Landing Page
These are the unprocessed AI images used for the landing page.
Luma AI Teaser
AI apps via Spaces
For comparison, an image using a free DALL-3 built on huggingface.co website.
DALL-E 3 via Microsoft Copilot
For comparison, some images were upscaled using the dgb.lol website.
Technique
Since I'm using the free version, I have to guide the AI carefully to create what I envision. For instance, I'll start with a prompt to draft something in a comic style inspired by Alan Moore’s ‘Promethea.’ My geeky background—thanks to decades of reading manga, comics, and graphic novels really helps. Once I've saved the image, I start a new topic, sharing specific details about a painter—during my travels, I explore various museums, from modern art to anthropology and design. Providing the AI to Jonas Burgert’s artist page, prompting an insightful analysis of the art style. Finally, I upload the saved picture, extract the relevant visual elements from the summary, and instruct the AI to recreate the image.
Original vs. Upscaled
The left-hand side shows the original image, while the right-hand side displays the processed image using the free tools available on the dgb.lol website.
Why ?!
Yes, I know it is obvious that I like mohawks and fishing. However, it is easy to see how sophisticated the DALL-E 3 image generator is if water, hair details, and white surfaces are included in the prompt.
Details
The picture on the left was enhanced before it was upscaled. The picture on top was just upscaled. In both pictures, the contrast differs, but the glaciers are more detailed in the originals.
Stable Diffusion
No ‘Upscale’ option was executed before the download. However, weighted prompts were used to fine-tune the composition.
Claude Sonnet 3.5
I experimented with WebSim.ai since I don’t have an API membership with Anthropic to test out the Artifacts feature with API publishing.
Midjourney
Unfortunately, I don’t have a paid subscription at the moment. However, when I dabbled with it years ago during the newbie trial, I managed to create some images. I was familiar with the prompting back then, and I assure you, I’ll revisit the section on exploring prompts to brush up on the specific MidJourney prompt syntax.
