Echoes Through the Void

Enhancing Original Content

Echoes Through The Void is an original novel written by DaoistShadowKeys, without AI.

As the novel chapters were created, ideas for enhancing the narrative through visual illustration also came into being, however, working on a single craft (writing) is already difficult enough, adding illustrations or visuals remained a frustratingly elusive endeavour as it limited what could be done creatively.

Fast forward to today and generative AI technologies have become very commonplace, with their current capabilities text, image and video generation are possible, trying out those technologies revealed that regardless of their impressive capabilities they lack that ethereal quality that the human touch provides with its full experience, creativity, context and uniqueness.

Image generation in particular has advanced to a point where I think it could be a useful tool to express some of the ideas I couldn't before, in fact many authors are already jumping into this train, however, most of the output seems to be AI Slop, so what is different this time around?

Looking at clearly AI generated content both at WebNovel and Amazon Kindle store, it is clear that a lot of the slop overly relies on AI to drive the work and narrative, the human touch is lost and it no longer serves a purpose to express any creative intent.

The way I'm approaching it is to use image generation to provide visuals for the novel, the novel itself is the pure creative expression while generated images only serve to illustrate, "but DaoistShadowKeys a lot of slop books also do that, how is this different?" you may ask, and I also noticed it, but I saw a specific pattern which is again the lack of human touch and care, very often slop images tend to look out of place (not just the extra limbs) but it is clear the low effort, as if somebody just asked the machine to "make some images for the story" and that was it.

What I'm refining is a creative process where generative technologies are just one tool, each image generated takes effort to define, the location, the characters, the context, for each character I have spent time refining character sheets, for each image generation it takes several tries to get exactly what I envision so it takes time to adjust the composition, defining scenes often takes characters, environments, objects and sounds to be combined, very much like creating a game but your output is a static image or a short clip.

Even so, the process is still early and while the outputs are much better than expected, I'm still not completely satisfied, generative technologies are good enough in some areas but in others they still lack and that is reflected still in the outputs regardless of the extensive work that goes into it, generative technologies are difficult to exactly steer but it is getting better, hopefully as time goes on I'll also document the process.

In short, I hope you enjoy reading the story as much as I did the first time I saw the fully functional version, I can expect with time to get better at this and provide more features like short video clips, community features (think comments) and more.