真摄影级出图!Midjourney V6 alpha测试视频 MJV6对比DALLE 3 谁优谁劣?使用Style raw参数在MJ中画出照片级图片 MJV6文本生成测试 Style raw用法解析

氪學家
1 Jan 202409:16

TLDR本期视频介绍了Midjourney(MJ)新推出的V6版本,并与DALL-E 3进行了比较。MJ V6是经过半年技术沉淀后的跨版本更新,相较于V5.2有显著提升,特别是在写实风格图片的生成上。视频中,通过使用Style raw参数,展示了V6在生成照片级图片方面的强大能力。同时,V6在文本生成方面也有所进步,能够更准确地理解和生成提示词中的文本内容。尽管DALL-E 3在语义识别上更为强大,但在写实风格图片的生成上,MJ V6的表现更为出色。视频最后鼓励观众亲自体验V6版本,感受其带来的震撼。

Takeaways

  • 📅 MJ V6版本经过半年的更新,相较于V5.2,带来了跨版本的改进。
  • 🔍 V6版本在AI绘画领域中,面临来自DALL-E 3和SD XL Turbo等新产品的竞争。
  • 🚀 V6版本更新了更精确的提示词跟随、更长的提示词支持和改善了模型的一致性。
  • 🎨 V6增强了图像的提示和混合能力,提升了小幅文本的绘制能力。
  • 📈 V6默认生成的图片分辨率是1024x1024,放大后可达到2048x2048。
  • ❌ V6测试版不支持pan方向性拓图、zoom缩放拓图、局部重绘等功能。
  • 🔑 V6对提示词更加敏感,官方建议避免使用如photorealistic、4K8K等旧提示词。
  • 🌟 使用style raw参数或降低stylize值,可以获得更加写实的风格。
  • 🖼️ V6在添加style raw参数后,生成的图片在写实感上超越了V5.2版本。
  • 📝 在MJ中生成文本需要将文本内容置于英文双引号中,V6在文本绘制上有所进步。
  • 🆚 在写实风格图片的对比中,MJ V6版本的图片在细节上优于DALL-E 3。

Q & A

  • Midjourney V6 alpha版本相较于V5.2有哪些显著的更新?

    -Midjourney V6 alpha版本相较于V5.2版本有更精确的提示词跟随、更长的提示词支持、改善了一致性和模型的知识、提升了图像的提示和混合能力、小幅文本的绘制能力增强,以及改进了放大器,包括subtle和creative模式。

  • 为什么Midjourney V6 alpha版本发布比预期晚?

    -Midjourney V6 alpha版本发布晚于预期可能是因为在这半年中AI绘画圈出现了许多技术突破和新产品,Midjourney需要时间来研发并整合这些新技术。

  • 在Midjourney V6中,如何使用Style raw参数来生成更写实的图片?

    -在Midjourney V6中,可以通过在提示词后添加Style raw参数来生成更写实的图片。例如,如果提示词是'one girl',可以将其修改为'one girl Style raw'。

  • Midjourney V6 alpha版本在文本生成方面有哪些改进?

    -Midjourney V6 alpha版本在文本生成方面支持将文本内容置于双引号中,以生成含有文本的图片,且在文本的绘制上有了显著的进步,如手写感觉的增强和元素的丰富性。

  • Midjourney V6 alpha版本与DALL-E 3在写实风格图片生成上有哪些差异?

    -Midjourney V6 alpha版本在写实风格图片生成上,尤其是在服装、皮肤纹理、毛发细节以及景深关系方面,表现得更为出色,而DALL-E 3虽然在语义识别能力上更强,但在写实感方面略逊一筹。

  • Midjourney V6 alpha版本在文本生成时需要注意哪些问题?

    -在Midjourney V6 alpha版本中生成文本时,需要确保文本内容置于双引号中,并且确认提示词后方有V6.0参数,以确保使用的是V6模型。

  • Midjourney V6 alpha版本不支持哪些功能?

