Perchance - Create a Random Text Generator @lemmy.world Khudozhnik @lemmy.world 2mo ago

Embedding one input inside another

hello people... I have a bit of a weird question. Say I have two dropdown lists named "Btype" and another named "Environment" Is there a way to implement the [Environment] input into a [Btype] input. I'm wondering if and how this can be achieved. I tried the following, yet it doesn't work:

Btype label = building type type = select options Default = , House = Generate a house with garden in the style of [input.Environment], House with 2 floors and slanted roof,

All help is much appreciated, thanks.

You're viewing a single thread.

12 comments

You're using the t2i framework, I'm guessing? I don't use it myself, but having a look at it... Maybe this will work: [this.getParent.Environment._element.value]... something like that. That will find the actual select box in the page and and get its current value. Maybe.
- sorry... your suggestion still triggers an error on my end... yet I found an effective, yet code-heavier, solution to avoid errors... conditional prompting::
  
  Line-Art style if [input.Lighting] != "": prompt = [input.description], (line drawing:1.8), (no gradients:1.8), clean lines, no color, black and white, visible hatching, high contrast, high resolution, high detail, intricate details, 4k, wallpaper, concept art, pen on paper, [input:Lighting] else: prompt = [input.description], (line drawing:1.8), (no gradients:1.8), clean lines, no color, black and white, visible hatching, high contrast, high resolution, high detail, intricate details, 4k, wallpaper, concept art, pen on paper, negative = [input.negative], low-quality, deformed, text, poorly drawn, 3D, color, gradients, greyscale, shading
  
  It seems to me you are accessing the other inputs with :? It should be [input.Lighting] not [input:Lighting] if you have an input with:
  
  Lighting label = Lighting type = select options ...
  
  Also, you can have the lists in the prompt like so:
  
  Line-Art style prompt [input.description], (line drawing:1.8), (no gradients:1.8), clean lines, no color, black and white, visible hatching, high contrast, high resolution, high detail, intricate details, 4k, wallpaper, concept art, pen on paper, [input.Lighting != ""] ^[input.Lighting] [input.description], (line drawing:1.8), (no gradients:1.8), clean lines, no color, black and white, visible hatching, high contrast, high resolution, high detail, intricate details, 4k, wallpaper, concept art, pen on paper, ^[input.Lighting == ""] negative = [input.negative], low-quality, deformed, text, poorly drawn, 3D, color, gradients, greyscale, shading
  
  So that you can use Dynamic Odds in selecting which prompt to use if input.Lighting was added/selected.
  
  You could also link the generator so we can directly see any problems.
  
  What does that even do? I've never seen code like that in perchance documentation or anywhere else. Does that actually do anything?
  
  Got a link to a generator that uses this code? It might stop the error but not actually do anything; if [input.Lighting] != "": would just be an item within the Line-Art style list. 🤷
  
  If you could give me a link to the code I gave not working, I could look at the error and figure it out for you.
  
  https://perchance.org/ai-architecture-generator Don't mind the clutter in the styles... I'm trying to cook a soup from those! I added your suggested snippet into the "House" btype (building type)... where I tried to implement the selection of Rtype (roof type)... I hope I done it right, but I doubt it.
  
  Ah, you're actually putting it in a different place that I thought you were. Also, the values of the dropdowns don't work the way I thought they did 😂
  
  It seems perhaps [input.Rtype] is what you're meant to use with that system to get the current prompt from that. (It's a bit complicated honestly--I use my own plugin so I'm not experienced with t2i... 😅)
  
  What exactly were you trying to do? If it's just to include the value in the other part of the prompt, the above should work (I guess). But you did that before right--what happens if you do that?
  
  Yes it did actually work... But VioneT was bright enough to remind me of the case sensitivity, which I totally missed in that one. You were right to assums I just wanted to include a value in another prompt, and by doing so, simply combine prompts before they are sent to the output. I'm not sure, but I think that way it'll be easier to manage the weight of certain inputs. So, if I'm correct (like I did in the gen, and as far as my brain lets me figure things out for itself), if I embed a certain input into all options of a different dropdown, with [input.Rtype], I could take that one out of the imageOptions prompts list, right... or else it will be double? By doing that I hope to create a more "workable" promptline for the AI to interpret. But I dunno... I'm just trying to make it make sense in my own brain, tbh.
  
  Yeah, looks like what you did by commenting that out of the prompt is about right. 👍

12 comments