Using Vision FX 1.0 AI Plugin with Paintshop Pro 2023

[sheng-si]

Perhaps words like Salvador Dali - maybe in conjunction with the words extremely detailed - are sufficient to trigger the original warning?

No Salvador Dali is in one of the prompts given by corel on their tut page

[sheng-si]

What is this NSFW nonsense? I tried to fiddle with a photo of my wife and me, and two out of the five results say they're Not Safe For Work. I'm not happy about some computer program making those kinds of decisions. Is there a way to de-Bowdlerize this thing? Or did I just waste my money?
And no, there was no nudity or anything like it in the original image.
AI BS Image1.png

A way to solve NSWF https://www.distinctplugins.io/faq

[sheng-si]

I've been using Vision FX for a few weeks now, and I'm still lousy with the text prompts. And I too get the NSFW images and was never able to understand why; it just happens sometimes.

I think a lot of both situations may be related to the 'starting seed.' What is that, precisely? Is it the image I'm applying the effect to? Some image pulled at random from the net? I have no idea. The number changes on its own.

If anyone knows what the starting seed is I'd love to know it! Thanks!

maybe solving NSFW https://www.distinctplugins.io/faq

[sheng-si]

How do I remove the NSFW filter in Vision FX?
Not possible within vision fx to modify or remove.
Because they are using
Stable diffusion 1.5 model
The algorithm works independent from distincs plugins - maker distributor Vision FX.

Found something, NOT YET TRIED MYSELF

https://pinter.org/archives/14569

[sheng-si]

The solution on pinter.org did not help because the named file is not on my pc
BUT

In file
Pipeline_stable_duffusion_img2img.pyc

I find the following txt.

You have disabled the safety checker for a by passing `safety_checker=None`. Ensure that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered results in services or applications open to the public. Both the diffusers team and Hugging Face strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling it only for use-cases that involve analyzing network behavior or auditing its results. For more information, please have a look at https://github.com/huggingface/diffusers/pull/254

and also this
Classification module that estimates whether generated images could be considered offensive or harmful.
Please, refer to the [model card](https://huggingface.co/CompVis/stable-diffusion-v1-4) for details.
feature_extractor ([`CLIPFeatureExtractor`]

I think someone with more knowledge can help us Maybe

[Kathy_9]

So, if AI decides your image is NSFW, does it only warn you and allow you to proceed or does it not deliver your image?

Asking again in case more people have encountered the NSFW and know the answer.

[LeviFiction]

As I don't have the plugin, preferring a free stand-alone option, I cannot help.

[CloudyBright]

It produces an NSFW occasionally but doesn't stop the program from producing four more images. I haven't had an NSFW appear twice during the same pass.

[sheng-si]

So, if AI decides your image is NSFW, does it only warn you and allow you to proceed or does it not deliver your image?

Asking again in case more people have encountered the NSFW and know the answer.

No it do not stop the plugin. In stead of five previes you get mostly four WIth on one the text NSFW.
I found some more info on github how to remove it in stablefusion but I could not find the same text in visionfx files. Nearly same text but not what was explained.

[sheng-si]

As I don't have the plugin, preferring a free stand-alone option, I cannot help.

I hoped you are more familiar with possibilities in stanble fusion and the explanation to remove something as I found on github and huggingface.
Mostly because I assume something can also happen in stand alone AI based on stable fusion.

[LeviFiction]

Sorry sheng-si but this is more related to the plugin and how they did everything than just the sd1.5 model itself. InvokeAI (the program I use) has a simple slider button to turn off NSFW monitoring. It can also load additional refined models others have trained. It's just very different in its implementation than the Vision FX plugin. Though as I don't have a suitable graphics card, it runs on my CPU and is very slow. 5 minutes to get through 15 iterations/steps so I also don't use it often.

[Trst001]

This plugin looks interesting, but I only have a plain RTX 2060 with 6GB of RAM. So, I suppose it will not run on my PC, or ?

[revmatthew]

The Vision AI will not work with a number of issues. It crashes and then it NSFW labels even things like classical statues. Sigh. Money not well spent.