Something Different - GenAi for Tour Portrait

[jimlosaltos]

This technology gets controversial when used to create more realistic images to mislead people. I won't go there, but these samples hint at where the technology is headed. None of my attempts for this set aim to "look like a real photograph of an actual event".

Here, I've directed Midjourney to "imagine" (which is the literal opening instruction in Midjourney) Roger Federer as a Renaissance artist in his studio. Seems apt both because Fed has an artistic game, the demeanor of a creative genius, and I can see him painting in retirement <g>. The broad-brimmed hat was called "a wide-awake hat". The headline was generated separately in Firefly.

filedata/fetch?id=103035&d=1704391149&type=thumb

This combines a real photo I took of Rafa with an imaginary background to depict Rafa literally sailing into the sunset of retirement. Rafa's Babolat from my photo is replaced with a sailor's tiller. Adobe Firefly, its GenAI, is integrated into Photoshop. Here I gave it a series of text instructions to remove the background, then create the sunset over an ocean. I repeatedly generated that over and over until I got one that was close enough. Then I re-generated the sailing ship ropes, then integrated the two versions in Photoshop, and finally generated the birds. Whew. What's impressive to Photoshop users is that Firefly creates new content that matches the direction of lighting, shadows and tone in the original photo, something that would take hours to match by hand, even if you have the skills and the patience.

filedata/fetch?id=103036&d=1704391149&type=thumb

P.S. I preferred this earlier one of an old, retired Fed, but I was told he was less recognizable with the beard <g>

filedata/fetch?id=103034&d=1704391149&type=thumb

[jimlosaltos]

As Monty Python famously said, "And now for something completely different."

January is traditionally used by publications for forward-looking articles, so we went with something different, a collection of tennis images created with Generative Ai. I view this as similar to Tour Portraits we've done comprising abstract images, just with different tools. This collection endeavors to show the variety of tools and different types of image output. For those concerned with GenAi's impact as "photography" (which is not what I'm sharing here}, I included a link to a NYT article at the end of this post.

All the images in this tour portrait were created with three GenAi systems: DALLE-E, which is from the highly-publicized OpenAI corporation that created ChatGPT and is partnered with Microsoft; Midjourney, an independent company with a more complex system giving the user more control over its output, and Firefly, Adobe's approach, which is quite different from anyone else's I'm aware of, integrating GenAi into its current photo editing software, including Photoshop. All of these images were created last July, so each of the tools has since evolved on in subsequent releases.

Please click here to see the January TPN Tour Portrait:
https://www.tennisplayer.net/article...tte-ai-tennis/

Here are a couple of samples to illustrate potential uses. I'll post a few more with explanations in the next couple of days.

This image was generated by Firefly using a text prompt starting with "1920's female tennis player, art deco style, block colors".

filedata/fetch?id=103022&d=1704319467&type=thumb

For contrast, this DALLE-E generation imagines tennis as part of a mythical, medieval knight's joust, making sudden death take on a different meaning. Other text parameters set style, level of detail, armour and the output results from numerous iterations.

For discussion of GenAI photography impact:
NYT: https://www.nytimes.com/2023/12/26/opinion/ai-future-photography.html?unlocked_article_code=1.I00.vlNG. YJXolEWuj5CC&smid=url-share

filedata/fetch?id=103023&d=1704319467&type=thumb