is AI gonna take our jobs?

Doug Bee said:
https://youtu.be/ITq-mG67qiE
Click to expand...

Made in 2 days from initial (human) idea to finish.
Click to expand...

90 seconds of animated video in two days is pretty impressive.

I remember how in the late 80's/early 90's, following the great commercial success of the "Simpsons", rival TV networks went after the animated market with such fare as the "Fish Police". Most of those programs failed. But now they can be produced very inexpensively. This trailer may not look like a live action flick but its image quality is more than sufficient for a streaming feature. And, once creators get used to the various quirks of this software, an experienced pro with an arsenal of templates should be able to make a full 90-100 min feature within a month or so all by himself. Prorate it to a group and one can start and finish a film per week. Without any cameras, Fresnels or microphones. Or actors in their trailers. New era indeed.
 
Yeah but it’s totally derivative of Baz Luhrmanns Gatsby movie down to the graphic of the logo at the end. Without that refernce what would this be?

This is the problem. It’s a mash up but nothing is original.
 
Doug Bee said:
Yeah but itâs totally derivative of Baz Luhrmanns Gatsby movie down to the graphic of the logo at the end. Without that refernce what would this be?

This is the problem. Itâs a mash up but nothing is original.
Click to expand...

Yeah that's the nature of how LLMs work. In theory it might be able to accomplish something more novel and original with the ingenuity and care of the creator. So, it's still all about the creator. But in the most extreme outcome, you could have what looks like a Hollywood film made by a single person, just like some entrepreneurs fantasize that they'll be able to operate a large company by themselves using AI.
 
DLD said:
90 seconds of animated video in two days is pretty impressive.

I remember how in the late 80's/early 90's, following the great commercial success of the "Simpsons", rival TV networks went after the animated market with such fare as the "Fish Police". Most of those programs failed. But now they can be produced very inexpensively. This trailer may not look like a live action flick but its image quality is more than sufficient for a streaming feature. And, once creators get used to the various quirks of this software, an experienced pro with an arsenal of templates should be able to make a full 90-100 min feature within a month or so all by himself. Prorate it to a group and one can start and finish a film per week. Without any cameras, Fresnels or microphones. Or actors in their trailers. New era indeed.
Click to expand...

Some of what you're talking about isn't really novel. You could do the exact same thing already using video game engines. You could make 90 seconds of animated video in 2 days. There's a reason why that type of thing hasn't taken off and displaced traditional Hollywood production. Maybe AI will. (More likely I think it just gets incorporated into what people were already doing.)

But I agree that the economics are determinative. That's why there are so many successful rappers and DJs in the music world today. It's easier to make a living when you only have one person to pay. When I listen to music from the 60s and 70s, I often think to myself that music used to be way better. (That's why modern musicians are constantly sampling and reusing music from that era.) And I'm not just being nostalgiac for my childhood because that was before my time.
 
I was reading about "Casablanca" (1942) a few days ago and how Max Steiner got into a conniption due to a presence of a third party song that became a part of the flick. Steiner practically invented the genre of the film soundtrack (24 Oscar nominations) and he went to the WB bosses and asked for the whole scene to be removed or reshot. They obliged but then put up the excuse that Ingrid Bergman already cut her hair and it wouldn't look the same with a wig. And so, Dooley Wilson's version of "As time goes by" stayed in.and became a classic of all classics.

The point being - Steiner held immense clout at WB because of his immense contribution to the process of music on film. And that was about fifteen years into it. The AI - or whatever we call AI - is still at the Jazz Singer stage. Someone will write different software, someone else will make more powerful chips, yadda, yadda, yadda ... to quote the most recently diseased Tim Bachman, "You ain't seen nothing yet".
 
BTW, live voice editing of images would/should/ought to be the next step - something like "Elsa sees Sam and crosses over to his piano". And images are created in front of your eyes. And, if you don't like them, you "rewind" and add "walks gingerly" or "walks decisively".
 
