AI- and procedurally-generated artwork have made great strides as of late, and already there has been lots of excitement about how that might change the field of game development. There is much that has been said (and remains to be said) on the topic as it pertains to us here, so we’ll start with brass tacks.
Moving forward, we’ll be taking the following stance:
- You are welcome to share and post AI art on this forum so long as it is clearly marked as such.
- FEU does not recognize ownership of AI-generated art, and will treat it as F2U and F2E with no attribution necessary. That said, using an AI-generated piece as a base for your artwork is perfectly fine, as it is for any other F2E modification.
We recognize that point 2 will be controversial, with lots of edge cases. We are intentionally taking a hardline stance for the time being, with the expectation of doing some fine-tuning as the field develops.
A good rule of thumb is that, if I could take your model, references, and prompt(s) and create the same piece, you haven’t done enough to make it “yours”. A (non-exhaustive) list of things that we do not consider to pass this bar is:
- Prompt fine-tuning
- Reference selection fine-tuning
- Repaletting
- Formatting for GBA insertion with frames
- Cropping
To put it more succinctly – While you are not ever obligated to share pieces that you’ve generated, should you do so, we will automatically consider them to be F2U and F2E unless you have somehow creatively modified it in a way unique to you.
Note that these rules will also apply to things like FEMapCreator’s automap function and the Portrait Creator’s random mode moving forwards.
Of course, there is lots of philosophical discussion to be had about the ethics of training data, creative endeavor, and so on to be found on so many platforms with so many caveats that there is no way that I can cover it all in-depth here. If you’re interested in my personal thoughts, or if you generally have questions about the policy, please feel free to reach out.