DreamLyrics™ Play-by-Post Forums Modern/Historical DUC: Durham University Campus OOC:Student Chatter

I am around and available.

I'm around an ready to go.

Busy today but I'll try and get a major move post up this week!

Haven't forgotten you guys. Christmas prep and looking after Tusk is taking up a lot of my time. Plus we have a visit to an osteopath this afternoon. Please bear with me.

No worries, Gypsy! Have fun with Tusk and Christmas prep! Doing a work event here. Who killed the Gingerbread Man crime scene event.

Sounds like fun!

Don't sweat it. I've been crazy busy. And to think that I'm retired!

I know. Wonder how I had time to work!

OK, Tusk has improved quite a bit and I am starting to get my head above water. I have an opportunity on Sunday to get a start post up. Then I have a new computer to set up but can't start that until Tusk has done his bit which will not be for about a week! Fingers crossed!

Just glad to hear that Tusk is doing better. That is what is important.

Agreed!

I've been playing with DiffusionBee AI to see if I can create character photos. It's all very weird and a high learning curve to get something that looks decent. Here are a couple examples:

Original magazine photo of Sora Choi, model: https://www.dreamlyrics.com/forums/gallery/725/full/18892.jpg


AI attempt to reconstruct: https://www.dreamlyrics.com/forums/gallery/725/full/18893.png

It took me between 8 - 12 attempts to get this one.

Thoughts? Has anyone else played with this tool?

I have read some articles about AI generated art in general, but have not tried any yet. It is certainly an interesting set of tools, but there is a debate from artists and other content providers that remains to be seen how this will all fall out.

Pretty good attempt though.. impressed.

Has anyone else tried Night Cafe? I've played with it a bit and for face shots it's fun. https://creator.nightcafe.studio/

Dugan,

It was actually after reading several articles on the AI debate that prompted me to take a look. I agree that there are some major ethical dilemmas:

Context of use: There is a major delineation between commercial use (and sale) vs. personal use (including fan-fiction, such as ours on Dreamlyrics.) Historic and/or journalistic representation vs. illustration also needs to be very clearly defined.

Emulating the art-style of copyrighted materials - by feeding those specific art examples into the AI model - and then commercially charging for use of that AI seems to be a potential violation of copyright law. This also could include the many sources/people themselves (models, celebrities, etc.) that have been input as examples.

[NOTE: I have been using the free, open-source AI engine DiffusionBee as opposed to a pay-service like Lensa or Lexica.]

I think the Charlie Warzel quote from Dugan’s link hits these examples: “Like others, I also have questions about the corpus used to train these art tools and the possibility that they are using a great deal of art from both big-name and lesser-known artists without any compensation or disclosure to those artists,” Warzel writes.

However, beyond the legal ramifications, I’m also seeing how the manifestations of sexism, racism, etc. (And truthfully probably remind me of my own personal biases that I continue to work through.)

Some interesting observations in these regards testing out prompts from the AI community:

Christina Hendricks (Caucasian-American) brings up more frequent and accurate look results than a model like Sora Choi (Korean) or Micah Gianneli (Filipino-Spanish-Australian).

So many big boobs!!!! Sigh….

More accurate rendering in painting and illustration styles than realistic photography. Could this be a tech limitation or sourcing?

I’m curious that one of the biggest critics I’ve read regarding AI art generation has been the artist, Greg Rutkowski. However, his D&D style art is super common in the AI prompts I’ve seen. {Return to legal arguments…}

All and all it has been an interesting experiment for someone that doesn’t have artistic skills. For my two-cents of use in fan fiction, it’s about the same as clipping webpics to represent a character. However, I hope to stay on the right side of this topic - and give fair due to the sources when appropriate.

Would love to continue to hear more thoughts.

Mina,

I'll take a closer look. Thanks for sharing!

- L.

OOC: sorry - completely swamped with work, including on-boarding new staff. I'll add something just to keep things going.

I have put up the competition ratings & rules.

As Skye and Levon are fighting first, can you eat make 3 rolls of 1d4? We will then add your rating and determine the results. I am happy for you to write the posts describing your bouts.

I didn't see a dice roller set up, so I used FFH:


2023-05-08 00:01:21
n/a die roll for DUC
Lionhawke rolled 1d4,1d4,1d4 and got 4, 1, 2

2023-05-08 00:00:56
n/a die roll for DUC
Lionhawke rolled 3d4 and got 12

I thought to save time and did the 3d4 first - then thought there might be a reason you wanted them separate. Take whichever you wish.

Thanks for your patience while I was working and then family obligations.

Yes, as each roll is for a round. Need to roll another d4 as you aced the first roll. And no worries.

Dice roller is there now

2023-05-08 18:00:17
Skye
Lionhawke rolled 1d4 and got 4

Thanks. Need Levon's rolls now