What AI Image Generator Should YOU Be Using??

Matt Wolfe
19 Oct 202348:29

TLDRIn this comprehensive review, various AI image generators are evaluated based on accuracy, creativity, realism, and other criteria. Mid Journey excels in creativity and realism but has usability and cost drawbacks. Dolly 3, while accurate, is highly censored and expensive. Google and idiogram offer free options with decent performance. Leonardo stands out for its low censorship and versatility, making it the best value overall.

Takeaways

  • 🎨 AI image generators like Mid Journey, Dolly 3, Firefly Image 2, Stable Diffusion XL, and Google's generative search experience are evaluated based on accuracy, creativity, realism, and other factors.
  • 🚀 Mid Journey and Dolly 3 (especially in Chat GPT) excelled in accuracy, closely adhering to the prompts given, with Dolly 3 scoring a 9 out of 10 for its prompt adherence.
  • 🌈 In terms of creativity, Mid Journey led the pack, offering the most vibrant and imaginative images, followed by Stable Diffusion XL and Leonardo.
  • 🖼️ For realism, Mid Journey's raw style was the most convincing, closely followed by Firefly Image 2, which provided very realistic facial features.
  • 📚 When evaluating illustrations, Mid Journey's Nii mode and Leonardo performed well, providing detailed and stylistically coherent images.
  • 🏷️ For logo and vector creation, Google's generative search experience surprisingly outperformed, offering simple and flat vector images that closely matched the prompt.
  • 🎨 Text within images was best handled by Dolly 3, Google, and Idiogram, which were able to incorporate text accurately onto signs or objects within the generated images.
  • 🔄 For tiling textures and backgrounds, Mid Journey and Stable Diffusion XL (via Leonardo) were capable of creating seamless, tilable patterns.
  • 🚫 Censorship varied among the platforms; Mid Journey and Google were less restrictive, while Firefly showed the most censorship.
  • 💰 In terms of pricing, Dolly 3 (Bing Image Creator version) and Idiogram offered free options, while Mid Journey required a subscription, and Dolly 3 (Chat GPT version) was the most expensive.
  • 📊 Overall, Leonardo emerged as the best value, offering a balance of features, performance, and a free tier, making it versatile for a wide range of use cases.

Q & A

  • What are the main AI image generators discussed in the transcript?

    -The main AI image generators discussed are Mid Journey, Dolly 3, Firefly Image 2, Stable Diffusion XL, Google's generative search experience, and Idiogram.

  • What criteria were used to evaluate the AI image generators?

    -The criteria used for evaluation include accuracy, creativity, realism, illustrations, logos and vectors, textures, background usage, censorship in images, usability of user interfaces, and pricing.

  • How did the AI generators perform in terms of accuracy?

    -Dolly 3 performed the best in accuracy, scoring a 9 out of 10, followed by Mid Journey with a 5.5 when using the raw style. Idiogram and Firefly 2 scored a 6.7 and 6.5 respectively, while Google and Stable Diffusion XL scored a 7.2 and 6.5.

  • Which AI image generator was found to be the most creative?

    -Mid Journey was found to be the most creative, especially when using the raw style, followed closely by Stable Diffusion XL and Leonardo, and then Dolly 3.

  • In terms of realism, which AI image generator scored the highest?

    -Mid Journey raw was considered the most realistic, scoring an 8.5, followed by Firefly 2 with an 8, and then Mid Journey without using raw, which scored a 7.5.

  • How did the AI generators handle the task of creating illustrations?

    -All the AI generators performed decently at creating illustrations, with Mid Journey, Leonardo, and Firefly 2 being particularly noted for their quality. Google struggled with prompt adherence but still managed to create decent illustrations.

  • Which AI image generator was able to create tilable textures effectively?

    -Mid Journey and Stable Diffusion XL (using Leonardo) were able to create effectively tilable textures, scoring a 10, while Dolly 3, Bing Image Creator, and Idiogram struggled with this task.

  • How did the AI generators handle the inclusion of text within images?

    -Dolly 3, Google, and Idiogram were able to include text within images effectively, while Mid Journey struggled with this task. Firefly 2 and Leonardo had some issues but were closer to getting the text right.

  • What were the usability scores for the AI image generators?

