Limited-Time Offer: Save 40% on Annual Plans!🎉

Nano Banana & Qwen: The New KINGS! Try them FREE!

Theoretically Media
19 Aug 202518:53

TLDRThe video explores two new free AI image editors: Quen Image Edit from Alibaba and the mysterious Nano Banana AI. Quen ImageEdit, an open-source model, excels in semantic visual editing and bilingual text accuracy. Nano Banana, speculated to be from Google, impresses with its ability to generate additional camera angles and handle complex image references. The host also discusses updates from Clling, including the upcoming first frame and last frame feature in Cling 2.1, and a new video-to-music feature from 11 Labs. Notion AI is highlighted for its organizational and creative capabilities in AI filmmaking.

Takeaways

  • 🚀 Two new AI image editors, Quen Image Edit and Nano Banana AI, are available for free trials.
  • 🎨 Quen ImageEdit from Alibaba is an open-source model that can be run locally or on various platforms, offering impressive low-level and high-level semantic visual editing capabilities.
  • 🍌 Nano Banana is a mysterious new AI editor speculated to be from Google, excelling in image reference editing and generating additional camera angles within a scene.
  • 🌟 Nano Banana demonstrated superior performance in creating cinematic shots and maintaining style consistency compared to other models like GPT Image.
  • 🤖 Notion AI, introduced by Notion, offers features like AI meeting notes, research mode, and built-in chat mode, which can be useful for organizing and generating creative content for AI filmmaking.
  • 🎥 Notion AI can help create detailed character profiles, world-building elements, and even generate prompts for creative assets like cities.
  • đź“… Notion AI can generate a visual task board and a production schedule for filmmaking projects, helping streamline the production pipeline.
  • 🎉 Cling 2.1 is set to release the first frame, last frame feature, which is expected to be a powerful addition for editing.
  • 🎵 11 Labs has released a new feature that converts video to music, generating contextual music for scenes.
  • đź‘€ The presenter highlights the potential of these new tools in creative fields, especially for independent filmmakers and content creators.
  • đź”— All tools mentioned offer free trials or access, making advanced AI capabilities available to a wider audience.

Q & A

  • What are the two new AI image editors discussed in the video?

    -The two new AI image editors are Quen ImageEdit from Alibaba and Nano Banana.

  • How does Quen ImageEdit stand out from other AI image editors?

    -Quen ImageEdit is open-source, allowing it to run locally or on multiple platforms. It excels at both low-level and high-level semantic visual editing, such as rotating characters and handling text in different languages.

  • What is unique about the Banana image generated by Quen ImageEdit?

    -The banana generated in the Quen ImageEdit is a little exaggerated in appearance, looking more GMO-like than organic. Despite this, it successfully replaces the original object in the image with a banana.

  • How does Nano Banana compare to other AI image editors?

    -Nano Banana is known for its powerful reference image editing capabilities, such as generating additional shots from a scene and maintaining stylistic consistency. It is highly regarded for its ability to create multiple perspectives of a scene based on a single imageAI image editors comparison.

  • What is the speculation surrounding the identity of Nano Banana?

    -There is speculation that Nano Banana could be developed by Google, possibly connected to its Gemini 3 AI models, although this has not been confirmed.

  • What kind of image editing is Nano Banana particularly good at?

    -Nano Banana excels in generating multiple camera angles from a single image, producing highly consistent and detailed new perspectives within a scene.

  • What are some notable features of Quen ImageEdit?

    -Notable features of Quen ImageEdit include its ability to handle text generation in multiple languages, including Chinese, as well as its proficiency in low-level edits like removing objects and high-level tasks such as character rotations.

  • How does Quen ImageEdit handle the manipulation of text in images?

    -Quen ImageEdit can accurately generate and manipulate text within images, including stacking text, which is a challenging task for many AI models.

  • What is a standout feature of Nano Banana in terms of scene editing?

    -A standout feature of Nano Banana is its ability to generate additional scenes or shots, even from minimal input, which is useful for creating cinematic sequences or exploring different perspectives of a single moment.

  • How does Notion AI contribute to AI filmmaking?

    -Notion AI helps organize AI film production by generating task lists, creating character descriptions, and providing prompts for world-building, all within an organized knowledge base that keeps creative content easily accessible.

