Some casual chat (Patreon)

This style is a bit different from the preview. Because in the end, I decided to stick to the original work's four toes and fleshy paw pads. I felt that if the furry characters also used the usual, generic human soles, it would look too weird. Also, I ultimately went with a style that features finer line art. Everyone, remember to join Discord; just before the comic update, I'll release the work-in-progress versions first. You can also get a glimpse of what my usual manual image editing workflow looks like.

While working on the XXL Woofia character series, I found that if you want the best results, sometimes it's best to train your own LoRA models. Occasionally, LoRAs in the AI community come with some strange issues—like failing to train the details of the character's clothing, or using someone's artwork which causes the generated images to consistently carry that artist's personal habits. When training, I only use the original character reference images and generate clothing detail assets using closed-source models, which avoids these problems. Maybe I can share some of these models later.

For the next human daddy update, I'm planning to do characters from Okinawa Slave Island. Some people really want to see muscular, beefy bears, and I suddenly realized it's actually been a while since I've done that type. I know the response to that audio-enabled video was just average before; LTX Video definitely has obvious flaws, and the audio it generates sounds too fake—the only bright spot is probably that it can generate 10-second long videos (any longer and the video visuals break down). But the good news is that there will be an open-source Chinese AI model coming out later called MiniMax, which might bring a lot of surprises then.