I Asked Photoshop AI to Zoom Out Infinitely. Here’s What Happened.

Joe Scott
15 Apr 202422:36

TLDRThe video explores the capabilities of generative AI tools integrated into the 2024 update of Photoshop, specifically the generative expand and generative fill features. The creator, while not a Photoshop expert, demonstrates how these tools can be used to expand images and fill in missing areas based on surrounding content, which can be particularly useful for creating thumbnails for videos. Intrigued by the potential, the video documents an experiment where the generative expand tool is used repeatedly to zoom out infinitely from a single image, resulting in a fascinating and somewhat surreal visual journey. The video also touches on the ethical considerations of AI image generation, the impact on artists and photographers, and the potential for AI to increase productivity in various fields. Finally, the creator discusses the challenges of editing the AI-generated content into a seamless video and the potential for AI to subtly alter our perception of reality.

Takeaways

  • 🎨 Photoshop's 2024 update includes generative AI tools like Generative Phill and Generative Expand, which can fill in missing parts of an image based on surrounding context.
  • πŸ“ˆ The author, while not a Photoshop expert, has significant experience using the software for thumbnail creation and advertising work, rating their skills at around a 6 or 7.
  • πŸ“ The generative expand tool simplifies the process of creating 16:9 images from vertical photos, which are challenging to work with due to their aspect ratio.
  • πŸ” Content-aware fill was a previous solution for filling in image areas, but generative expand offers a more seamless and efficient approach.
  • 🧠 AI's role as a time-saving tool is highlighted, with the potential to automate and predict image content based on surrounding visual cues.
  • πŸ”„ The experiment of continuously using generative expand on a single image led to unexpected and abstract results, illustrating the tool's creative potential.
  • πŸ–ΌοΈ The generative AI tools within Photoshop, powered by Adobe's Firefly model, are trained on images from the Adobe Stock catalog, which addresses ethical concerns regarding image rights.
  • 🚧 There are concerns about the impact of AI-generated images on the job market for artists and photographers, as well as the potential for misuse in manipulating visual content.
  • πŸ€” The author ponders the self-reinforcing nature of AI training, questioning if the Firefly model is training on AI-generated images, leading to a homogenization of outputs.
  • πŸ“š The video concludes with a discussion on productivity and the integration of AI into everyday tools, suggesting that the technological singularity is already happening through incremental AI integration.
  • πŸ“± An app called Imprint is mentioned as a solution for digesting book insights in a time-efficient manner, offering interactive courses that distill key concepts into quick lessons.

Q & A

  • What is the main topic discussed in the video?

    -The main topic discussed in the video is the exploration of generative AI tools within Photoshop, specifically the generative expand feature, and the experiment of using it repeatedly to create an 'infinite zoom' effect.

  • What is the generative expand tool in Photoshop?

    -The generative expand tool in Photoshop is an AI-powered feature that allows users to expand an image from a selected subject outward, intelligently filling in the surrounding areas based on the content around the edges of the image.

  • How many times did the author perform the generative expand on the image?

    -The author performed the generative expand on the image around 100 times to observe the changes and progression of the image.

  • What was the initial observation after performing the generative expand multiple times?

    -The initial observation was that the image started to look like a melting ceiling and the arm appeared elongated, indicating that the AI was making unusual choices in filling the expanded areas.

  • Why did the author reduce the image size back to 4K resolution?

    -The author reduced the image size back to 4K resolution because the size of the image was becoming unsustainably large with each expansion, which could lead to performance issues.

  • What is the ethical consideration mentioned regarding AI image generation?

    -The ethical consideration mentioned is the use of copyrighted work from artists, photographers, and illustrators without their permission in training AI models, potentially leading to lost revenue and job opportunities for these creative professionals.

  • What is Adobe's response to the ethical concerns of AI image generation?

    -Adobe's response is the introduction of Firefly, their generative AI model, which is trained using only the images in the Adobe Stock catalog, ensuring that the artists whose work is used have already received compensation.

  • What challenges did the author face when trying to create a smooth transition between the expanded images?

    -The author faced challenges with the velocity of scaling and the varying speeds at which the image expanded at different percentages of scale, making it difficult to time the transitions accurately.

  • How did the author attempt to solve the problem of uneven transitions between images?