Outlines

00:00

null

The speaker introduces two new free AI image editors: Alibaba's Quen ImageEdit and the mysterious Nano Banana, highlighting Quen first. Quen ImageEdit is open-source, allowing local or platform-based use. It is tested through various examples: converting a gun into a banana, removing objects like hair strands from plates, colorizing black-and-white images, and performing complex edits such as rotating characters or transferring styles. Quen excels at maintaining visual consistency while handling both low-level and high-level semantic edits. Additionally, it demonstrates impressive bilingual text-editing capabilities, producing accurate translations and realistic text incorporation into images. Real-world demonstrations include anime-style transformations, character turnarounds, tributes to designer Joe Caroff, and fashion magazine-style edits. Overall, Quen ImageEdit proves effective at precise visual modifications, though occasionally outputs overly stylized or artificial-looking elements.

05:01

🍌 The Rise of Nano Banana: Google’s Possible Secret Weapon

Nano Banana emerges as a mysteriousQuen ImageEdit overview new AI image editor, speculated to be developed by Google. It demonstrates powerful reference-based editing, capable of combining multiple inputs such as characters, carsJSON code correction, and outfits into coherent results. Accessible via the LM Arena battle page, Nano Banana is tested through side-by-side comparisons with other models, showing its superior ability to follow prompts and maintain scene coherence. Its standout feature is generating alternative camera angles and cinematic shots from a single input image, effectively simulating multi-camera film coverage. Examples include sci-fi dystopia scenes rendered with different perspectives and realistic photo edits like removing crew members from behind-the-scenes Kubrick photography. While highly capable, occasional inconsistencies, such as subtle pose shifts, appear in certain generations. Despite these, Nano Banana often outperforms competitors like Quen and GPT Image models in realism, prompt coherence, and stylistic consistency.

10:03

🕵️‍♂️ Nano Banana’s Style Consistency and Speculation Around Google

Further testing Nano Banana, the speaker examines its ability to preserve facial accuracy, style, and scene coherence. It successfully inserts the creator into a cinematic spy setting and maintains stylistic consistency across animated scenes by rotating characters while keeping them visually coherent. This raises speculation that Nano Banana may be tied to Google’s Gemini 3 or upcoming imaging technologies, possibly integrated into future Pixel devices. Its classification as a 'nano' model hints at potential use in on-device editing. While awaiting confirmation at Google’s upcoming event, the creator underscores Nano Banana’s impressive abilities and positions it as a potential game-changer in real-time or mobile-based AI image editing. The section transitions into a sponsored segment on Notion AI, showcasing how it can enhance AI filmmaking workflows by providing organization, task management, and creative support through integrated AI-powered features.

15:03

đź“‹ From Notion AI to Cling and 11 Labs: Workflow & Media Innovations

The final section pivots to Notion AI, where the creator demonstrates how it can function as a creative and production hub for AI filmmaking. Notion AI assists with meeting notes, project organization, character development, creative world-building, and task management, even generating production schedules and visual task boards. Its integration allows creative ideas (like fantasy cities and character backstories) to remain connected across a knowledge base while also handling logistical details like calendars and deadlines. After the sponsorship, attention shifts to Cling 2.1’s 'first frame, last frame' feature, which enhances visual storytelling by enabling better start/end frame handling, a key upgrade for animation and film sequences. Finally, 11 Labs introduces a 'video to music' feature, which generates contextual background music from uploaded videos. While not on par with professional composers, it produces genre-appropriate audio that complements video content. The segment concludes with optimism about ongoing innovation in AI creative tools and excitement about future updates.

Mindmap

Keywords

đź’ˇAI image editors

AI image editors are software tools that use artificial intelligence to manipulate or generate images. In the context of this video, the main theme revolves around exploring and comparing two new AI image editors—Quen ImageEdit and Nano Banana. These tools allow users to make changes to images based on text prompts, such as transforming objects or altering styles. For example, in the script, the host uses Quen ImageEdit to change a gun into a banana in an image, demonstrating the editor's ability to perform semantic visual editing.

đź’ˇQuen ImageEdit

Quen ImageEdit is an open-source AI image editor developed by Alibaba. It is highlighted in the video as a powerful tool for both low-level and high-level semantic visual editing. This means it can handle tasks like removing small objects from an image or making more complex changes like rotating characters and applying style transfers. The video shows examples of using Quen ImageEdit to remove a strand of hair from a plate and to create a hyper-stylized anime image, illustrating its versatility and accuracy in maintaining the original image's context while making the requested edits.