    -Midjourney V6 alpha版本作为测试版,目前不支持pan方向性拓图、zoom缩放拓图、局部重绘、样式协调器和提示词反求功能。

  • Midjourney V6 alpha版本在图像分辨率上有哪些提升?

    -Midjourney V6 alpha版本默认生成的图片分辨率是1024x1024,经过放大后,图片的分辨率可以达到2048x2048。

  • 为什么Midjourney V6 alpha版本不建议使用photorealistic这样的提示词?

    -Midjourney V6 alpha版本对提示词更加敏感,官方建议避免使用photorealistic、4K8K等提示词,因为这些词在V6中被定义为垃圾提示词,可能会影响出图效果。

  • Midjourney V6 alpha版本在风格和提示方面有哪些变化?

    -Midjourney V6 alpha版本的提示方式与V5不同,对提示词更加敏感,推荐使用style raw参数或较低的stylize值来获得更加写实的风格。

  • Midjourney V6 alpha版本在官方社区中有哪些更新说明?

    -Midjourney V6 alpha版本在官方社区中的更新说明包括更精确的提示词跟随、更长的提示词支持、改善一致性和模型知识、提升图像提示和混合、小幅文本绘制能力增强以及改进放大器模式。

  • Midjourney V6 alpha版本在测试中表现出了哪些优势和不足?

    -Midjourney V6 alpha版本在测试中表现出了在写实风格图片生成上的优势,尤其是在细节和真实感方面。不足之处可能在于作为测试版,一些功能如pan方向性拓图等还不支持。

Outlines

00:00

📈 Introduction to MJ V6 and AI Art Development

The video begins with a greeting and an acknowledgment of the two-month gap since the last MJ tutorial update. The host mentions the release of MJ's new V6 version, an alpha release that marks a significant update from the previous V5.2 version. The host highlights the advancements in AI art during the six-month hiatus of MJ updates, including the release of DALL-E 3 by OpenAI, SD XL and SD XL Turbo models, and Adobe's Firefly 2. The host expresses concern over potential user loss for MJ during this period but is eager to explore the new features of MJ V6 and its potential to make a strong comeback. The video then instructs viewers on how to switch to the V6 model and suggests visiting the official MJ community for more information on the update.

05:02

🎨 Testing MJ V6's Realism and Text Generation Capabilities

The host proceeds to test MJ V6's capabilities by comparing its performance with the previous V5.2 version. They discuss the improved realism in images generated by V6, especially when using the 'style raw' parameter, which is shown to produce highly realistic images as opposed to the somewhat 'softened' look of V5.2. The host also touches on the technical aspects of using the 'style raw' parameter for better control over the output and to achieve a more realistic style, contrasting it with the previous auto-styling tendencies of MJ. A side-by-side comparison with DALL-E 3 is conducted to evaluate the realism of the images produced by both AI models. Additionally, the host tests MJ V6's ability to generate images with text, following the updated instructions for including text within quotes. The results are favorable, with MJ V6 demonstrating a clear improvement over V5.2 in both text generation and overall image quality. The video concludes with an invitation for viewers to experience the new features of MJ V6 and a thank you note for watching.

Mindmap

Keywords

💡Midjourney V6

Midjourney V6 refers to the latest version of the AI image generation software, which is in its alpha testing phase. It represents a significant update from the previous V5.2 version and is compared against other AI models like DALL-E 3 in the video. The term is central to the video's theme as it discusses the advancements and capabilities of this new version.

💡DALL-E 3

DALL-E 3 is an AI model developed by OpenAI, known for its powerful semantic understanding and image generation capabilities. It is one of the competitors compared to Midjourney V6 in the video, showcasing the advancements in AI image generation and how different models interpret and generate images from textual descriptions.

💡Style raw

The term 'Style raw' is a parameter used within Midjourney V6 to generate more photorealistic images. It is a key concept in the video as it demonstrates how the AI can be directed to produce images with a specific artistic style, and it is used to illustrate the improvements in V6 over previous versions.

