Realistic AI images

[vbl2007]

Trying to keep up with the times, I installed a program for working with and generating images and videos using artificial intelligence (AI).
Using and testing various AI models and corresponding prompts, I tried to achieve the most realistic images and representations of people, interiors, events, and settings in various conditions.
How and what I got is up to you to judge.
Enjoy.

 

[JRR]

Amazing what Ai (and you) can do.

Many, at quick glance, didn't look Ai produced, a few did

While I personally didn't care for some of the images, they were are all well done

Congrats

 

[vbl2007]

8 minutes ago, JRR said:

While I personally didn't care for some of the images, they were are all well done

JRR, thanks for the feedback! Indeed, quite often the AI doesn't understand (can't, won't, or imposes its opinion), and you have to repeat it several times to achieve the desired effect.

[mhwarner]

What program are you using?  I personally am more interested in using AI to automate my own  still photos into videos and several programs have done a decent job but most have limits which change on a regular basis.  I don't mind paying if I could find one that was consistently good with any limits stated up front.  

Also, I really like many of the transitions used in this video.  Have you shared any of these in the past or would you be willing to share now?  Thanks in advance!

[vbl2007]

15 hours ago, mhwarner said:

What program are you using?

mhwarner, The computer program for working with and generating images, videos, and audio using AI on prompt is called ComFyUI. It's free.
The program itself is relatively small, 5-5.5 GB, but that's only half the battle.
The sheer number of models and libraries it contains is what takes up the most space.
Models for anime, children, men, women, formal models, erotic and semi-erotic, for different body parts (separate models for the face, arms, legs, body, figure, gait, etc.), body structure for men and women, children, animals, and so on.
And all this is separate for images and videos.
All this takes up a huge amount of space on your computer. My relatively small selection of models took up 2 TB (terabytes), and that should fit on a single drive.
But that's only half the battle.
The program, like the AI itself, is very sensitive to computer power.
Weak, and even mid-range, computers are not suitable for modern AI. My images took 0.5-5 minutes to generate, depending on the AI model's complexity.
But videos... from 10 minutes to over 2.5 hours, and I didn't expect them to last any longer and cancelled the generation.
When generating long and large videos, the video card heats up to 90 degrees Celsius, which is very harmful.
Moreover, (according to some rumors) the program was developed for new and modern high-performance video cards, such as the Nvidia RTX 4090-5090 and newer.
And your computer's memory (RAM) should be at least 32 GB, and preferably 64 GB or more.
Therefore, for image generation, photo editing, and short video generation, it's best to use AI-powered platforms with processing power thousands of times greater than any computer. Sites like Qwen and Grok offer a wide range of plans and payment methods.
There you can improve photos, animate them, colorize old photos, and much more.
Good luck!