    -Leonardo received the highest usability score of 9, followed by Firefly 2 with an 8, and Dolly 3 inside of Chat GPT with an 8. Mid Journey was given a 5 due to its user interface being in Discord, which can be overwhelming.

  • Dolly 3 with Bing Image Creator and Idiogram both received a 10 for pricing as they are currently free to use. Google's generative search experience is also free, scoring it a 10. Mid Journey scored a 6 due to its cost, and Leonardo scored a 7.5 because it offers a free tier and more image generations for its paid plan compared to Mid Journey.

    -null

  • Which AI image generator was considered the best overall value?

    -Leonardo was considered the best overall value, scoring a total of 75.5 across all categories. It excelled in most areas except for text inside images.

  • What was the final verdict on Dolly 3 inside of Chat GPT?

    -Dolly 3 inside of Chat GPT performed the worst among all the generators discussed. It requires a $20/month subscription for Chat GPT Plus, has censorship issues, and does not perform as well in several categories including realism, textures, and backgrounds.

Outlines

00:00

🤖 Overview of AI Image Generators

The paragraph introduces a variety of AI image generators available, highlighting the challenge in selecting the right tool for specific use cases. It mentions popular generators like Mid Journey, Dolly 3, Firefly Image 2, Stable Diffusion XL, and Google's generative search experience. The video's aim is to determine the best tool based on criteria such as accuracy, creativity, realism, illustrations, logos, vectors, textures, usability, and pricing.

05:02

🎨 Testing Accuracy and Prompt Adherence

This section focuses on testing the accuracy of AI image generators by comparing how well they adhere to specific prompts. The test involves using various tools like Mid Journey, Dolly 3, and others with different prompts to see how closely the generated images match the requested descriptions. The paragraph discusses the results, giving scores to each tool based on their accuracy in rendering the prompts.

10:03

🌈 Creativity and Image Diversity

The paragraph delves into the creativity of AI image generators by evaluating their ability to produce unique and diverse images from broad prompts. It compares tools like Mid Journey, Dolly 3, and Google, noting their strengths and weaknesses in creating colorful and contrasting images. The discussion includes personal opinions on the creativity level of the generated images and assigns scores accordingly.

15:05

🖼 Realism in AI-Generated Images

This part assesses the realism of AI-generated images using a specific prompt about a couple holding hands in front of the Eiffel Tower. The paragraph compares various tools, including Mid Journey, Dolly 3, and Firefly Image 2, based on how realistically they depict people and locations. The evaluation considers the level of detail and the convincing nature of the images, with scores reflecting the perceived realism.

20:06

📚 Illustration Styles and Anime Characters

The paragraph discusses the ability of AI image generators to create illustrations, particularly in anime style, using the prompt of an anime girl with braids in neon streets. It compares tools like Mid Journey, Dolly 3, and Firefly Image 2, evaluating the quality of the illustrations, the level of detail, and the adherence to the prompt. The summary includes scores based on the visual appeal and stylistic consistency of the generated images.

25:06

🏷️ Logos, Vectors, and Text in Images

This section evaluates the capability of AI image generators to create logos and vectors, as well as incorporate text into images. The paragraph tests tools with a prompt for a simple flat vector image logo of a wolf and compares the results in terms of simplicity, style, and effectiveness in text rendering. The discussion highlights the strengths and limitations of each tool in producing logos and text-inclusive images, with assigned scores reflecting their performance.

30:07

🌟 Textures, Backgrounds, and Tiling

The paragraph focuses on the ability of AI image generators to create textured, tiling backgrounds. It tests the tools with a prompt for colorful circuitry and evaluates their success in generating seamless, repeatable tiles. The discussion includes a comparison of Mid Journey, Dolly 3, and others, with scores assigned based on the effectiveness of tiling and the quality of the textures produced.

35:10

🚫 Censorship and Content Restrictions

This section explores the censorship and content policy restrictions of AI image generators when creating images with celebrity faces and intellectual property logos. The paragraph compares how different tools handle prompts involving recognizable figures like Tom Hanks and characters like Stormtroopers, discussing the level of censorship and the ability to generate such content. The evaluation includes scores that reflect the generators' willingness and capability to produce restricted content.