    -The author initially attempted to solve the problem by adjusting the scaling velocity in Adobe Premiere and using Adobe After Effects for more precise control over the Z-space. However, this proved to be time-consuming and still resulted in some stutter effects.

  • What alternative tool did the author consider to simplify the 'infinite zoom' process?

    -The author considered using an app from a company called VideoLeap, which offers an 'infinite zoom' feature that could potentially simplify and speed up the process.

  • What is the author's final takeaway from the experiment and the role of AI tools in productivity?

    -The author's final takeaway is that AI tools are meant to save time and increase productivity, integrating gradually into existing programs and processes. However, there are potential dangers, such as the ethical concerns of AI image generation and the risk of AI-generated content replacing human creativity.

Outlines

00:00

🎨 Photoshop's Generative AI Tools

The video discusses the integration of generative AI tools in the 2024 update of Photoshop. These tools include generative Phill, which allows users to replace selected parts of a photo with desired objects, and generative expand, which fills in the space around a subject. The speaker shares their curiosity about the potential for weird results when using these tools repeatedly and their extensive experience with Photoshop, primarily in creating thumbnails for their videos. They also touch upon the challenges of working with vertical photos in a 16:9 format and how the generative expand tool simplifies this process.

05:03

🧠 AI's Impact on Creativity and Originality

The speaker explores the implications of AI-generated images, particularly the ethical considerations when these tools use copyrighted work without permission. They mention the advent of Dolly from OpenAI, which combined natural language processing with image generation. The discussion also covers Adobe's Firefly, a generative AI model trained only on images from the Adobe Stock catalog, which is presented as a more ethical alternative. The video also includes an experiment where the speaker uses generative expand repeatedly to create an 'infinite zoom' effect, pondering the potential for AI to change the landscape of image creation and the concerns regarding the authenticity and originality of artwork.

10:04

πŸš€ AI Tools for Increased Productivity

The video highlights how AI tools like Photoshop's generative expand feature can significantly increase productivity, particularly in tasks such as thumbnail editing. The speaker shares their personal proficiency level with Photoshop and editing, and how they attempted to create a seamless infinite zoom effect using Adobe Premiere, but faced challenges due to the varying speed of image scaling. They eventually enlisted the help of a friend skilled in Adobe After Effects to achieve the desired outcome, albeit with some imperfections. The speaker emphasizes the gradual integration of AI into existing tools and processes, suggesting this is the true path towards a technological singularity.

15:34

πŸ“š Leveraging AI for Knowledge and Productivity

The speaker reflects on their experience with an app called Imprint, which distills key insights from popular books into interactive courses that can be completed in minutes. They discuss the app's effectiveness in learning concepts like 'atomic habits' and the '5-second rule' for combating procrastination. The Imprint app is praised for its visually engaging lessons and actionable advice. The speaker also acknowledges the potential dangers of AI, such as the ease of manipulating images or the loss of revenue for artists due to AI-generated content. They conclude by encouraging viewers to share their thoughts on the topic and to try out the Imprint app for themselves.

20:35

🌟 Embracing AI for Personal Growth

The video concludes with a promotion for the Imprint app, offering a 7-Day free trial and a 20% discount on subscriptions for viewers. The speaker shares their enthusiasm for the app and how it can serve as a 'cheat code' for self-improvement. They also invite viewers to subscribe to their channel for regular content and express their gratitude for the viewers' engagement. The speaker hints at returning to their usual video format in the future but appreciates the audience's interest in this experimental video.

Mindmap

Keywords

πŸ’‘Generative AI

Generative AI refers to artificial intelligence systems that are capable of creating new content, such as images, text, or music, that did not exist before. In the context of the video, the host discusses the integration of generative AI tools in Photoshop, which allows users to generate new image content based on existing parts of an image or text prompts. This technology is showcased through features like generative expand and generative fill, which are used to create thumbnails and expand images in a way that appears seamless and natural.

πŸ’‘Photoshop

Photoshop is a widely used software for image editing and manipulation developed by Adobe Inc. It is a professional tool that offers a wide range of features for working with images, including retouching, graphic design, and 3D rendering. In the video, the host talks about the 2024 update to Photoshop, which includes new generative AI tools that make it easier to create and edit images, particularly for tasks like generating thumbnails for videos.