đź’ˇNano Banana

Nano Banana is a mysterious new AI image editor that has recently appeared and is speculated to be developed by Google. It is described in the video as an impressive model for image reference editing, capable of generating additional camera angles and maintaining coherence in scenes. For example, the host demonstrates Nano Banana's ability to create multiple shots from a single image, such as generating close-ups of characters in a scene. This feature makes it stand out as a powerful tool for creating cinematic sequences from still images.

đź’ˇSemantic visual editing

Semantic visual editing refers to the process of making changes to an image based on the meaning and context of the visual elements within it. In the video, this concept is central to the discussion of Quen ImageEdit's capabilities. For instance, the editor can understand and modify specific objects or characters in an image while keeping the rest of the scene intact. The example of changing a gun into a banana showcases how the AI interprets the prompt and executes the change in a way that maintains the overall coherence of the image.

đź’ˇLow-level and high-level editing

Low-level editing involves making small, detailed changes to an image, such as removing a strand of hair or adjusting minor elements. High-level editing, on the other hand, involves more significant transformations, like rotating characters, changing the style of the image, or generating new perspectives. In the context of the video, Quen ImageEdit is praised for its ability to perform both types of editing effectively. For example, it can remove a small object from an image (low-level) and also rotate characters or apply style transfers (high-level).

đź’ˇText editing in AI models

Text editing in AI models refers to the ability of these tools to modify or generate text within images. The video highlights Quen ImageEdit's bilingual text editing capabilities, which are demonstrated through examples where text is accurately inserted or modified in images. This feature is particularly useful for creating images with specific text elements, such as fashion magazine covers or signs, as shown in the script where the host mentions Quen's ability to maintain text on street signs when transforming an image.

đź’ˇImage reference editing

Image reference editing is the process of using multiple reference images to guide the AI in generating a new image that combines elements from these references. Nano Banana is noted in the video for its exceptional ability in this area. For example, the host mentions how Nano Banana can take separate references for a location, a car, a character, and an outfit, and generate a coherent image that incorporates all these elements. This capability makes it a powerful tool for creating complex and detailed images based on specific references.

đź’ˇCinematic scene generation

Cinematic scene generation involves creating a sequence of images that can be used to form a short video or scene. Nano Banana is shown in the video to be particularly adept at this, as it can generate multiple camera angles from a single image. The host demonstrates this by creating close-ups of different characters in a scene, effectively generating a mini cinematic sequence. This feature is significant for filmmakers and content creators who want to quickly generate multiple shots for a scene.

đź’ˇAI filmmaking

AI filmmaking refers to the use of artificial intelligence tools to assist in the creation of films or video content. The video touches on this concept through the discussion of using Notion AI to organize and generate creative content for an AI film project. The host explores how Notion AI can help in organizing production pipelines, generating character descriptions, and even creating a visual task board for pre-production, production, and post-production phases. This highlights the potential of AI in streamlining and enhancing the filmmaking process.

đź’ˇNotion AI

Notion AI is a feature within the productivity app Notion that integrates artificial intelligence to assist users in various tasks. In the context of the video, Notion AI is used to help organize and generate content for an AI film project. It can summarize meeting notes, create character backstories, generate prompts for creative tasks, and even develop a production schedule. The video demonstrates how Notion AI can be a valuable tool for both creative and organizational aspects of filmmaking, making it easier to manage complex projects.

Highlights

Two new AI image editors, Quen ImageEdit and Nano Banana, are available for free trials.

Quen ImageEdit from Alibaba is an open-source model that can be run locally or on various platforms.

Quen ImageEdit is effective at low-level semantic visual editing and high-level semantic editing.

Quen ImageEdit supports bilingual text editing with high accuracy.

Nano Banana is speculated to be developed by Google but has not been officially confirmed.

Nano Banana excels at image reference editing with multiple references.

Nano Banana can generate additional camera angles and cinematic shots from a single image.

Nano Banana maintains style consistency and handles real-life photography well.

Notion AI introduces features like AI meeting notes, research mode, and built-in chat with advanced models.

Notion AI can be used as a hybrid creative and production office for AI filmmaking.

Notion AI generates creative content like character backstories and world descriptions.

Notion AI creates visual task boards and production schedules for organizing filmmaking.

Cling 2.1 is set to release the first frame and last frame feature, enhancing animation capabilities.

11 Labs releases a new feature that converts video to music, providing contextual audio for scenes.

The video to music feature from 11 Labs generates music that matches the genre of the video.