They may lose, but the net effect may be that many artists "opt out" from future models, effectively crippling them unless the technology improves enough to compensate. Apparently, a reduction in the amount of data used for training models 2.0 and above is what hurt their ability to draw people.
Your skill has improved dramatically. And probably your models. Make sure to learn about Controlnet and Openpose extensions. If you can install them, do it! It will revolutionize the quality of the work you can put together for you will be able to control the character pose better, layer in the background AND affect the lighting through img2img in a tenth of the time you currently do. All inside the software. It's terrifying and awesome to realize what this software is doing now, and even moreso in a few months. The software capability seems to be following Moore's Law.Not a lot of time to work on this lately, but here's a recent version. I'm going for a more realistic moon lately and trying to represent a scene at the very end of the book (extra points if you know who the man is).
View attachment 13930
Thanks! I read up on the extensions you recommended and they sound great. I'll have to check on how to run them on a Mac.
Thanks for the warning! I think most of these models would probably have to be converted to run on a Mac, although I could be wrong.
Thanks for the comments! The novel is a historical fiction thriller with American ex-pats, set in the deserts of the Middle East; the crescent moon is the symbol of Islam. An earlier version of the cover featured a souk, somewhat similar to what you're describing, but I decided to go with a scene from the book in the end.
It looks like I should be able to run at least some of the ControlNet functions fairly easily; I'll check it out tonight.
They say 2.1 is better, but I have not jumped to it yet. So many problems surrounding it. I'll wait till either 2.2+ or 3.0 to come out... if it comes out.
It is, and for a small team of people. I was just showing how far the tech had come in 6 months.This AI-video implementation currently looks rather time consuming. Give it a few months.
Even so, I have to start the next book cover real soon. Thanks to the new tools with easypose and controlnet, I should be able to get what I want faster than I did last time and with much more accuracy.
I hired a freelancer on Reedsy. I've also used a graphic designer on a freelance site called Upwork and had good success (though that was not for a book.) I've never used Fiverr but that's another freelance site. I had a good experience on Reedsy. I could look at artist's portfolios and pick a few I liked then receive quotes for the work. If you're looking to spend a little less, perhaps you could find a college graphics design program and some students who would do the work to build their portfolio?