40:12

💰 Usability and Pricing of AI Image Generators

The paragraph discusses the usability and pricing of the AI image generators compared earlier. It evaluates the user interfaces, customizability, and cost-effectiveness of each tool, providing scores based on ease of use and affordability. The discussion highlights the trade-offs between cost, features, and usability, helping users decide which tool might be the best fit for their needs and budget.

45:13

📊 Conclusion and Recommendations

The paragraph wraps up the comparison by summarizing the overall performance of each AI image generator across various criteria. It highlights the best value option, the most creative and realistic tools, and the least censored versions. The summary provides a clear recommendation on which tools to use in different situations, taking into account factors like accuracy, creativity, realism, usability, and price.

Mindmap

Keywords

💡AI image generators

AI image generators are artificial intelligence systems capable of creating visual images based on user inputs or prompts. In the context of the video, they are compared for various use cases such as accuracy, creativity, and realism. Examples mentioned include Mid Journey, Dolly 3, Firefly Image 2, and Stable Diffusion XL.

💡Prompt adherence

Prompt adherence refers to the ability of an AI image generator to accurately follow and interpret the user's instructions or prompts to create an image. It is a critical aspect when evaluating the performance of AI image generators, as it measures how well the AI understands and executes the given task.

💡Creativity

In the context of AI image generators, creativity refers to the ability of the AI to produce unique, innovative, and aesthetically pleasing images from vague or broad prompts. It is a measure of how well the AI can go beyond literal interpretations and add its own artistic flair to the output.

💡Realism

Realism in AI-generated images refers to the degree to which the images appear lifelike and could be mistaken for photographs or real-world scenes. It is an important criterion for evaluating AI image generators, especially for uses where believability is crucial.

💡Illustrations

Illustrations, in the context of AI image generation, refer to the creation of visual art that interprets a story, concept, or idea, often with a stylized and non-photorealistic approach. AI-generated illustrations can vary in style from cartoonish to semi-realistic.

💡Logos and vectors

Logos and vectors in AI image generation involve creating simple, scalable, and often geometric designs that can be used as symbols or icons. Vectors are important for branding and can be resized without losing quality, making them ideal for logos.

💡Text in images

Text in images refers to the integration of written words into visual artwork or photographs. For AI image generators, this capability is significant as it allows for the creation of images with captions, slogans, or other textual elements.

💡Censorship

Censorship in AI image generators pertains to the systems or policies in place that restrict or prevent the creation of certain types of content, such as images of celebrities or copyrighted IP. This can affect the versatility and usability of AI tools in various contexts.

💡Usability

Usability refers to how intuitive, user-friendly, and efficient an AI image generator's interface is. It encompasses the ease with which users can input prompts, adjust settings, and receive outputs, significantly impacting the overall user experience.

💡Pricing

Pricing refers to the cost associated with using an AI image generator. It is a critical factor for users, especially those looking for free or cost-effective solutions. The video compares various pricing models, from free tiers to paid subscriptions.

Highlights

The video compares various AI image generators, focusing on their accuracy, creativity, realism, and other specific use cases.

Mid Journey is considered one of the best AI image generators, but its raw style is often needed for higher prompt adherence.

Dolly 3 is being called the 'Mid Journey killer' due to its accuracy and adherence to prompts.

Firefly Image 2 is said to be just as good as Mid Journey and Dolly, with its new version bringing improvements.

Stable Diffusion XL is praised for its high level of customization.

Google has integrated an image generator into its generative search experience.

Idiogram was top of the AI art world a month ago for generating text inside images.

The video tests the AI generators on prompt adherence, creativity, realism, illustrations, logos, vectors, textures, backgrounds, censorship, usability, and pricing.

Dolly 3 excels in accuracy, closely following prompts and delivering high-quality images.

Mid Journey's raw style is particularly effective for creating realistic and detailed images.

Firefly Image 2 has improved significantly and is now on par with Mid Journey and Dolly in terms of creativity and quality.

Google's generative search experience allows for image generation but has limitations in certain areas like text generation.

Idiogram offers a free and uncensored platform for AI image generation, though its quality varies.

Leonardo, part of Stable Diffusion XL, provides a high level of customizability and is least censored among the tested tools.

The video concludes that Leonardo offers the best value overall, followed by Mid Journey and Idiogram.