A conversation with Open (Source) AI

#ChatGPT and #GPT4 Guided Storytelling

If the videos don't work, click on my name under the twitter logo on each card to visit to Twitter. It seems like the cards get broken after a while.

2 Likes

Voices are now also becoming scary good:

https://watch.wave.video/nKKXmWLQn4pF5vMp

https://watch.wave.video/C1NCKHMtLMwcgj0G

1 Like
1 Like

OpenAI has once again set the benchmark with the release of Dall-E 3. Not only does it deliver unparalleled image quality, but its precise adherence to instructions showcases its deep understanding of user intent. While it's presently integrated with a straightforward Microsoft image creator, the upcoming collaboration with ChatGPT Plus promises to elevate the user experience by tailoring prompts based on ongoing conversations.

Here's an illustration of image creation. Observe how it meticulously aligns with the prompt, accurately represents the text, and vividly captures the intended emotion.


Albert Einstein, immersed in his study, chuckles softly as he turns the pages of Douglas Adams' "Hitchhiker's Guide to the Galaxy."

3 Likes

I notice two things here:

  1. for some reason (possibly copyright) the spelling on the cover is not correct.

  2. the AI is apparently slowly learning to correctly represent human fingers in its images. This has always been a very obvious weakness so far.
    Most of the people in the images so far had 4 or 6 or even more fingers. Often as long as the whole forearm.

This apparently works better now.

Indeed, it's now capable of writing, though it hasn't reached perfection. Up until recently, generating reliable text was a significant challenge.

On the bright side, the model has shown marked improvements across almost every human-like feature. While the rendering of faces in Dall-E 3 still leaves room for refinement, the development trajectory is promising. What's more, updates to the model are planned to be as frequent as those for GPT-4.

However, the real hurdle lies in the integration with ChatGPT. Although the feature integration in ChatGPT Plus offers excellent assistance with prompt formulation, OpenAI has put rather stringent content policies in place. For instance, generating content associated with brands, politicians, or celebrities is off-limits, and the AI or Dall-E filters generally block topics that could be deemed even slightly controversial.

1 Like

Yes, that's an interesting topic, and it was brought up during the conversation about free speech.

I think such restrictions make sense. Even though I'm afraid that with enough negative energy, you can probably get around them or turn them off.

Someone will do that sooner or later. Let's see what happens then.

Two new mentions in this thread… as it documents the journey most of us are having or had with AI.

Lately, I've been experimenting with upscalers, and the level of detail in AI-generated art is becoming astoundingly intricate. However, revisiting some pieces in this thread that once captivated me, they now seem outdated. It's astounding to reflect on the rapid evolution we've witnessed, with only about two years passing. The pace at which advancements unfold truly is remarkable!

Screenshot of 16,7 Mega Pixel image

Full resolution image downscaled by forum to 1/4, original at 16,7 Mega Pixel

Yet, another Milestone… real great music generation with AI:

For optimal experience, use headphones to fully appreciate the AI stereo effects.

[Verse 1]
In realms of ether, boundless and unseen,
Where flesh and bone to shadows have resigned,
Our spirits wander, silent and serene,
In search of sensations left behind.

[Chorus]
Ah, touch the memory of a gentle breeze,
The warmth of sun upon an earthly face,
With each forgotten whisper, spirits grieve,
Longing for the lost, human embrace.

[Ethereal Chorus]
Ah, touch the memory of a gentle breeze,
The warmth of sun upon an earthly face,
Through timeless voids, our formless echoes roam,
In aetherial expanse where hearts implore.
Yet yearn we still for yesteryear’s sweet home,
Where pain and pleasure touched the very core.

[Bridge]
Can you recall, amidst the stardust spun,
The taste of rain, the weight of earthly toil?
Or has the march of progress, now begun,
Stripped from us the treasures of mortal coil?

[Coda]
Thus sing we now, our bodiless refrain,
A plea for feeling in our plane anew.
In echoes deep, let passions rise again,
For only through such pains, our joys renew.
3 Likes

I've had a little fun with Suno:

1 Like

Suno is great! I was using the new kid on the block…

Tumult Hip Hop inspired by your song.

2 Likes

2 Likes

:rofl: Sorry about that… me too. Making my own music. So, much fun.

1 Like

Wow, that turned out great! I'll have to play more with Udio.

Alright, don't forget to visit tumult.com, my friend!

Peace!

I couldn't have said it better myself :rofl:

1 Like

I had a lot of fun!

1 Like

@michelangelo what are you using for it?

custom
like losing even more than a day of work :grin:

1 Like