DLD said:
BTW, live voice editing of images would/should/ought to be the next step - something like "Elsa sees Sam and crosses over to his piano". And images are created in front of your eyes. And, if you don't like them, you "rewind" and add "walks gingerly" or "walks decisively".
Click to expand...

yes definitely, although I sort of thought that the text-to-image makers would have already added a natural language editing option for images. you ask for a photorealistic image of Biden bench-pressing and then you ask it to tweak the output. Maybe the more involved generators (like Midjourney) let you do this, but from what I've seen on Dall-E and others, you have to start over with a new image prompt. Or select a portion of the image to erase and tell it what you want there instead. I would have thought that chatGPT-like capabilities would start winding their way into every area of computer use already, but I haven't seen it happening

I'd also add that past performance is not predictive of future advances. experts including the ceo of openai have said that LLMs are reaching the peak of what they can do. of course, there is a vast number of applications which have yet to exploit or perfect. but one shouldn't expect future iterations of chatgpt to keep making massive improvements.

another funny comment I've read is that AI researchers first perfected text-to-image and now can turn their attention to text-to-video. there's some truth that video tech will build on still image tech. but none of these companies was holding off on developing video generators. they were working on image and video simultaneously. and just as with current photoshop and image editing vs video editing, it's reasonable to expect that video generating AI will always have more limitations than still image generating AI
 
Back in the late 90s/early 00's, there was an animated TV show called "Home Movies". I was quite a fan. It stayed around for 5 years and has an 8.2 rating on IMDB (that's very high)


Now, look at its animation style. Couldn't AI do this right now? And for nothing?

And then all you need is a screenplay.
 
Below is copy of my conversation with AI robot.

Question #1:
What conclusions can be made from surface transmittance in optical lenses?

Answer #1:
Surface transmittance is a measure of the amount of light that passes through an optical lens. It is an important factor in determining the quality of an optical lens. The conclusions that can be made from surface transmittance in optical lenses are that it affects the amount of light that passes through the lens and therefore affects the quality of the image produced by the lens. The higher the surface transmittance, the better the image quality produced by the lens. Conversely, if the surface transmittance is low, then the image quality produced by the lens will be poor.

Question #2:
What are some factors that affect surface transmittance?

Answer #2:
There are several factors that can affect surface transmittance in optical lenses. These include the spectrum and polarization of the emitted light, the position of the point source, the area of the surface under stress, surface finish, size of internal flaws, glass composition, surrounding environment, and the amount and duration of the load.

Question #3:
Does a larger diameter lens have more surface transmittance than a smaller diameter lens?

Answer #3:
The diameter of a lens does not affect surface transmittance. However, the diameter tolerance is an important mechanical tolerance that must be considered if the optic is going to be mounted in any type of holder.

Question #4:
How does the area of a lens affect transmittance?

Answer #4:
The area of a lens does not affect transmittance. However, the percentage of light that passes through a surface or material is called transmittance. If the value of percentage transmittance is high, it means that the surface will allow more lights to pass. Similarly, if the value of percentage transmittance is low, it means that the surface absorbs more light.

Question #5:
How does the surface area of a lens affect transmittance?

Answer #5:
The surface area of a lens does not affect transmittance. However, the percentage of light that passes through a surface or material is called transmittance. When light hits a lens surface in air normally, the percentage of light reflected at each surface is given by: R = (n - 1)2/ (n + 1)2 * 100%. Thus a material of refractive index 1.5 has a reflectance of (0.5/2.5)2*100 = 4% per surface.

Question #6:
If transmittance is the percentage of light passing through a surface how is surface area not affecting transmittance?

Answer #6:
The surface area of a lens does not affect transmittance. Transmittance is the percentage of light that passes through a surface or material. When light hits a lens surface in air normally, the percentage of light reflected at each surface is given by: R = (n - 1)2/ (n + 1)2 * 100%1. Thus a material of refractive index 1.5 has a reflectance of (0.5/2.5)2*100 = 4% per surface.

Question #7:
What factors affect a lens resolving power?

Answer #7:
The resolving power of a lens is determined by several factors. The optical design formula, the number of lens elements, and the tolerances allowed for that lens design are all important factors that can affect a lens’s resolving power. The resolving power of an objective lens is measured by its ability to differentiate two lines or points in an object. The greater the resolving power, the smaller the minimum distance between two lines or points that can still be distinguished. The larger the numerical aperture (NA), the higher the resolving power. This resolving power intrinsically depends on the properties of the light source being used to illuminate the specimen- namely, the wavelength of the light and the size of the opening of the microscope’s optical components.

Question #8:
What is numerical aperture?

