
This LORA is based on a dataset of 1700 images from NASA's Astronomy Photo of the Day archive, making it ideal for generating stunning space-themed imagery. It's capable of producing telescope photos, beautiful auroras, and composite space images from Earth, while also handling more specific subjects like rockets, space suits, and planets. Astrophotography terms such as nebula, Oort cloud, supernova, red dwarf, and black hole also work really well with this model.
Key Details:
-
Keyword:
nasapotd
-
Trained on Image Sizes: 1080x720, 1024x1024, and 1280x720
-
What You Can Create: The LORA excels at generating a wide range of space imagery—telescope shots, cosmic phenomena, auroras, planets, and rockets. It responds especially well to keywords related to astrophotography and astronomy, such as nebula, black hole, and supernova.
Tips for Best Results:
-
Sampler: For most images, Euler tends to work best. However, if you're trying to generate more abstract or unrealistic images and having trouble with prompt adherence, switching to Heun or Heunpp2 can help rein things in and bring better results.
-
Adjusting Strength: Start with a strength setting of 1.0. If you find yourself getting repetitive images (e.g., the same galaxy shot), drop the strength down to around 0.7 and adjust from there to get the variation you want.
-
Keyword Placement: The position of
nasapotd
in your prompt affects how strong the LORA’s influence is. Putting the keyword at the beginning makes the LORA more prominent, while placing it towards the end dials back its impact.
Watermarks tend to show up fairly often, so keep that in mind.
描述:
训练词语: nasapotd
名称: nasapotd_rank16_bf16.safetensors
大小 (KB): 149681
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success