Recent posts

#1
Games / Re: Connections
Last post by jt512 - Today at 06:23:30 PM
Connections
Puzzle #259
🟪🟪🟪🟪
🟨🟨🟨🟨
🟩🟩🟩🟩
🟦🟦🟦🟦

Blues more or less by default.
#2
Podcast Episodes / Re: The Skeptics Guide #972 - ...
Last post by arthwollipot - Today at 05:26:17 PM
Yep, but it's here and it's here to stay. The only thing we need to do now is work out how to live with it.
#3
Podcast Episodes / Re: The Skeptics Guide #972 - ...
Last post by daniel1948 - Today at 04:52:43 PM
Thanks. That's very interesting. Not something I have an interest in doing, but it's interesting how much more flexible those apps are, compared to what Google offers online for randos like me on the internet.

I still think it greatly expands the opportunities for disinformation.

FWIW, I've been saying for a while that the other big way AI will continue to make our lives worse will be replacing human service providers with AI versions that are not quite as good as an actual human.
#4
Podcast Episodes / Re: The Skeptics Guide #972 - ...
Last post by Erog - Today at 04:09:17 PM
Thank you I did not know.

For AI.
Well web-based AI is not only wildly different but also lacking in almost ALL of the controls.  your experience is not surprising but really does not actually show what's possible.  Also, another thing people think is that actually making the very cool images shown is "easy"... you can mash the render button 1000s of times and hope you get something you like ...  but to get what you want like anything else you have to learn what you are doing, and how to use the software and these online sites really limit people to just playing.

There are many ways do download and run "text to Image" AI locally and have good UIs to control them with.
A1111 and ComfyUI are a few..  This allows A lot more control, so much more it's hard to describe... Think disposable wedding camera vs a digital DSLR  . . . with Lightroom/photoshop. Yes, it's that much .. the disposable can get lucky and get a good shot but the pro will get what they want.

For instance, I can draw out basic elements (think child drawing) to show the AI where to put the sun, mountains, lake, road etc background.  Pose a stick figure for the subject to get the exact pose you want.  render a batch of very similar images all with the very slightest differences varying the weight of different prompts you are interested in changing and do this automatically. You can Paint out a part of the image and give that area a new prompt that will merge with the original. Have different prompts for different parts of the image and have it assemble automatically or make masks for you to do it manually.
And that is a short list.

The big thing is like you said this really is NOT AI as most people think it is and will not/ can not do anything more than what it is doing. ie text to speach or text to image is just that.

politically and media-wise? IMO sure it will be an arms race like every other... but people or the media really do not NEED AI images, sound or video to do its job. IMO not even close as media already does it and has for decades and people eat up the crap just as happily..
   
#5
Podcasts / Re: Chicago Extravagansa Aug 1...
Last post by puddingface - Today at 02:32:40 PM
I don't have tickets yet but planning to. We are from far west suburbs so likely will stay the night in Chicago. Might go to the live podcast, but we're also going to extravaganza and podcast+ in Dallas this spring so not sure.
#6
Podcast Episodes / Re: The Skeptics Guide #972 - ...
Last post by daniel1948 - Today at 01:58:46 PM
Quote from: Erog on Today at 11:51:08 AMHate to say it guys its not that hard to make a limited change.  In fact there are many ways to go about it from the simple of using the same seed and adding a prompt or weighting that prompt more.
To inpainting which gives you a crazy amount of control and allows you to keep what you want and change what you don't.  or have generation happen with a previous image, a stick figure for the pose, or even just a sketch for the seed image. 
And many other ways.

It inpainting would be possible in a movie but you would have to do some more work. (effectively rotoscoping, but with the added benefit that you can also generate your own mask/depth map automatically) 

Crazily the technology to make the video is not far from what you wanted.  ie using the previous frame as a basis. and the problem they surmounted was the jumpiness that happened when you did not.
there is even the ability to have two separate prompts being used simultaneously directing different parts of the scene and then unified.

All cool stuff. and just a FYI.

Kinda looks like you're addressing the podcast hosts. They don't read this forum. But you can email them, and they do read, and sometimes respond to, their email.

I played briefly with Bard's image creation. It was very frustrating because it seemed to focus on one point in my prompt. When I asked for A and B it just gave me A. When I said, "No, I want A and B" then it just gave me B. When I gave it a picture of me and asked it to incorporate that in the image, it refused. I guess they're concerned that people will try to make compromising pictures of others, so that's understandable. But I wanted a picture of me as Daniel in the Lion's Den, and it wouldn't make it.

It wasn't much fun.

Earlier I tried having a conversation with it and that was no fun either. I was not impressed.

A.I. will be a disaster, not because it will become Skynet, but because it will create chaos in politics by making it much more efficient to promulgate disinformation. And it cannot be stopped or controlled because the internet spans international borders.
#7
Forum Games / Re: Visual Counting
Last post by DevoutCatalyst - Today at 12:26:24 PM
Sorry Leonardo, I'd rather have a Volvo.
#8
Forum Games / Re: Visual Counting
Last post by Morvis13 - Today at 12:18:24 PM
#9
Podcast Episodes / Re: The Skeptics Guide #972 - ...
Last post by Erog - Today at 11:51:08 AM
Hate to say it guys its not that hard to make a limited change.  In fact there are many ways to go about it from the simple of using the same seed and adding a prompt or weighting that prompt more.
To inpainting which gives you a crazy amount of control and allows you to keep what you want and change what you don't.  or have generation happen with a previous image, a stick figure for the pose, or even just a sketch for the seed image. 
And many other ways.

It inpainting would be possible in a movie but you would have to do some more work. (effectively rotoscoping, but with the added benefit that you can also generate your own mask/depth map automatically) 

Crazily the technology to make the video is not far from what you wanted.  ie using the previous frame as a basis. and the problem they surmounted was the jumpiness that happened when you did not.
there is even the ability to have two separate prompts being used simultaneously directing different parts of the scene and then unified.

All cool stuff. and just a FYI.
#10
Recipes / Re: An easy-to-make recipe
Last post by bachfiend - February 24, 2024, 06:52:13 PM
Quote from: daniel1948 on February 24, 2024, 06:27:05 PM
Quote from: arthwollipot on February 24, 2024, 06:22:22 PM
Quote from: daniel1948 on February 24, 2024, 05:03:50 PMThe fact that you just dump all the food into one container and nuke it suggests to me that you're not particular about food preparation. You just want it to be easy to make, and easy to clean up afterwards.

To be clear, there's absolutely nothing wrong with that. For some people, food preparation is a pleasure, a pastime to be indulged and enjoyed whenever possible. For others, it's an annoying delay to the necessary function of getting nutritional calories in one's gullet. It's okay either way.

Personally, I'm a little bit of both.

Exactly. I was talking about what's meant by a "foodie."

I am particular about food preparation. I don't want to overcook the very tasty vegetables I eat and destroy their flavours.  I admit 'foodies' are keen to develop new flavours in their food by cooking, with such processes as the Mallard reaction (I generally don't cook at a high enough temperature except in making toasties). But the added flavours are due to melanoidins:


https://en.wikipedia.org/wiki/Maillard_reaction

https://en.wikipedia.org/wiki/Melanoidin

which reduces the nutritional value of the food (not much of a problem - people are eating too much animal protein anyway), and may also have deleterious health effects.

Cooking to enhance flavours might also shorten you life.  More pleasure in a shorter time.