Dear C-Keen.
I’m also seekin’. Seekin’ answers!
I’m curious as to where the line is drawn in mixed cases, since I’m experimenting with both pixel art painting and txt2img/img2img models, and have had thoughts of combining the two to generate extensions of my own works.
Relevant examples of unclear cases to clarify:
- Starting from an original work as the core “seed”, outpainting the world by expanding the canvas around it, continuing the work based on that input.
- Same as 1., then manually adding further edits after doing the txt2img outpainting.
- Starting from own original non-generated work of art, using some style transfer to generate a much similar edition in a pixel art style.
- Starting from own text prompt to generate some pixel art, then manually editing that.
I understand that this is about appreciating artists. Pixel art is a craft with a rich history, and it’s a dogma, which can be helpful and fun for artists to get going. Generating the works seem meaningless from that perspective, but I’d argue, nonetheless, that all of the examples above (not just any gen. pixel art) are a continuation and natural development of the craft, which has already been changing through the times. From analog embroidery through digital ages of computers and software. Should we keep insisting on crafting the traditional way, or can we use modern tools? How many colors are allowed, if we want to stick to earlier pixel art traditions here?
As I see it, all of the listed examples require a certain degree of artistic work, and couldn’t have existed without that, but use txt2img or img2img generation as tools as part of the artistic process, experience and output. One could argue that these pieces of art represent the current state of the craft, and that artists working with these tools should not be excluded from here. But on the other hand I fully respect the existence of the opinion that they should, to collect and adore only the 100% manually painted works within this community.
I find it easy to understand your decision regarding 100% generated pixel art, but the next question that rises is how will you point out the generated ones that are being posted? Except for obviously incorrect stuff like a gradient “pixel” challenging the technical limitations of real works. But, then again, what if an artist manually draws such glitches by faking a lower resolution than the actual image file and breaking it, shouldn’t that be allowed?
Please don’t get me wrong. I’m not here to oppose any decisions or defy one ruling over another. I just think these questions are interesting to ask. Have you thought of how to actually enforce this? If you don’t want 100% generated works here, how will you make sure to find those that are, and only those? Or if you want absolutely 0% generated pixels in here, how will you find the ones with any at all, like in the five examples I provided? The community hivemind? How to avoid false accusations?
I’m looking forward to hear your thoughts, whatever they are. Best regards.
[…]
But again, that’s specific to that field.
[…]
But that’s getting too technical.
[…]
But still, without context, it’s hard to be precise.