💡Photorealistic

Photorealistic refers to the quality of an image appearing very similar to a photograph. In the context of the video, it is used to describe the level of detail and realism that the AI models can achieve. Despite being discouraged by Midjourney V6's official documentation, the term is tested in the video to see how it affects the output of the AI.

💡Semantic recognition

Semantic recognition is the ability of an AI to understand the meaning of words and phrases in context. It is a crucial aspect when comparing AI models like Midjourney V6 and DALL-E 3, as it determines how accurately the AI can interpret complex textual prompts and generate corresponding images.

💡Consistency and knowledge

Consistency and knowledge in the context of AI models like Midjourney V6 refer to the ability of the AI to maintain a coherent style and draw upon a broad base of knowledge when generating images. The video discusses how the V6 version has improved in these areas compared to its predecessors.

💡Text generation

Text generation is the ability of an AI to create and include textual elements within an image. The video tests this feature by asking the AI to generate images with specific text, such as 'hello world' on a sticky note, to evaluate the accuracy and style of the text output.

💡Discord

Discord, in this context, is a platform where users can access and interact with the Midjourney V6 AI model. It is mentioned in the video as a place where users can switch to the V6 model and start generating images, highlighting the community aspect of AI image generation.

💡WebUI ComfyUI and FOOOCUS

WebUI ComfyUI and FOOOCUS are user interface programs that enhance the user experience when working with AI models like Midjourney's SD XL Turbo. They are part of the advancements in the AI art ecosystem that the video discusses, showing how these tools can improve image quality and efficiency.

💡Adobe Firefly

Adobe Firefly is Adobe's AI platform, which has been updated to the second generation as mentioned in the video. It competes with other AI image generation models and is significant due to Adobe's large user base and the integration with Photoshop, making it a noteworthy player in the AI art space.

💡Community announcements

Community announcements are updates provided by the developers of Midjourney on their official community platform. They are important for users to understand the latest features and improvements in the V6 model. The video references these announcements to explain the new capabilities of the AI.

Highlights

Midjourney V6版本经过半年的沉淀,推出了全新的alpha测试版,相较于V5.2版本有了显著的更新。

V6版本在AI绘画领域中,面临来自DALL-E 3和SD XL Turbo等竞争对手的技术突破和新产品上市。

V6版本的更新包括更精确的提示词跟随、更长的提示词支持、改善一致性和模型知识、提升图像提示和混合、小幅文本绘制能力提升。

V6默认生成的图片分辨率是1024x1024,放大后可达2048x2048。

V6版本不支持pan方向性拓图、zoom缩放拓图、局部重绘、样式协调器和提示词反求功能。

V6对提示词更加敏感,官方建议避免使用photorealistic, 4K8K等提示词,推荐使用style raw参数或较低的stylize值以获得更写实的风格。

通过添加style raw参数,V6版本在写实风格图片的生成上表现出色,细节接近实拍照片。

V6版本的MJ在与DALL-E 3的比较中,在写实风格图片生成方面表现更优。

V6版本的MJ在文本生成能力上有所提升,尤其是在英文文本的准确度和多样性上。

V6版本的MJ在文本绘制上展现了丰富的元素和手写感觉。

V6版本的MJ在半年内对图片质量进行了显著优化,提供了更多的功能和改进。

V6版本的MJ在测试中展现了对写实风格图片的优化,特别是在服装、皮肤纹理、毛发细节和景深关系上。

V6版本的MJ在文本生成时,需要将文本内容置于双引号内以确保准确性。

V6版本的MJ在与DALL-E 3的比较测试中,虽然DALL-E 3在拼写准确度上略胜一筹,但V6在元素丰富性和写实感上更胜一筹。

V6版本的MJ在官方社区中有详细的更新说明,便于用户了解新功能和改进。

V6版本的MJ在测试中展现了对长文本提示词的理解能力,提升了图像的质量和一致性。

V6版本的MJ在测试中,通过style raw参数的优化,实现了更清晰的图像输出和更高的控制度。

V6版本的MJ在测试中,展现了对写实风格和文本生成的显著进步,鼓励用户亲自体验新功能。