It's all about words.

Me: 

Machine: 

Okay, so Peter Weir it's not, but it's a start.

Image from here:

Made a video using the prompt "I am behind him in the moving van, out of shot. He looks at me with a slow blink and turns away, then turns back to me and smiles with the whitest teeth against his tanned skin, his chin up. The van is moving quickly and with a slight sway, and his long limbs jolt with the van's movement. He is beautiful."

I took the last frame from that:

... to make the next video with the sequential prompt "His dazzling smile persists even as he talks to me, his toned body moving with the bumps of the moving van. He turns slightly more to me, and lazily rests his lightly haired right arm along the back of his seat rest, all the while chatting and occasionally looking towards the direction of the van's movement. He lets his head fall back and lets out a laugh. His arm falls away and he faces away again." 

I took both videos and turned them into GIFs. (I joined them into one, but it was too large to load.)

Cannes Film Festival, here I come!

Sure, it's AI slop, but it's good-looking and nicely hung AI slop. 

Each was generated from the original image, top left, then built on each incremental change with a new prompt: remove mobile phone, fold arms, hands on hips, steam-filled locker room, guys in towels, etc.

... and then there is the Silliness Algorithm, that assumes that users are 12 years old and on TikTok! Hardly elegant.

Removing backgrounds instead of adding them:

The InterVarsity All-round Kissing Freestyle event!

So, to Candy's original premise of backgrounds, I compared manual removal of fixtures with the removal by prompt mode, using the completely misnamed https://ezremove.ai/magic-eraser/ tool. (This tool makes free use thoroughly difficult to access or use. Loading free credits was by claiming one lot each after 15 and 30 minutes wait - a total of 45 minutes before having enough credits to do anything meaningful. Though it's a pain in the arse to use, it produced favourable results in the long run.  *gritted teeth* )

Base image:

My goal was to remove the pool fixtures (bars and railings) and towels to leave the water polo player emerging from the pool. First, manual removal of fixtures with a stylus and the BRUSH feature; it took five separate reloads to remove pool fixtures (1-3):

(4-5):

Fairly unsatisfactory for the time and effort expended.

Next, the prompt mode:

Result (single click):

Second prompt:

Result (one click):

So, for all the finess of manual removal, try the prompt first.

Dance, boy!

Not AI, but using conventional cut-and-paste tools I have this lad boogying for a moment longer.

The prompt:

The result:

The reason:

https://pixelbin.io/ was given these prompts:

     (i) Render the water polo player smiling confidently, with realistic well-built muscular arms, chest and big thighs beneath a narrow waist.

     (ii) Using the reference image only have the water polo player smiling confidently, with realistic well-built muscular arms, chest and big thighs beneath a narrow waist.

     (iii) Have the figure in the reference image smile confidently, with realistic well-built muscular arms, chest and big thighs.

and this reference image: 

these were the results generated:

Appalling disregard for the prospective client's wishes, whom they expect to sign up! Such chutzpah!

This may be of interest to someone, an image-to-prompt function where a prompt for an existing image may be refined for changes you'd like to make. https://vheer.com/app/image-to-prompt 

From this:

To this:

Prompt: "The two water polo players break from their kiss to run their hands over each other's pectoral and abdominal muscles appreciatively."

A random success! While I was testing free online AI image-to-video sites with standardised image and prompt, there was this output from a site which rejected the image as inappropriate but then spat this out:

Prompt: "The two water polo players kiss while running their hands up and down each other's backs and trunks."

Image:

What a nugget!

We do the thinking so you don't have to...

Verdict: don't bother!

When too many Grey Speedo Boys are barely enough!

I'm still fixated on Grey Speedo Boy (GSB), so here are some reworkings of images in to GIFs, and the tools and prompts used.

Scrubbing an image

Removing people, objects and backgrounds from an image, from thus:

to this:

to this:

using these: https://www.online-image-editor.com/ , https://objectremover.com/ and  https://www.iloveimg.com/remove-background gave me this:

(Well, that was steamy!)

Prompting AI generation

Using https://www.easemate.ai/image-to-video-generator , https://ezgif.com/ and https://ezremove.ai/video-watermark-remover/ with the prompts written in the image-to-video fields (screenshots above) gave me these:

(and another that just won't upload, probably due to file size).

I can tell just by the way he walks that the blond boy is a fanatical bottom.

Meta AI will only let you do a 20-second video.

1. Merge two images

+=

2. Convert to video with the prompt "Two water polo players admire each other's physiques, then move together to touch each other's chests, in front of a consistent static background. The water polo player on the left tenderly strokes the abdomen of the one on the right, who drapes his right arm over the shoulders of his mate, and moves his feet apart to be open to the tactile advances of the water polo player on the left." 

3. Result (after removing watermark):

Banksy tribute: prompt, "The shot opens with the stencil-art gavel-weilding judge in the background wall image repeatedly striking at the protester who holds a placard in defence of the attack, while a legal figure walks by and a cyclist in black lycra wearing a black backpack passes from screen-left to screen-right, oblivious to what is happening in the background."

I think that my patience has reached its limit with these AI programs.

1. Two images merged 

2. Prompt 1 - my words, "Swimmer leaps fluidly from the ocean and re-enters with dolphins in formation, from screen right to screen left."

Prompt 2 - my words revised, "Swimmer breaches the ocean swell and dives underwater with dolphins in formation, from screen right to screen left."

Prompt 3 - generated by easimate's Enhanced Prompt function.

3. Results, these triumphs of algorithmic genius:

4. With some trimming, I chose this:

If anyone has a picture they would like to see come to life. Post it here