LTX2.3/MiniMax Free Generation!

I turned off generation charging. I didn't know it was on must have been a autojoin thing. I may start doing early release just to help cover some small costs but not per generation.

MINIMAX Notes:

This is a preview early release of the H3 version of daring's deepthroat lora. Use I2V or FL2VA while the T2V is "ok" its not great.

Just fixing movement and the horrible crunch noise from the base model this round. The sound isn't perfect but it better here.

I have not tested with the 4 step lora yet. Please let me know if it works ok

MiniMax H3 audio

There was a bug in the samplers for H3 audio it has been fixed in ComfyUI v0.31.0. From my tests and seeing community feedback it is significantly better.

Things to come:

• More/different angles

• Better prompting around sound

• Smaller lora: i hope i started with rank 64, LTXV traumatized me.

• Long videos with no action.

Generated with the BF16 Model and Text encoder rank64. I'm still testing the rank so this is a big boy. Ask me if you have any questions.

All gallery images have workflows and prompts were all the same used fl2va_pruned_int8_convrot and nvfp4_awq for inference.

The prompting is pretty flexible just posting my test prompt.

Prompt:

integrated_multimodal_description: [Shot 1] The blonde woman deepthroats the penis in front of her, slowly pressing her face against the man's hips then pulling back to repeat the motion. She pushes her nose to the base of the shaft and pauses, her face showing a slight wince as the shaft presses into her throat and she lets out a short glottal "guhk" before retracting her head. Her lips remain sealed around the shaft as her jaw and throat bulge with each deep stroke. Saliva and spit drips down the shaft.

overall_soundscape: Heavy nasal breathing punctuated by wet. smacking and slurping on every stroke. Thick saliva sounds, "guhk" each time the shaft bottoms out in her throat.

non_diegetic_music: N/A