Devlog: Experimenting with the style
- Style 1 32
- Style 2 15
- Style 3 13
- Style 4 84
While I'm unable to work on the script, I can still work on the pictures, which led me to experiment a bit with the style.
The problem is that the model I'm using right now is completely unable to generate any decent backgrounds, so any scenes that require anything more complicated than "a bed", "a wall" will take a lot longer and still look... not the greatest. So, here's what I've been working on:
This is an image using the original model, I really like how the character look, but
it can be a bit of a challenge to not make her have big boobs. And background is... could be better
1
This one is the image using different, more popular model, it has that "detailed ai" look to it, some people seem to really enjoy it. Also backgrounds are a lot more detailed without additional prompting, which is really nice
2
This is the same popular model, but with the lora that removes shadows. I must admit there's something to it. The sharp look combined with detailed backgrounds has appeal
3
And finally, combined approach. Using one model for backgrounds and another for characters. Seems to combine the best of both worlds with cool looking characters and detailed backgrounds. The downside is that results can be inconsistent and it can be a bit tricky to naturally integrate them together, also it takes a lot longer, so if I stick to it will inevitably affect the developing time. All in all, I need to test it more.
4
So, what do you think? Which style speaks to you the most? Vote in the poll, leave a comment