Using Craiyon and DeepAI to test text-to-art AI prompts | Why Try AI
TLDRIn this episode of 'Why Try AI', the host introduces two free AI art generators, Craiyon and DeepAI, which allow users to test text-to-art prompts without registration. The video demonstrates how to refine prompts for generating consistent and desired images, despite the lower quality compared to paid models. The host guides viewers through the iterative process of improving prompts, emphasizing the importance of consistency over perfection, and concludes with a satisfactory result ready for more advanced AI tools.
Takeaways
- 🎨 There are numerous AI art generators that can create impressive art, but many require sign-up or payment for extensive use.
- 🆓 Introducing two free AI tools, Craiyon (formerly Dali mini) and DeepAI, which require no registration and allow quick testing of text prompts.
- 📝 Both tools focus on the text prompt as the core of communication with AI art generators, which is an essential skill to master.
- 🔄 They provide multiple images per text prompt, helping to assess consistency and the effectiveness of the text prompt.
- 🚫 Caveat: The quality of images from these free tools is not as polished as paid models, so they are not suitable for benchmarking the best results.
- 🔍 Consistency is key; the goal is to refine the text prompt for the desired outcome rather than seeking perfection from these tools.
- ⚠️ Be aware that results from free tools may not directly translate to paid models due to differences in algorithms and AI training.
- 📝 It's important to document each text prompt and the resulting images to learn how adjustments affect outcomes.
- 🤖 The process involves iterative testing, starting with minimal information and gradually adding details to refine the prompt.
- 🖼️ An example test case is given where the AI is asked to create a dark, dystopian image of a giant mouse over a city, illustrating the iterative process.
- 🔧 Experimentation with adjectives and specific terms helps guide the AI towards the desired aesthetic, even if some attempts don't work as expected.
- 🎭 Trying to mimic a specific artist's style, such as Frank Miller, can be a strategy to refine the AI's output, but it may not always improve the result.
Q & A
What is the main topic of the video episode of 'Why Try AI'?
-The main topic of the video episode is to demonstrate how to use free AI art generators, specifically Craiyon and DeepAI, to test out text prompts for generating art without the need for sign-up or registration.
Why is the presenter encouraging the audience to try AI art generators?
-The presenter encourages the audience to try AI art generators because they can create impressive art that looks like it was drawn by a real artist, showcasing the amazing capabilities of this technology.
What are the limitations of some AI art generators mentioned in the video?
-Some AI art generators require sign-up steps, a registration process, or even a waiting list. Additionally, they may offer free credits initially but eventually require payment to generate more images.
How do Craiyon and DeepAI differ from other AI art generators in terms of user experience?
-Craiyon and DeepAI differ by being 100% free and requiring no registration. They focus on the text prompt and eliminate additional settings or options that might be overwhelming for casual users.
What is the advantage of getting multiple images for each text prompt in Craiyon and DeepAI?
-The advantage is that it allows users to see if their text prompt is consistent in producing the desired result across different images, which is helpful in refining the prompt for better outcomes.
What are the two caveats mentioned about using Craiyon and DeepAI?
-The two caveats are: 1) The quality of the images generated by Craiyon and DeepAI is not as polished as more sophisticated models, so expectations should be managed. 2) Success with a text prompt in these tools does not guarantee the same outcome in other AI generators due to differences in algorithms and models.
What is the process the presenter uses to refine the text prompt for generating the desired AI art?
-The presenter starts with a minimal amount of information, generates an image, and then iteratively adjusts the text prompt based on the results, noting which prompts work and which do not, until the desired aesthetic is achieved.
Why is it important to document the text prompts and the resulting images?
-Documenting the text prompts and resulting images is important for tracking the iterative process, understanding how different elements of the prompt affect the outcome, and refining the prompts for better results in the future.
What was the initial text prompt the presenter used to test the AI art generator?
