER免费阅读，每周AI通讯，我们独家的"AI雷达"前沿报告每年6次，全面存档访问，访问我们的评论部分。
现在订阅
本文由 AI宝库自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Adobe is testing how well AI agents can use editing tools like a Photoshop professional would
Key Points
- Adobe Research and UC San Diego built EditAgent, an AI system that breaks image editing tasks into sub-steps and pairs a general-purpose editor with specialized tools for things like sharpening or zooming.
- A standard language model evaluates image variants generated in parallel, using measured statistics and close-up crops. With this data, it outperforms evaluation models that were specifically trained for the job.
- EditAgent produces more accurate results than previous AI agents but needs about ten times the compute. The verified output can then be used to train faster editing models.
Adobe researchers built an AI agent for image editing that pairs a general-purpose editor with specialized tools. A standard language model compares results using measured data and picks the best one.
AI agents for image editing break complex requests into steps and typically retry failed steps with the same tools, while a separate model judges the output by eyeballing it.
Researchers at Adobe Research and UC San Diego want to change that with EditAgent, a system that works more like a pro in Photoshop, according to a new paper. It picks the right tool for each task and compares the result against the original using hard numbers.
General-purpose editors fail at too many tasks
In a test of 600 image edits, blind-rated by five professionals, the general-purpose editor Flux-2-Klein-9B failed to produce a usable result 42.7 percent of the time. Specialized tools solved roughly two-thirds of those failures, especially for sharpening, rotation, and zooming. More retries with the same editor barely helped.
The researchers also tested whether an agent does better by refining results step by step or by generating multiple variants in parallel. Iterative refinement piles up small errors because each round builds on an already altered image. Professionals preferred the parallel variants generated from the original in 78.5 percent of the comparisons where they picked a side. The agent needed only half as many evaluation passes. EditAgent is built on these two findings.
Hard numbers turn standard models into better judges
EditAgent plans a request in steps and, for each step, pits the general-purpose editor against matching specialized tools. A standard language model then compares all variants in a single pass.
For each variant, the model gets a verification record with a close-up of the edited area, a change map of altered regions, and measured statistics on brightness and color. In one example, a photo was supposed to be gently brightened. The numbers showed the general-purpose editor overexposed it and shifted the colors warmer. The specialized tool stayed within spec.
The researchers tested this on their 600-edit benchmark. The judge model had to pick the best result from several candidates, and they counted how often it chose an image the professionals also rated as usable. For comparison, they ran models like EditScore that were fine-tuned on human ratings of image edits.
Without the verification record, none of the seven language models tested could match the best fine-tuned model. With it, all seven beat every fine-tuned model. Qwen3.8-27B made a usable pick 81.3 percent of the time, compared to 74.8 percent for EditScore-72B. For the researchers, this suggests that what a judge model sees can matter more than how it was trained.
After each round, EditAgent also checks whether later edits accidentally undid earlier ones. In the paper's example, the instruction "let in more daylight" makes a previously frosted window clear again. The agent catches the conflict and fixes it.
EditAgent improved the open-source editors Qwen-Image-Edit-2511 and Flux-2-Klein-9B across three benchmarks and outperformed seven other AI agents. On the EdiVal benchmark, the Qwen variant in its third editing round edged past the commercial models Seedream 4.0 and GPT-Image-1.5. But the agent needs about ten times as much compute as a single editor, and its performance depends on which tools are plugged in.
The agent generates training data for faster editors
The researchers also use EditAgent as a data source. Because the agent only keeps verified results, it produced about 26,000 training examples. Editors fine-tuned on that data performed better across all three benchmarks. Training data generated directly by FLUX, on the other hand, actually made FLUX worse. That tradeoff could make the agent's high compute costs worth it over time.
Adobe added AI agents to Photoshop, Premiere, and other Creative Cloud apps in June, starting as a public beta. Users describe what they want, and the agent handles the individual steps. How well these agents pick the right tools and check their own output is a question that applies directly to Adobe's shipping products. Black Forest Labs has since replaced the general-purpose model tested in the paper: the new Flux 3 Image Klein has 5.6 billion parameters, smaller than Flux 2 Klein, yet it scores higher than all other open models and even the commercial Flux 2 Pro in Black Forest Labs' internal tests.
AI News Without the Hype – Curated by Humans
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.
Subscribe now
Auto-fetched and translated by AI Gem Box. All rights belong to the original author.