This is a port of my dispatch dataset from LTX to H3. It's trained on the fl2va model. All samples are t2v on 20 steps.

Right now only 3 characters are trained, but I want to add in all of them. I wanted to do this so I could see if H3 can do multiple characters in 1 lora like Wan/LTX does. So far the answer is yes it can, but the voices are strange. Consider this like a proof of concept beta version.

The voices are trained, along with music from the game. But when you have multiple characters in 1 scene speaking, their voices tend to mix. Especially invisi and blonde blazer in the same scene.

The voices also are like 90% there, but still not super accurate. I also have seen this same problem when using the ref base model. It doesn't like to have multiple of the same gender voice in the same scene. I will keep training and tinkering to see if I can find a solution or not.

Style Trigger:

"Stylized 3D cel-shaded comic style."

Invisigal:

char_invisi, a woman with short dark hair accented by a purple streak, wearing a purple jacket over a dark top and distressed black jeans

Blonde Blazer:

char_bb is a woman with long blonde hair and a blue mask, wearing a blue and yellow superhero suit with a yellow cape and a red diamond-shaped gem on her chest.

(no costume, powerless)

char_bb has long, wavy dark brown hair and wears a strapless evening dress with long dark blue opera gloves

Robert Robertson:

char_rr has short brown hair and wears a light blue button-down shirt with the sleeves rolled up to his elbows and dark trousers with a beard with stubble.