andypuk1980

joined 3 months ago
sorted by: new top controversial old
When will the quality for image gen increased its been long in c/perchance@lemmy.world
[–] andypuk1980@lemmy.world 1 points 1 month ago

I've noticed the past few weeks that when selecting the oil painting style in the text to image generator, it just keeps putting in all kinds of random styles that I don't want, cartoon, photo, anime, so yes the image quality is good when it gets it right, but right now it's just throwing out all kinds of unwanted random stuff, some good, some bad.

0
Painfully slow image generation (lemmy.world)
 

Trying to use the text to image generator and it's so so painfully slow for me, it doesn't matter whether I use Firefox or Chrome, disabling ad blockers makes no difference, clearing cookies and cache hasn't made any difference, it's just so slow generating images, no matter what time I go on to use it. Firefox seems to be the worst with the fetch failure and invalid access code errors, is there anything I can do to make it go faster and stop these errors from popping up? Is there another browser that's better for Perchance other than Firefox or Chrome? I'm also getting this error Firefox - An error has occurred somewhere in your code (in lists or HTML) "onKeyUpMouseUpTouchEnd/<@moz-extension://4f0a3e99-bb2d-414f-91a0-e8b033445489/js/inject.js:58:9" and "onKeyDownMouseDownToushStart@moz-extension://4f0a3e99-bb2d-414f-91a0-e8b033445489/js/inject.js:49:7" which I assume is to do with other browser extensions but no idea what these errors mean or what extensions it's referring to. If anyone has any ideas or advise on resolving this that would be great.

Text to image generator problem with inconsistency despite detailed prompt in c/perchance@lemmy.world
[–] andypuk1980@lemmy.world 3 points 3 months ago

Yeah I've tried a few different techniques, even bullet points combined with parenthesis, but everything just keeps moving around into the wrong places. Adding a wristwatch starts putting a big dial on the belt buckle, making them wear different types of shoes or boots just gets swapped around, even tattoos swap around and get placed on the wrong body part. It is frustrating but I've just generated 6 images, picked the best one that is closest to what I requested, opened a new session and copied and pasted the seed over and generated 6 images of that one picture and just keep refreshing until something looks ok.

2
Text to image generator problem with inconsistency despite detailed prompt (lemmy.world)
submitted 3 months ago* (last edited 3 months ago) by andypuk1980@lemmy.world to c/perchance@lemmy.world
3 comments fedilink
 

So I just recently started playing around with the text to image generator, however I’m having no end of trouble trying to create an image of two young ladies wearing light blue jeans and white tank tops talking to each other, I'm asking it to generate at least 6 different images in painterly style, using square for the shape to try and get something reasonably accurate and decent, I’ve specified what other clothes and makeup they are wearing, each different from the other, for example one is wearing a black leather belt on the jeans with a square silver buckle, the other a dark brown leather belt with a gold square buckle, gone into alot of specific details, even naming the two young ladies, however the generator keeps swapping the belts around, it’s also doing this for the makeup and jewellery. Everything else seems to be pretty consistent such as the hair colors and styles, eye colors, where they are positioned, the scene setting etc. So my question is, how do I word the prompts to stop the swapping? Do I need someone to take a look at my positive and negative prompts and correct any mistakes? If anyone can answer this I will be the happiest AI text to image generator newbie in the world :)