[Feature Request]: Syntax change to include Step Range in Prompt Editing

[vidiotgameboss]

Is there an existing issue for this?

• [X] I have searched the existing issues and checked the recent builds/commits

What would your feature do ?

Allow for Prompt Editing to be used from X to Y steps more intuitively.

example: a cat [with fluffy fur:10:15] wearing a hat

This would ask for the model to make the cat wearing a hat up until step 10 after which the model is asked to make a cat with fluffy fur wearing a hat up until step 15, after step 15 the model is once again asked to just make a cat wearing a hat. The result would be that with fluffy fur wouldnt be sampled as much as the other parts of the prompt and in return the fluffy fur would not be as intense.

Currently the only way you seem to be able to do merely mimic this is by using [with fluffy fur:15] as a negative prompt which would kick in after step 15 and remove the prompt but now the end result would be affected by the fact that there's now two prompts trying to neutralize eachother - which is not the same as not existing in the first place - and also by the order in which you've negatively prompted. It is a good alternative until this is implemented and it provides more customizability but the results won't be the same as what I've proposed above.

Proposed workflow

The base syntax would be changed from [from:to:when] to [fromword:toword:fromstep:tostep]

Additional information

No response

[hananbeer]

changing the format would break compatibility. I think the correct way should be change the behavior if 4 items specified and retain behavior for less items.

[ClashSAN]

[lion:bird:girl: , 6, 9] you can use an extension, it works with better syntax https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/7254