Sign In

Nathalie Emmanuel

22
198
9
Type
LoRA
Stats
198
Reviews
Published
May 31, 2024
Base Model
SDXL 1.0
Training
Steps: 10,080
Epochs: 8
Trigger Words
nthlemmnlmssdr woman
Hash
AutoV2
9DF0000738

Nathalie Joanne Emmanuel (born 2 March 1989) is a British actress. Emmanuel began her acting career appearing in theatre in the late 1990s, acquiring roles in various West End productions such as the musical The Lion King.In 2006, she began her on-screen career by starring as Sasha Valentine in soap opera Hollyoaks, after which she appeared in various British television series until her debut film appearance in Twenty8k.

Emmanuel gained international recognition for her role as Missandei in the HBO fantasy series Game of Thrones (2013–2019), and continued her career with supporting roles in Maze Runner: The Scorch Trials (2015) and its sequel Maze Runner: The Death Cure (2018), the Fast & Furious films Furious 7 (2015), The Fate of the Furious (2017), F9 (2021) and Fast X (2023), and had lead roles in Army of Thieves (2021) and Megalopolis (2024).

Please tell me what you think of the file, the size and the details. I will use that to assess if the process is worth it.

  • I do not use adetailer or a refiner for the sample images posted above. (I do think adetailer is amazing and probably should be used for most artwork. I don't use it here just to showcase what the base file can do, but would absolutely use it if I was making art with my files)

  • My Files are designed to be work on the base model at weight 1.0. For other derived SDXL models, try weights around .70 to .80

  • This file heavily favors photorealistic style. It is possible to prompt for more varied artistic styles(see sample images), but know that you may need to increase the weight of the style portion of the prompt to get better results

  • I prefer DPM++2M SDE karras or DPM++SDE karras for these files. I think they produce better photorealistic results

  • As with all sdxl loras, the complexity of the prompt can reduce the likeness. All Samples images are at 1.0 strength, but feel free to adjust it higher if you are starting to lose fidelity due to a complicated prompt