black-forest-labs

/

FLUX.1-dev

like 4.49k

Hi Team,

I hope this is the right place to share this. I wanted to offer some constructive feedback on the model's current capabilities.

Firstly, it's crucial for the model to better understand and represent women. Implementing classifications like Type 2b, 2c, and 2a for hair and standard bra sizing (A, B, C, D) is essential. The model also needs to incorporate fashion sizing and terminology more accurately, as it currently seems to lean towards a specific look that feels quite limiting. The renders often have a flux-like appearance that doesn’t align with diverse representations.

Additionally, the software doesn’t seem to cater well to people of color. For instance, it struggles with conditions like vitiligo and tends to interpret terms like "chubby" or "thick" as overweight, which isn't always the case. There's also a recurring issue where the model defaults to features like a cleft chin or a European bone structure, particularly in up-close or straight-on portraits. This is rarely seen in women from certain ethnic backgrounds, such as Korean, Chinese, Japanese, or Filipino.

Moreover, there's an assumption that Chinese characters must align with an oriental style, which is inaccurate in most cases. This type of stereotyping can be problematic and doesn't reflect the diversity within these cultures.

I hope this feedback is helpful as you continue to refine the model. Including more diverse perspectives and experiences in your development process could greatly enhance its inclusivity and accuracy.

Thanks for your hard work on this, and I look forward to seeing the improvements!