Using Vision FX 1.0 AI Plugin with Paintshop Pro 2023

[Morticia]

LeviFiction, I just tried that, and here's what the slightly larger upscaled image looks like:

Sailing the Vast 2.png

[GenJerDan]

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

[JoeB]

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

First, your image didn't show up, just the image title AIBS Image1.png.

Second, did you get that warning as a popup or something from VisionFX plugin or when you posted (or tried to post or send) the image somewhere? If the former, did it prevent you from continuing with working on the image or were you able to continue your work on it despite the NSFW warning? If the latter, the warning was from wherever you tried to post or send it.

[LeviFiction]

I would also ask if the message popped up as soon as the plugin loaded, or as a result of your prompt? If it happened after making a prompt, then the prompt would be more apt than the image.

Ethical AI strives to avoid potentially harmful results. And an AI generator that takes in photos of people and produces output in the wrong hands could result in something unexpected or bad. So, often, code is brought in to make the determination to avoid such outcomes if possible. But it's not actually smart, it looks for patterns and produces probabilistic outputs. It will get things wrong from time to time. Even worse, it can be fooled and manipulated with a few choice words.

[GenJerDan]

First, your image didn't show up, just the image title AIBS Image1.png.

Second, did you get that warning as a popup or something from VisionFX plugin or when you posted (or tried to post or send) the image somewhere? If the former, did it prevent you from continuing with working on the image or were you able to continue your work on it despite the NSFW warning? If the latter, the warning was from wherever you tried to post or send it.

Odd that the image didn't. No matter.

No, no popup or anything. In the preview boxes at the bottom, instead of an image there was the slashed-circle warning symbol over "NSFW", so it wasn't even my photo that was 'naughty', it was the plugin creating its own version and then spanking itself for it.

[GenJerDan]

I would also ask if the message popped up as soon as the plugin loaded, or as a result of your prompt? If it happened after making a prompt, then the prompt would be more apt than the image.

Ethical AI strives to avoid potentially harmful results. And an AI generator that takes in photos of people and produces output in the wrong hands could result in something unexpected or bad. So, often, code is brought in to make the determination to avoid such outcomes if possible. But it's not actually smart, it looks for patterns and produces probabilistic outputs. It will get things wrong from time to time. Even worse, it can be fooled and manipulated with a few choice words.

Oh, the plugin rendered, but instead of the little preview boxes at the bottom, it had slashed-circle "NSFW" instead of an image.

[Lance]

I just downloaded Vision FX, and am toying with a photo, and so far, coming up with some interesting results. I am looking forward to finding out all the different things that I can do with the plug-in!
The more guidance I can get including ideas for Text Prompts to use for certain effects would be most welcome and appreciated.

[Morticia]

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!

[LeviFiction]

Morticia - this plugin is based on the StableDiffusion AI. I posted a link to a help document on the settings of StableDiffusion on Page 1 of this topic. StableDiffusion starts with an image of random noise and slowly builds up the result based on the number of steps. The Seed is the number used to generate the initial random noise. But it's only the starting point.

[Lance]

For the first time after starting to play with Vision FX, I got the NSFW result. The starting image was a photograph (my own) of a simple purple flower with green leaves. My prompts were 'In the style of Salvador Dali, extremely detailed digital art, mystical colors, beautiful lighting.' The warning came as soon as the first Preview completed the process. The image shows just fine in the main pane, the warningis in the Preview thumbnail. The second Preview came up without issue. So, there is nothing "questionable" in my image or in my prompts or in the actual final image at all. I don't get it.

[JoeB]

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

[LeviFiction]

Again, the issue is the AI. AI is not perfect it's not smart. It's trained on data and it uses probabilities and randomness to make determinations. It will get things wrong. The original Stable Diffusion includes a safety checker to either blur out or warn when it thinks it finds something that might be objectionable. Nothing actually has to be objectionable, the AI just has to "think" it is. There are entire papers on how Stable Diffusion's saftey checker works. It's mainly centered on sexually-explicity material and it does this through what's known as "sentiment analysis" which figures out how closely related various words are to each other using maths I don't understand. Your prompt and the result don't need to be bad. Your prompt just needs to have generated words that are close enough to 17 categories the developers thought were bad. It's a crap shoot.

[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?

[mabebo]

Hello, when using vision fx it suggests 5 different results. Ich can only save one of them. Is there a possibility to save more than one? Thanks for your help!

[sheng-si]

Hello, when using vision fx it suggests 5 different results. Ich can only save one of them. Is there a possibility to save more than one? Thanks for your help!

I also have that question in mind. It is Save and Close. So I think when I also want another of the examples, I have to start all over again. And processing takes some time.
Exampla of my 5 versions through screendump

VisonFX 2023-10-10 133251.jpg