Thanks for your very nice Styles custom node, I have it in use with all my workflows.

Since the update, all json files load. This can be convenient, but it also can lead to a very long dropdown to select a style ... too long atually to have good oversight

I know I can change and shuffle the json files, and there also is the search field, but I had this idea if maybe it would be possible to add a dropdown selector, where one can choose which styles show up (which json file will be used) in the second selector? And there can also be an entry that says 'all'. It would be a bit more convenient than the search field.

Greetings, Rudy.

more of a question than an issue.

Do you mind if people release variants of this node with different .json files.

I have a version with Fooocus styles included and would like to post it on git but only if you don't mind - I would link back to your original if you agree.

thanks

Mokkaboss1

The included styles in this node are pretty good, but there're actually endless styles, it's impossible to include all, and if you include more the list will become too long to use easily. I propose to add a custom style prompt input, so that user can use an artist style, such as "breathtaking {prompt} by Anna Dittmann," weights can also be easily added by user. This should work as long as the simple format is followed, i.e., the style contains a {prompt} placeholder, which is simple to validate. When there're values in both the selected style and custom style, the custom one should take precedence. Please consider adding this feature., which will make this node the most powerful prompt styler.

Hello there...

Love your custom nodes for ComfyUI but have a request...
Could you possibly add a way to randomize the style on every gen ?
And perhaps an option to go thru them sequentially...

THanks for your time !
And hope to hear from you.

Peter Lunk

reddit.com/user/MrLunk

风格里有一处少了个“.”，就是
{
"name": "sai-texture",
"prompt": "texture {prompt} top down close-up",
"negative_prompt": "ugly, deformed, noisy, blurry"
}这一句，
应该为
{
"name": "sai-texture",
"prompt": "texture {prompt} . top down close-up",
"negative_prompt": "ugly, deformed, noisy, blurry"
}
我用它在sd webui里生成xy图时这一项报错所以发现了
另外在您的插件能否在comfyui里用xy表格来使用，我没有找到方法

Well, as title says it will be useful to have possibility to tweak this value. It can give us more flexible way of creating prompt

I'm getting the following error on any style I select. What do I need to do to fix this?

Prompt outputs failed validation
SDXLPromptStyler:

• Value not in list: style: 'AR Art Variant_Uncategorized' not in (list of length 106)

Thank you

But with advanced nodes, you can read the defined style.

It would be great if the following two additions could be made:

1. Stack the styles. It would be great to offer the ability to select up to three styles instead of having to chain them together.
2. Output text of selected style as a String. In addition to the positive and negative prompts, it would be great to output the name of the selected style. This could be used to name files/directories when saving images.

Thanks for the custom node as I use it all the time!!!

Dear twri,

This project is AMAZING, and I have never thought simple yet sophisticated prompts can have such effective and dramatic on any images.

We are building a super cool image generation product for the masses, and I would love to hear your advice on prompt engineering. Please reach out to me at [email protected]. I am dying to speak with you.

Sincerely,
teteted

Awesome node, thanks for building this. It is exactly what I was looking for.

I have one JSON file that will not load. It is the sdxl_styles_fooocus.json file that is attached. I have put other JSON files in the same directory that have successfully loaded into the node. For some reason I cannot determine, this file will not load.

Is there a name length limit or something else I am missing?

Thank you!

Any chance this can be added? I was about to take all the styles and create my own wildcard but maybe this isn't something too hard to implement so I can keep on using the nodes.

what is "sai"?

It would be really cool to have a 'random SAI style' option that randomly selects from one of the SAI styles. I think this is what clipdrop does today.

Hey there,

I got this issue; all the settings were filled, and the inference worked fine in Comfyui. This issue is thrown by API.

{ "error": { "error": { "type": "prompt_outputs_failed_validation", "details": "", "message": "Prompt outputs failed validation", "extra_info": {} }, "node_errors": { "259": { "errors": [ { "type": "required_input_missing", "details": "style_positive", "message": "Required input is missing", "extra_info": { "input_name": "style_positive" } }, { "type": "required_input_missing", "details": "style_negative", "message": "Required input is missing", "extra_info": { "input_name": "style_negative" } } ], "class_type": "SDXLPromptStyler", "dependent_outputs": [ "17", "346", "187", "103" ] } } } }

Hi,

First your node is really great. I would like to know if it's possible to make like in Fooocus v2, a multi select style that will be merged as in Fooocus v2 ? It's great and work very well to get a specific style and result and could be great in your node.

Hope it'll be possible one day :)
Thx

Matt

like this one here

Greetings,
Thanks for sharing the prompt styler, how do you implement the script: sdxl_prompt_styler.py(Any examples would be nice inclusion). Looking forward to hearing with you.
Best,
Andy

You can add a file filtering option

Why can't this node behave and be compatible and act like a conditinong node directly? I'd love to use this for 1.5 workflow too but I can't seem to connect this node to a Ksampler positive and negative input respectively. This would be super useful. What am I missing here?

Hello,

I did add a feature on my fork of sdxl_prompt_styler adding a new option to the advanced node. It is used to copy the content of G text to L, deduplicating existing tokens found on both prompts.

This is used mostly by convenience because some models behave better with the same text in G and L. It kinda acts as a "overpowered text control" option.

Do you want a PR of this feature ?

hi,
first of all, thank you for your work. I like to use the tool, but it also took me quite a bit of work after I started using it.
Since I do a lot of testing, I often used the "send to txt2img" function in PGN Info as well, only to find that my prompt looked like this

<parameters
ethereal fantasy concept art of ethereal fantasy concept art of professional 3d model professional 3d model professional 3d model 1girl, best quality, masterpiece, ultra highe res,raw photo,supercute,detailed...>

By the way, the same thing happens in the Negative prompt, so now, every time I use SDXL Styler, I have to clean up my prompt afterwards before I can use it.
It would be nice if you include a function in the script that.
First, checks if the keyword is already included and second, also removes the redundant (deselected style keyword) from the prompt when it reappears in the prompt, and third, doesn't insert it multiple times in a row each time you sent the prompt with txt2img to the prompt.

many thanks and best regards
LordKa

can the sdxl prompt styler be used for sd 1.5? if not can can those prompt presets in your repo be used for sd1.5?