In AI a seed is a series of numbers that tells the AI how to generate an image. It is a starting point for the random number generator. The AI model is trained by learning from a set of input data, trying to identify patterns and relationships between the data points. Once the model has been trained, it is able to generate new data that is similar to the training data by using the patterns and relationships it has learned.
By generating a random seed the AI is able to generate different images each time, even if all other values stay the same. This allows infinite possibilities and an endless variety of images to be created each time.
If you use the same seed, prompt and settings, you can get the same image every time.
Seeds play a critical role in the creative process. Artists can use different seeds to explore variations on a theme, or share their favourite seed values with others to explore, compare and evaluate different AI generated results.
How to use seeds in NightCafe?
You can use the random seed generator for every creation. When duplicating or evolving, we recommend using the same seed if you just want to tweak the prompt or other settings, but if you want to "roll the dice" again and see what you get, the best way to do that is to simply use a different seed number.
Some algorithms also treat seeds differently than others. When using the Stable algorithm, the seed is deterministic, so creations with identical settings and the same seed will produce the same result, but identical settings and a different seed will produce a different result. Although, using CLIP guidance gives the seed less determinism, and two identical creations may still result in different images.
When using the Coherent algorithm, changing the seed even by one digit can generate varying effects and is worth experimenting with.
When using the Artistic algorithm, changing the seed usually doesn't have too much of an effect but can change the location of the focal points.
Tip: Use "try multiple" to create more than one artwork with the same settings but different randomized seeds.