Autoplay
Run Locally
nafasp_lora_v1_500.safetensors
296 MB · SafeTensor · bf16
Downloads are served by Civitai. Some creators ask for a Civitai login first.
About This Model
MINIMAX H3 · LoRAs
by AdaptiveVision - Natural Face & Speech Minimax H3
Creator notes
Face and speech dynamics for Minimax H3
Full face muscle dynamics with clear English speech
Trained on T2V but supports also I2V
720x1280 resolution with 15-30 steps
Weight 0.6-0.8 is recommended
Pro Tip: to prevent voice drift, use fixed seed and start your text with the same word (hey guys,)
Prompt: she is speaking into a mic, she maintains eye contact with the viewer saying: 'text here'
SHA256
74150D1D047FE8FB97C861CAFB887CE1D030B11055F50F91EB93A0B169DA297F
ModelTrends.ai Model ID
#26974
Flag content
Example renders
Prompts and settings shared by the people who made these renders on Civitai. Pick a render to see how it was made.
Settings
Sampler
Euler
Steps
8
Guidance
1
Seed
1111
Size
1472x2207
Prompt
REF2VA - 12 seconds - Aspect ratio 2:3 - <PICTURE 1> at the 00:00.000s - <PICTURE 2> as grip reference for (S1).
[REFERENCE]: <PICTURE 1> is the first frame and establishes the subject (S1), a human female paladin with blonde hair, wearing polished white and gold plate armor over fine fabrics; she holds a kite shield with reinforced metal edges and a glowing ethereal yellow sword. <PICTURE 2> serves strictly as a technical reference for the grip of the shield, showing how (S1) holds the metallic handle with her hand while a leather strap secures her forearm to the wooden surface. The environment is a military camp at dusk; background soldiers are visible but heavily blurred with deep bokeh to maintain focus on (S1). Clean 3D stylized fantasy art style with hand-painted textures and solid volumes.
[CORE IDEA]: A professional yet comedic tactical instructional video featuring an experienced Paladin instructor (S1) teaching a student (the viewer); military training camp atmosphere; high-impact physical blocking and technical shield maneuvers; cinematic tutorial style with dynamic camera work and sudden rhythmic cuts.
[PROCESS]:
[SHOT 1] At 00:00.000, the camera begins in a medium shot, pushing in slowly at medium speed towards (S1). The Paladin (S1) stands firmly, holding her shield high near her chin, tilting it slightly outward to demonstrate defensive positioning; she looks directly into the lens with an intense, professional gaze. With a deep, authoritative and commanding voice, the Paladin (S1) says: [English] "Listen well, boy! In hard moments, keep your guard high and let the blows rain down!" At 00:04.000, she abruptly tilts the shield outward to deflect an invisible thrust with a sharp metallic clink.
[SHOT 2] At 00:05.000, the camera cuts via Hard Cut/Tape Jump to a low-angle medium shot; (S1) deeply flexes her knees and takes a heavy, grounded step forward, planting the base of the kite shield against the dirt with massive force to create an immovable barrier. With a deep, authoritative voice, the Palotin (S1) says: [English] "We call this the Shield Wall! Not even an elephant passes through here!" At 00:08.000, she locks her arm, creating a static and imposing defensive posture.
[SHOT 3] At 00:09.000, the camera cuts via Hard Cut/Tape Jump to a tracking side shot; (S1) rotates her torso 45 degrees backward with high speed, pivoting on her left foot while extending the shield laterally in a wide fan shape to cover her flank. With an energetic and instructional tone, the Paladin (S1) says: [English] "Watch your flank! There are no friends behind your back!" At 00:10.500, she completes the rotation with a sharp, disciplined stop.
[SHOT 4] At 00:11.000, the camera cuts via Hard Cut/Tape Jump to an extreme close-up of (S1)'s face and shoulder; she takes an explosive step toward the camera, performing a rapid shield bash feint with the flat rim while her glowing yellow sword flashes near her shoulder. With a gritty, experienced tone, the Paladin (S1) says: [English] "With this, you give them a good scare and make your enemy run!" At 00:12.000, the shot ends with an intense stare as the camera performs a quick pull-out to black.
[AUDIO]:
Non_diegetic_Music: N/A.
Overall_Soundscape: Distant sounds of a military camp; clashing steel from background soldiers, muffled shouts, and heavy wind drafts through canvas tents.
Timed_Sound_Effects: At 00:04.000, a sharp metallic impact against the shield rim. At 00:05.000, a heavy, deep thud of wood hitting the earth. At 00:09.000, fast armor sliding and fabric swishing during the pivot. At 011.000, an explosive air rush (whoosh) following the shield bash feint, followed by a natural fade out.
Settings
Steps
30
Size
640x1152
Prompt
Young African American woman is making TikTok style video when traveling in Japan. She says: 'Hey, would you like to hear 5 makeup tips for black women?'.
Settings
Steps
30
Size
640x1152
Prompt
Local TV news studio, medium close-up of a female news anchor facing camera, shoulders-up, soft key light, shallow depth of field, news desk and city skyline screen behind her. She speaks clearly in natural American English, calm professional tone, small nods, natural blinks, lips match the words. No music swell, only studio room tone and her voice.
Dialogue:
"Good evening. Downtown traffic is moving again after the bridge reopened, and crews say the evening commute should stay clear."
Settings
Size
1088x1728
Prompt
A young woman is looking at the camera. she is placing an order. There are french fries in a container on the counter top.
The young woman says in a snobby voice, "Don't judge me by what I eat."
The camera zooms on on the young woman's face.
The young woman says, "I can eat french fries all day and still have a great body.".
The young woman reaches down and grabs a french fry from the container and eats it.
The young woman says passionately, "this tastes so good!".
The young woman grabs another french fry and eats it.
Similar Models
AI Models for Image, Video & Text Generation
natural face speech h3 is a LoRA for the MiniMax H3 family listed on ModelTrends.ai, a read-only catalog of open source AI image, video and text models. Compare it with other MiniMax H3 models, check its heat score to see how it is trending, and open it on PirateDiffusion or BitVector to try it.
Browse by Model Family
Text / LLM Models
411
Anima Models
509
Chroma Models
14
Flux Models
1,071
Flux 2 / Klein Models
1,237
MiniMax H3 Models
331
Hunyuan Models
197
Ideogram Models
20
Krea2 Models
674
Qwen Models
17
Qwen2 Models
36
Ltx2 Models
150
Stable Diffusion 1.5 Models
6,465
Stable Diffusion XL 1.0 Models
15,432
Zimage Models
804
Wan Models
919

Model
Trends
.ai
© ModelTrends.ai
|
Made in Japan
|
© 2026
