Hi~!

I’m getting familiar with Perchance and the text to image plugin, so I modified a script from the guides to test tags blending a bit more. I’m trying to visualize task blending, but when the prompt is not fixed (contains other lists that are being evaluated than the blended tags). In that case, I would like in each row to only change the blended tags, and to keep the same seed and evaluations of the other lists the same (e.g., here I want the same animal). But I didn’t manage to do that yet, as the animals are changing within the same row. I wonder if there is a way to partially evaluate the prompt a first time for each row and a second time for the blended tag.

Another way would be to do string formatting to handle the tag blending instead of using perchance lists (with a trick like this: blending_value = [blending_values.selectAll[counter]]), but I didn’t see a part about that in the guides, and I’m not familiar with JavaScript at all.

Do you have some solutions for this?

Here’s the generator I started: https://perchance.org/7zk7cvxzzl

I coded it to blend 2 dimensions of tags within a matrix, I use only 1 dimension here, so it should not be relevant.

Notes: The idea behind this is to visualize the tag blending on a sample of the possible prompts rather than a fixed one. Here, there are 6 animals, so we could try them all on a bunch of seed, but in practice, it’s not doable, that’s why I’m doing the sampling in the first place.