πŸ’‘Thumbnails

A thumbnail in the context of video content refers to a small, representative image that is used to preview and promote a video. Thumbnails are crucial for attracting viewers' attention on platforms like YouTube. The host mentions using generative AI tools in Photoshop to create thumbnails for over 800 videos on his channel, highlighting the efficiency gains from using AI in this task.

πŸ’‘Content-Aware Fill

Content-Aware Fill is a feature in Adobe Photoshop that automatically fills in selected areas of an image with content that resembles the surrounding pixels, making it appear as if the filled area was always part of the original image. The host discusses how this feature has evolved with the addition of generative AI, allowing for more sophisticated and seamless filling of empty spaces in images, particularly useful for creating 16:9 aspect ratio thumbnails from vertical photos.

πŸ’‘16:9 Aspect Ratio

The 16:9 aspect ratio is a widescreen format that is commonly used in video and television production. It is favored for its ability to fit well on both widescreen and standard screens without cutting off the image. The host talks about the challenges of creating thumbnails with this aspect ratio from vertical photos and how AI tools in Photoshop help to overcome these challenges.

πŸ’‘Adobe Stock

Adobe Stock is a subscription-based service that provides access to a large collection of stock photos, videos, illustrations, and other visual assets. It is used by designers, marketers, and content creators to find high-quality images for their projects. The host mentions using Adobe Stock to find a photo for a video about de-extincting animals, which he then uses with the generative expand tool in Photoshop.

πŸ’‘Generative Expand

Generative Expand is an AI-powered feature in Photoshop that allows users to expand the canvas of an image and fill in the new areas with content that is predicted by the AI based on the existing image data. The host uses this tool to create a larger image from a smaller one, maintaining the visual coherence of the original subject.

πŸ’‘Infinite Zoom

Infinite Zoom is an experiment conducted by the host where he uses the generative expand tool repeatedly on an image to see what happens as the AI continues to create new content. The result is a series of images that progressively zoom out and generate new content, creating an effect of an infinitely zooming out scenario. This experiment explores the capabilities and limitations of generative AI.

πŸ’‘Firefly

Firefly is Adobe's generative AI model, which is used within the company's software, including Photoshop. It is trained using images from the Adobe Stock catalog, ensuring that all images used for training have licensing rights, which addresses some ethical concerns about AI training on copyrighted material. The host discusses how Firefly is integrated into Photoshop as part of the generative AI tools.

πŸ’‘Adobe Premiere

Adobe Premiere is a professional video editing software used for film and video production. The host mentions using Premiere to attempt an editing task where he wanted to scale and time images in a sequence to create a smooth transition effect, but found it challenging due to the complexities of velocity and scaling in the software.

πŸ’‘Adobe After Effects

Adobe After Effects is a digital visual effects, motion graphics, and compositing application used for video post-production. It is more advanced than Premiere and allows for more complex visual manipulations. The host's friend Mark uses After Effects to edit the infinite zoom experiment after the host faced difficulties with the velocity and scaling issues in Premiere.

Highlights

Photoshop now includes generative AI tools in its 2024 update.

Generative AI tools like generative Phill and expand allow users to fill in selected areas with AI-generated content.

The AI predicts what should be in the image based on the edges and surrounding context.

The author has experience using Photoshop for thumbnail creation and advertising work.

AI tools in Photoshop can be a game-changer for designers, saving significant time on tasks.

Content-aware fill is an older feature that fills in selected areas with surrounding content.

The generative expand tool in Photoshop is simple to use and can create space for AI to fill in.

The author experimented with generative expand to create an 'infinite zoom' effect by repeating the process.

The experiment resulted in bizarre and unexpected image transformations.

AI-generated images are now prevalent in stock libraries, raising ethical and copyright concerns.

Adobe's Firefly is a generative AI model trained on licensed images from Adobe Stock, addressing ethical concerns.

The author struggled with editing the 'infinite zoom' sequence due to varying scaling speeds.

After Effects was used to align and edit the 'infinite zoom' sequence, which was time-consuming.

The final 'infinite zoom' video had some imperfections but showcased the potential of AI tools.

The author reflects on the gradual integration of AI into our lives and its impact on productivity.

Imprint is an app that provides interactive courses based on popular books, aiming to improve productivity and learning.

The author suggests that AI tools are not a singular event but a continuous integration into our daily processes.