This is an SD script designed to help anyone create better photos. Inspired by etcroot's SD Prompt Builder, huge shoutout to Dreamlabs for the original idea and implementation.
Welcome to the SD Prompt Enhancer Extension! You can now add the extension via your webui from this repo.
I'm taking requests to add community-made tag files to a public database so everyone can share their inputs! Here is the request form and here is the google drive folder with all of the approved tag files
For various reasons, I am not allowing tag files with any NSFW descriptors or content to be published
Having trouble coming up with just the right type of picture you want? Do you know vaguely what you're after but need some help with the specifics? I present the SD Prompt Enhancer!
Enhance your photos with with style, post-processing, lighting, and extra effects... mix and match to produce your desired output!
Take advantage of the camera options to get the best shot, frame, focus, and angle for your image.
You can choose what matters most! Employ the radio buttons to switch between Prompt, Image Attribute, Camera Settings, and Random priorities to more accurately represent the type of image you're after.
This is an active project and I will continue to add features in the future. Please leave any feedback and comments as I need ideas to touch up this script.
TO INSTALL
You can now install from the extensions tab in your webui!
Please look at the Installation section of the README.md file in the github repo linked above. This holds the most current information regarding installation of the script.
Description
Added tag database (csv file) that can be edited manually to add arbitrary user tags
FAQ
Comments (34)
Is it easy to expand the tags and categories yourself?
Yes and no. I say no because there is no UI support for this feature yet. I say yes because with v1.0 you can to go into the script and edit the dictionaries that the script references. the script uses the dictionary keys as the options for the dropdowns, and the dictionary values as the actual prompt strings. You can absolutely add an arbitrary number of key-value pairs to any dictionary you want to add more tags to the categories,.
For categories, you would have to make a new dictionary with your key-value pairs and create a new gradio Dropdown element to support the functionality. It's certainly a lot harder to add your own category than tag, but it's not impossible.
@xerminator13 So technically it is somewhat similar to https://github.com/some9000/StylePile ?
I been using https://github.com/blue-pen5805/sdweb-easy-prompt-selector loads tags from .yml files. It has few annoying things, but pretty decent. https://i.imgur.com/SgHdA7k.png
Looking forward what you can manage to make out of it :P Good tag extension with easy tag/category adding would be really nice.
@Potatovision I suppose it's in the realm of StylePile, but not nearly as involved or complicated at the moment. My main goal is to add the ability to add custom tags and categories within the UI, that's a really great feature idea.
I will say that this is meant to be a lite tag editor and will not nearly reach the complexity of something like StylePile for the moment, but that's certainly a goal to reach for.
@xerminator13 Too complex things are bad in the end. Stylepile is really good for someone who is starting with stable diffusion, but once you get better at prompting it becomes hassle to use.
In the end in my opinion perfect would be something you can manage yourself, add tags, categories. Nothing overly complicated. The easy prompt selector is pretty close. You can just add tags to prompt with click. But new tags must be added from file and the formatting must be really specific or it breaks down. Was fun after adding 500+ tags to 10 different files and it broke down 🤣 Took while to fix.
Another thing would be good, is to be able to share the tag/category files.
@Potatovision Makes sense. My current vision for this is to have two features implemented: Adding and removing tags and categories within the UI (no complicated formatting nor adding by file) and being able to add tags straight to the text box instead of having to run straight from the script.
Sharing the tag/category files is also an interesting idea I didn't consider, that's another good idea!
@xerminator13 Very nice 😎I think the tag sharing part is something that would make the script/extension shine, as the tagging/category creating is not solely on the creator of the script.
with the new version there is a csv. read the info
Nice, getting there 😎
This project is super cool, glad I was able to inspire this creation & thank you very much for crediting me as well! I'll make sure to link this project on my projects page in my next update for people who want to run it locally.
After adding options, I don't know what to do next. How to add the selection to the prompt word?
That feature is Work In Progress. As long as a tag is showing in the dropdown menus, when you press generate it will automatically be applied to your prompt, just not in the text box.
@xerminator13 hahahaha，I understand, it's great，thx
Great tool. This is very convenient.
I’m glad you like it! If you get the chance, I’d really appreciate an official review! Otherwise, please enjoy the script and suggest and features to be added later👍
This script is great!
It makes prompt input so much better depending on how you use it.
Thank you!
Great help thank you. really great product.
Very minor point hopefully in the next iteration it can pick up these additional prompts when doing send to img2img straight from text2image - they are picked up in the image browser chunks to png so can send from there as a get around but great work thank you
I completely agree with this feedback, I am working on this feature and hope to release it with v2.0
adding Staging Positions to camera settings would be amazing. heres a good list I found.
https://www.filmmakersacademy.com/glossary/staging-positions/
Thanks for the suggestion! I'll be sure to add these to the default tag collection. I would like to note that for the moment I don't want to add too many tag categories to the script and database just to maintain user friendliness, but I acknowledge adding your own tags to the database is not very easy at the moment.
How do I install it? I copied the file "prompt_enhancer" to the scripts folder but in the webui, scripts menu I don't see the "SD prompt enhancer" option.
For v1.5 you must have both the csv file and prompt_enhancer files in the "scripts" folder, or else prompt_enhancer will encounter an error and will not display within the GUI. It may also be named "SD Programmatic Prompt Enhancer" within the scripts menu on the webui
Why dont you put it on github aswell?
I use had similar script before "StylePile" but yours is a bit simpler, and I try to build a style list, and a bit cumbersome going to the Notepad filter bunch of stuff. xd
I'll second the github comment.
@thelustriai @SnG17 I’m working on v2.0 right now, it should hopefully be out within the next couple of days. It’s coming with some major backend and frontend upgrades worthy of a github branch, but otherwise I think v.1-1.5 will stay strictly on Civitai. Also idk how to enable script installing from a github link within the webui yet.
@xerminator13 Nice. Just ask on the repo, maybe someone can help you out.
@SnG17 Actually that’s a good point…
What is the difference between StylePile, Wildcards, and Prompt Enhancer? Kinda wanted to find "the right word for the job" rather than a creative roulette.
Hi, it's nice, but most of the words used in dropdowns are not really explicit, like in "camera shot", i don't understand any of the words. Is there a wiki or somewhere with exemples to show each elements how it impact images?
No, there is not. However, I pull these tags/concepts from a couple different online sources, so I can link those here
@xerminator13 Yes it would be useful to have an index with the words used, and to explain what effect they have
@xerminator13 please attach a list with the explanations of the abbreviations. thanks
@Jesse_F A link to the page with the camera settings is here. I cannot give explainations for the Image Attributes because I did not make those tags, the source for those is on the SD Prompt Builder that I linked at the top of my model description. I you want specific tags values you can read the csv files for the moment, I will consider adding an interface to view current tags and their values in the future.