-The initial text prompt was 'a giant mouse looking down on a city movie poster', aiming to create a dark, dystopian image.
How did the presenter adjust the text prompt to move closer to the desired outcome?
-The presenter adjusted the text prompt by adding terms like 'photo', 'realistic', 'scary', 'sinister', 'silhouette', and experimenting with different angles and descriptions to guide the AI towards the intended aesthetic.
What was the final outcome of the presenter's iterative process with the AI art generator?
-The final outcome was a set of images that closely matched the desired dark, dystopian aesthetic, with a giant mouse looming over a city in a black and white silhouette, which the presenter felt comfortable taking into a more sophisticated program for further refinement.
Outlines
🎨 Introduction to AI Art Generators
The video begins with an introduction to the use of AI art generators, emphasizing their ability to create stunning artwork that rivals human artists. The host encourages viewers to explore these tools despite the often required sign-up processes and potential costs associated with generating images. Two free AI tools, Crayon (formerly Dali mini) and Deep AI text to image generator, are highlighted as alternatives that require no registration and allow for quick testing of text prompts. The focus is on mastering the art of crafting effective text prompts, which is crucial for communicating with AI art generators.
🖌️ Iterative Process of AI Art Creation
The host demonstrates the iterative process of refining text prompts to achieve desired AI-generated art. Initially, the prompt 'a giant mouse looking down on a city movie poster' results in a cartoonish image far from the intended dark, dystopian theme. Adjustments are made to the prompt, such as specifying 'photo of a giant mouse' and removing 'movie poster' to enhance realism. Further refinements include adding 'silhouette' and 'looming over a city' to achieve a more sinister look. The host emphasizes the importance of noting down each prompt and the resulting images to understand how different prompts affect the outcome, ultimately aiming for consistency in the AI's interpretation.
🌐 Finalizing the AI Art Prompt
Continuing the iterative process, the host experiments with various prompts to refine the AI-generated image of a giant mouse looming over a city. Attempts include adding 'photo', 'scary', 'anthropomorphic', and 'from below' to improve the realism and sinister feel. The addition of 'laser eyes' proves unsuccessful, leading to a removal and a return to the previous, more effective prompt. The use of adjectives like 'intricate' and 'sinister' helps achieve a closer match to the desired aesthetic. The host also tries adding a style modifier 'by Frank Miller' but finds it inconsistent and removes it. Finally, the host tests the refined prompt on Deep AI, noting the consistency in the AI's interpretation and deciding it's ready for more sophisticated programs.
Mindmap
Keywords
💡AI art generators
💡Text prompts
💡Craiyon
💡DeepAI
💡Image iteration
💡Dystopian
💡Sinister
💡Photorealism
💡Silhouette
💡Intrinsic
💡Frank Miller
Highlights
Introduction to using free AI art generators to test text prompts without registration.
Craiyon and DeepAI as two free tools for quick testing of text prompts.
Both tools provide multiple images per prompt, useful for checking prompt consistency.
Craiyon offers nine images per prompt, while DeepAI provides four.
Initial prompt example: 'a giant mouse looking down on a city movie poster.'
Discussion on the importance of noting down text prompts and generated images for future reference.
Experimenting with prompt changes to achieve desired results.
Adding 'photo' and 'view from the ground' to the prompt for more realistic results.
Using 'silhouette' and 'seen from below' to achieve a darker, more sinister look.
Testing variations like 'torso of a giant mouse looming over a city.'
Including descriptive adjectives like 'scary' and 'anthropomorphic' for better results.
The importance of iteration and refining text prompts.
Trying artist-specific styles, such as adding 'by Frank Miller,' to influence the aesthetic.
Comparing results from Craiyon and DeepAI to check for consistency.
Finalizing a prompt to use in more sophisticated models for higher quality images.
Emphasizing the cost-saving and efficiency benefits of using free tools for initial testing.
Encouraging viewers to try the process themselves and share their experiences.