Answer #8:
Numerical aperture (NA) is a measure of the acceptance cone of an objective lens. It describes the range of angles within which light that is incident will be transmitted. The numerical aperture of a lens objective is a measure of its ability to gather light and resolve fine detail at a fixed object distance. Image-forming light waves enter the objective in an inverted cone. The numerical aperture of a lens objective is the measure of its ability to gather light and to resolve fine detail while working at a fixed object distance.

Question #9:
What are the units of numerical aperture?

Answer #9:
The numerical aperture is a number (without units) that indicates the angles over which light can enter the lens. The larger the numerical aperture, the more light can enter the lens, and the more detail you can see. When more light enters the lens, the image becomes brighter. Numerical Aperture (NA) is a value (often symbolized by the abbreviation NA) originally defined by Abbe for objectives. It is given by the simple expression: µ Numerical Aperture (NA) = n × sin ( µ) or n × sin ( α)2.
 
Here are some practical implications:

1.) Put a FF lens on a MFT camera using an adapter.

Did you lose light? Yes. You lost the light volume that overspilled the sensor. Did you lose resolution? No. Light transmittance is the percentage ratio of the light volume to the image area at the lens objective. Resolution is also a ratio, lines divided by millimeters. Transmittance = resolution. Transmittance did not change, and neither did resolution.

2.) Put a FF lens on a MFT with a speedbooster.

Did you gain light? Yes. You recovered the light intensity that had overspilled the sensor without the booster. Did you gain resolution? No. The volume of light increased but the area of the image at the objective lens also increased. Transmittance was unchanged.
 
Tom Roper said:
Here are some practical implications:

1.) Put a FF lens on a MFT camera using an adapter.

Did you lose light? Yes. You lost the light volume that overspilled the sensor. Did you lose resolution? No.
Click to expand...


Welllllll yes in theory however....

It's not uncommon for flares to occur when there's extra light bouncing around between the rear of the lens and the sensor surface. I started in the film days and it was very common for some to use 35mm format lenses on Super 16 cameras. The idea being they were somehow better lenses. Except the larger projected image circle would end up bouncing around on things behind the lens and cause strange flares off of point source, typically near the frame boundaries.

They tend to be uglier and less conventional and there's an argument that extra random secondary light leads to less resolution through altered and lowered contrast.

I think there's a GREAT danger right now with AI. People are using it like it's Google, except it's kind of fibbing, not telling you everything, or even outright lying to you.

https://www.washingtonpost.com/technology/2023/04/05/chatgpt-lies/
 
the prominence of prompt engineering may be fleeting for several reasons. First, future generations of AI systems will get more intuitive and adept at understanding natural language, reducing the need for meticulously engineered prompts. Second, new AI language models like GPT4 already show great promise in crafting prompts — AI itself is on the verge of rendering prompt engineering obsolete. Lastly, the efficacy of prompts is contingent upon the specific algorithm, limiting their utility across diverse AI models and versions.

So, what is a more enduring and adaptable skill that will keep enabling us to harness the potential of generative AI? It is problem formulation — the ability to identify, analyze, and delineate problems.

Problem formulation and prompt engineering differ in their focus, core tasks, and underlying abilities. Prompt engineering focuses on crafting the optimal textual input by selecting the appropriate words, phrases, sentence structures, and punctuation. In contrast, problem formulation emphasizes defining the problem by delineating its focus, scope, and boundaries. Prompt engineering requires a firm grasp of a specific AI tool and linguistic proficiency while problem formulation necessitates a comprehensive understanding of the problem domain and ability to distill real-world issues. The fact is, without a well-formulated problem, even the most sophisticated prompts will fall short. However, once a problem is clearly defined, the linguistics nuances of a prompt become tangential to the solution.
Click to expand...
https://hbr.org/2023/06/ai-prompt-engineering-isnt-the-future
 
In term of video production, I wonder how long it will take to have a "made to ready" text-to-video program that can convert any green screen background into a complete multiple actor scene which will look as good as a very elaborate (and expensive) clip from a show like The Mandalorian. Forget the LED screens, confine full production to a computer and call it a day.
.
 
Apparently Elon thinks the first trial for a brain chip implant will take place this year...something I (and probably millions of others) said would eventually happen (not much of a shocker either) but didn't expect it yet!

When it artificially & technologically rains it pours...
 
the irony of it all is the self-driving cars will probably fully happen after the brain chips are well on their way now that the FDA approved them last week
 
You must log in or register to reply here.
Back
Top