You can use both tags and captions, both together and individually, my advice is to use the expression tag between parentheses before the caption.

You can also mark it by adding "very" to the expression in brackets, and with additional tags to add after the caption, such as "open mouth", "scream", "smile", "laugh", "terrified" etc.

A tip, when using Flux (in any case, not only with my LoRa) do not overdo it by only putting tags, tags must be used wisely and should only mark concepts that you have not managed to obtain with the caption alone.

DOES IT WORK WITH ANIME?

Quite, but it was not designed for expressions in this style, if this model is appreciated and users want it you can also make an Anime Expression Helper.

EXEMPLES

SAD:

(sad), a close up of a beautiful young woman in a room with a sad expression, Her lips is twisted of sad. Her eyes were bloodshot from crying. She have red hair and freckles and is underwater (
very sad), a close up of a beautiful young woman in a room with a sad expression, Her lips is twisted of sad and she cry. She have red hair and freckles and is underwater
(very sad), a close up of a beautiful young woman in a room with a sad expression, Her mouth is open and twisted of sad. Her eyes were bloodshot from crying. is closed and tearfull. She have red hair and freckles and is underwater
(very sad), a close up of a beautiful young woman in a room with a sad expression, Her mouth is open and twisted of sad. Her eyes were bloodshot from crying is closed and tearfull. She have red hair and freckles and is underwater,scream

ANGRY:

(angry), a close up of a beautiful young woman in a room with a angry expression. She have red hair and freckles and is underwater
(very angry), a close up of a beautiful young woman in a room with a angry expression, Her mouth is open and show her teeth. She have red hair and freckles and is underwater

SCARED:

(scared), a close up of a beautiful young woman in a room with scared expression. She have red hair and freckles and is underwater.
(very scared), a close up of a beautiful young woman in a room with scared expression. She have red hair and freckles and is underwater. scream
(very scared), a close up of a beautiful young woman in a room with terrified expression. She have red hair and freckles and is underwater. scream, open mouth

HAPPY:

(happy), a close up of a beautiful young woman in a room. She have red hair and freckles and is underwater
(happy), a close up of a beautiful young woman smile in a room. She have red hair and freckles and is underwater
(very happy), a close up of a beautiful young woman laught in a room. She have red hair and freckles and is underwater
(very happy), a close up of a beautiful young woman laught in a room. She have red hair and freckles and is underwater, open mouth

SURPRISED:

(surprised), a close up of a beautiful young woman in a room with surprised expression. She have red hair and freckles and is underwater.
(very surprised), a close up of a beautiful young woman in a room with surprised expression. She have red hair and freckles and is underwater. open mouth

DISGUSTED:

(disgusted), a close up of a beautiful young woman in a room with disgusted expression. She have red hair and freckles and is underwater.
(very disgusted), a close up of a beautiful young woman in a room with disgusted expression. her mouth is distorted by the disgust she feels. She have red hair and freckles and is underwater.
(very disgusted), a close up of a beautiful young woman in a room with disgusted expression. her mouth is distorted by the disgust she feels, and her expression is grotesque. She have red hair and freckles and is underwater. open eyes

Trigger words

You should use sad , very sad, happy, very happy, angry, very angry, surprised, open mouth, scared, very scared, disgusted, very disgusted to trigger the image generation.

Download model

Weights for this model are available in Safetensors format.

Download them in the Files & versions tab.

Use it with the 🧨 diffusers library

from diffusers import AutoPipelineForText2Image
import torch

device = "cuda" if torch.cuda.is_available() else "cpu"

pipeline = AutoPipelineForText2Image.from_pretrained('black-forest-labs/FLUX.1-dev', torch_dtype=torch.bfloat16).to(device)
pipeline.load_lora_weights('Keltezaa/expression-helper-2-0', weight_name='expression_helper2.0.safetensors')
image = pipeline('`sad `, `very sad`, `happy`, `very happy`, `angry`, `very angry`, `surprised`, `open mouth`, `scared`, `very scared`, `disgusted`, `very disgusted`').images[0]

For more details, including weighting, merging and fusing LoRAs, check the documentation on loading LoRAs in diffusers