SDS|FILM/Film Photography|XL

The model is mainly trained in three main categories: Film photography, Computational photography, and Digital raw images, focusing on in-depth adjustments for Asian portraits. The current training dataset mainly consists of young Asian men, women, and a small amount of scenery. GPT4V is used for labeling, with some mixed annotations using Cogvqa and WD1.4. Within WD1.4 annotations, "1girl" and "1boy" have been modified to "woman" and "man" to prevent age misalignment.

Note: This version is the SD1.5 version, merged from other model pages. Furthermore, the model does not aim for comprehensive balance in training objectives, hence there may be noticeable functional and aesthetic biases.

Key training concepts and trigger words are as follows:

Digital raw images: Raw format

Film photography: Film photography

Some film models: Fuji C100 shooting, Fuji C200 shooting, Kodak 400 shooting, Kodak gold 200 shooting, Nolan 5219 shooting

Computational photography: Computational photography

Image quality rating: Mobile phone image quality, Landline image quality, Pager image quality

Initiation technique (can be modified based on actual conditions, including adding computational photography and the following ratings as negative, for maximum image quality improvement, and vice versa):

Positive trigger words: 8K, masterpiece, best quality:1.2, ultrahigh-res

Negative trigger words: anime, cartoon, 3D rendering, high saturation, facial blemishes, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, CyberRealistic_Negative-neg, (SkinPerfection_NegV15)

The last two are negative embeddings

clip skip: 1

Other parameter settings can refer to the example image

Issues: The handling of eyes by the 1.5 model has always been lackluster, for further enhancement, the XL model can be used to individually redraw the eyes.

The model is based on LEOSAM FilmGirl Ultra training, all copyright usage declarations continue from the previous model level: https://www.liblib.art/modelinfo/2c9398ced43b7346a41bbf465946341d

————————————————————————————————————————