SamuZai
ovidius
ovidius

patreon


AI Doctrine

Update: I added poll results, added a point about "no secrets", and added a point about "no paywalls".

Here are the results of the pool. I summarize this as "most of you are in favor of AI" with only a small minority of people who dislike it (22% overall).

I'd like to talk about my guiding principle for using AI: "Do what cannot be done"

Basically, I don't want to generate AI content that an artist can already make using other means. Anything in anime style, illustration, painting, etc could be done by a traditional artist (even animated) so I'm pretty much staying out of that*

Anything in realistic style could technically be done by a photographer (or videographer) + some models, but for the kinds of content we like here, it would be either outright impossible or require fantastic amounts of money/talent to make.

There is also using AI to help me with my own work. Technically I could pay another 3D artist to do that parts that I don't have time for, like the intros, dialogue sequences, etc. But I don't have infinite resources or time, and if AI means I can bring more to you guys, then I think it's worth it.

In summary, I will focus on using AI to:

Additionally:

* there are some cases where certain models/loras ONLY really work with anime or related styles, so I may play with those if the subject matter is really compelling, for example the 3 panel TF sequences.

Comments

no trouble, the answer I gave its a line frequently said by Charlie, in the episode 7, season 1 of two and a half men, give it a search, you'll know what I mean. I use it a lot in discussions

Leonardo Buccioli

I agree with both of you. Most people think of AI as the loads of crap images that bad apples are spamming out on various platforms. But when you find a good site like civitai, when you start to learn how you have to put into it to get good results, you see a different side.

Ovidius Naso

Leonardo, I apologize for that other user's behavior. It is not acceptable for anyone to harass another Patron like that.

Ovidius Naso

This comment is rife with insults and signs of entitled thinking. The fact that you decided to make this public instead of writing to me in private says a lot about your questionable motives. You can use your own platforms to go on your crusade, it is rather uncouth to try to use mine for that. Do not engage with me on this platform anymore.

Ovidius Naso

Most content at this point in time is slop created by using prompts with a generalized "style" created by data harvesting images (an untold portion of which might not be legally obtained) and then serving the generated image as is However, if someone were to use their own work as a data set for a "style guide" and then on top of that, hand edit the generated result to further refine the style and visuals, the end result is much higher quality and has no dubious origins in data Ethical concerns aside, it's the difference of quality between someone throwing a bunch of random veggies in a blender and calling it soup vs someone who hand picked said veggies and seasoned it properly after blending

Narcoleptic Bunny

I understand

Leonardo Buccioli

I've made my personal stance pretty clear in the comment I left on the poll in question, I will not support nor condone any artist who debase themselves and their craft through the use of LLMs and thus building on the stolen work of others. I understand that there are differnt types of LLMs out there and some are more benign than others, however, it is a very slippery slope. One great example of this is the artist "jackthemonkey" who went from having some "AI" generated voice stuff, to inserting it into his art and he's now on to creating entire works in both writing, music, animations and art (feel free to read all these with quotation marks as I don't count any of it as actual art), out of completely "AI" generated content, and charging for it. Sadly it isn't just like "any other tool" like some people seem to think, LLMs only work through the use of stolen content on a scale never before seen. Frankly, an artist should know better... and since it seems like you'll be using "AI" going forward, I'll be pulling my support. v.v

Meade Marmalade

I'm favor to AI use, it's just a tool like any other

Leonardo Buccioli

I'm ok with that.

Gt_the_hack

You know my position. I’m exited to see what you can do with a new tool!

Jack the Monkey

I definitely agree with the use of AI tools to assist. I think the hang up with most people have so far is mostly picturing the current AI content that's circulating at this time, which is probably the first thing that comes to mind.

PineappleVR46

Yes, the idea is to use it to make the intro and dialogue parts faster. The lip sync that I have access to is pretty primitive, and there aren't really tools in Poser for doing idle animations, so I end up doing all that by hand. I would love to have more story and talking, but that probably won't happen unless I can find something else (AI, VaM, mocap, etc) to make it more acessible

Ovidius Naso

Using AI to "enhance" your own work I think is fine, especially if the model is trained on your own existing body of work I would be against it if the goal was to make quick and easy slop, but it sounds like you intend to use it as an actual tool for work

Narcoleptic Bunny


More Creators