AI is creating art and that is true. When Dall.E was introduced, there was so much awe and criticism around whether it is generating the right set of images for the prompt we give. Dall.E2 is the improvised version of Dall E and it could not escape criticism either. The AI image generators precisely make images based on the prompts fed. Not everyone is aware of the jargon that exactly describes what is visible. For e.g, using terminology used for photography such as close-up, and bokeh effect brings out results closer. To help with how a prompt should be framed, Guy Parsons has come out with a prompt book with all the word guides, aesthetics, vibes, emotional prompt language, and effects like inpainting, outpainting, editing, adding variations, etc. It also teaches how to get the desired effects in photography by including proximity, angles, lighting quality, and lens type in the prompt. For chronologic effects, you can take a few hints from historical data.
The author says, the creators themselves do not have a clue about what the tool knows, because it was only trained on pictures with captions. The users have to figure out what it knows to find out what output they can derive. Dall E2 is not going to generate the perfect result in the very first go. Only refined prompts, exactly which the book is written for, will help you see the imagined pictures. The book contains instructions on how to use adjectives, word the prompt with an appropriate number of words using natural language, and also how to combine separate images.
More Trending Stories
- US$10k is Coming for Bitcoin! But What Awaits After That?
- Where is Andrej Karpathy Headed After Leaving Tesla’s ‘Autopilot’?
- Google’s AI is Not a Pro in Data Labeling! But the Comp Fails to Admit it
- How Law will Solve Metaverse Crimes? Who will be the Witness?
- AI Image Generators Like Dall.E and Imagen Steal Ideas from Humans
- From Model-Centric to Data-Centric, How the AI Ecosystem is Moving?
- Top 10 Mini Ethical Hacking Projects Aspirants Should Add in Resume
