Hunknown

I still have occasional problems with saving images to the public galleries. It often gets stuck at the “Saving…” message. After some time (few minutes to an hour), the problem seems solved, to appear again at the next save. During a 2-hour session with my generator, I can save about half of the images I’d like to. I wish I could give more details, but this is all I can report.

As far as I know, the Anti-Description has become useless, because the current engine doesn’t use it anymore. With regard with the full-body images of people with long hair, I have no problems creating them on the standard ai-tex-to-image-generator, see example:

The image has been created by this prompt:

A handsome tall Monacan man in his 30s, muscular, shirtless, very hairy barrel chest, barefoot, standing on a deserted beach looking at the Ocean, his long black hair flowing around his torso moved by gentle gusts of wind.

It’s all about how you mention the relevant bits within the prompt, I think. In this case, I specified he’s barefoot, and his hair flow around his torso.

Thank you, and don’t worry, I’m WAY over 18 😂 and you know the level of kinkiness of my own generator 😅

Sorry, PROBLEM SOLVED. I had to create many NSFW images in incognito mode, send some to the gallery and reload the page to finally be able to see the entire gallery 🤦‍♂️. Thanks for your help!

Dang! 😂 I didn’t think that someone could downvote my images on my own site 😅 Also, I didn’t know that the votes were effective worldwide, I thought it was a local thing. I lowered the hideIfScoreIsBelow to a negative number and my images appeared again, at least in normal mode; but I still miss most of the images in incognito mod: will you please tell me more about the safety lock 🔐? Where can I find it? Is it a local or worldwide setting?

EDIT to add: when running the tests in incognito mode, I purposely created some NSFW images, Perchance asked confirmation I was 18+, I said yes and permanently deactivated the PG-13 filter; but still I could see only 3 images out of 108 😒. Now I’ve retried the ignognito mode, I can see some of the newely added images, but I noticed they were all mildly erotic, nothing really NSFW: therefore, there must be a further filter somewhere preventing me to see the NSFW images…

Thank you again for your tips 😊 I noticed, too, that longer prompts give more problems, and we should keep in mind that ‘prompt’ means the actual prompt + the artStyle description appended to it. There are some artStyles on some Perchance generators that are longer than a long prompt 😂, so it’s important IMHO to use simple artStyles (like Cinematic or Professional Photo), or select “No Style” and write our own style directives in the prompt, keeping them as short as possible.

Oh, in case someon’s wondering, you can set the default filter (‘g’, ‘pg13’, ‘none’) in galleryOptions -> contentFilter, but if you choose ‘none’, it reverts automatically to G. On the other hand, ‘pg13’ is accepted and actually used as default filter.

I second that, it’s a good thing to have different levels of filtering. Now, if only the AI could actually understand what’s G, what’s PG-13 and what’s rated R… 😅 A men’s chest, neck to belly button, is definitely G, or kids shouldn’t even go to the beach 😂

Thank you for your insightful tips, I’m definitely a common user of Perchance t2i and have no real knowledge about Flux or how I should create the prompts to better instruct Flux to create what I want. That said, my personal ‘trial and error’ experience is different than what you just stated. For example, still today I notice visible differences when enclosing something in brackets; not sure whether Perchance makes any difference when using () or ((())), but still there’s a difference, especially if the bracketed element is in the last part of the prompt. I’m not sure what the SFG is 😅. If it’s what the default Perchance t2i interface calls ‘Guidance Scale’, then a value of 1-5 gives horrible results with realistic images; not so bad when it comes to render a painting or drawing. My prompts are quite complex, and I often use a Guidance Scale ranging from 10 to 20, to balance the adherence of the image to the prompt and the realistic output. Lastly, I’ve modified some of the standard Perchance styles to suit my needs, but the evergreen “Cinematic” style (and a few others) is just perfect as it is, and still gives outstanding results.

I tried to render a random landscape using the ‘Oil Painting - Old’ style and it came out pretty good, not ‘digital’ in any way; but the default generator is quite limited. For example, you can’t set the ‘Guidance Scale’ that is automatically set to the default value of 7, but for a painting-style should be lower, roughly 4 to 6. You can try again with the default generator, adding at the end of your prompt: (guidanceScale:::4) including the brackets and the three colons ::: OR you can try a more art-oriented generator like https://perchance.org/wallpaper-konbini where you can manually set the Guidance Scale and choose one of the many painting styles.

Your answers are enlightening, I’m so new to all of this. Thank you very much. I hope that, at least, my generator can be indexed and listed in Google searches, even if “private”. There are tons of NSFW Perchance generators, when searching Google.

Thank you so much for the clarification! I’m actually totally NSFW, and I’ve stated it clearly in the code, tags and description, which explains why I’m not allowed to be listed in the gen list.

As far as you know, does the NSFW-ban apply to the hub, too? Because I’ve registered my generator in the hub, when asked I replied it was a NSFW generator, the application was apparently accepted, but I’m still nowhere to be found in the hub… 😅