mFeelzGood said: This thread is crazy It is fun to see how many people posted, the randomness, how many views it got, and how many pages of replies it got.
This was an interesting idea. It got a lot of people interested and engaged.
I do like the idea of organizing now. It would be nice to start up a nice pattern. This would help people find what they are looking for rather than have to go through this thread page by page. People can follow threads for topics they are interested in. It is hard to think of the best pattern.
I'm inclined to agree with the above. As more and more people start producing and posting AI images, this thread will become untenable fairly quickly as those looking at the content may not be willing to scroll very far back. Consequently, image sets could quickly become "lost" and, ultimately, if people aren't viewing/liking them, there's little point in posting them.
It also seems to me that there's noticeably more interest in the pie/gunge images than there is in the muddy/wet ones, so people having their own threads for their style of content would mean that people wouldn't have to scroll through stuff they're less interested in to find things that they do want to see.
The only caveat I'd put on that is that people posting on threads they created themselves typically won't bump the thread back to the top (even the "Last Post" date won't change), so it won't be possible for others to know that a thread has been updated. Therefore, they'll need to either follow the thread or the person whose thread it is. That's fine, as long as people understand that they need to do that if they particularly like a thread.
I have suggested to MM that posts on threads which won't bump it back to the top with the current algorithm should be bumped if they contain at least X number of images (I suggested 10 to prevent people from spamming with single images).
For now, I think I'm going to generate my own thread for the themed content that I've been posting in this one. Ideally, I'd also move the other posts here on that theme across to the new thread to preserve the storyline (such that it is), but I don't think there's any way of doing that.
HideoKojima said:
chryslerfire said: i made this thread as a place for everyone to post whatever ai they can come up with but i do wonder if im posting to many images or filling up the picture feed it seems popular and there's alot of images on this thread now.
im not sure if its better to have a thread for each substance or outfit or whatever. maybe it is better to put it all in one place
I believe its fine, I was worried at first but now we can tag our content as "synthetic" which shouldnt clog up the actual feed? I might be wrong.
My main concern is like, pushing out actual creators out of the image archive and stuff, I would hate it if someone worked hard on an amazing scene, only for it to be pushed aside for AI stuff, but if the tag works how I think it does, I think were ok.
mFeelzGood said: I had a similar concern because the synthetic filter is disabled by default, so the majority of users are seeing this on the Wall. I was also concerned that a producers scene might get less views because it gets bumped down on the Wall.
I was thinking, it might be nice to have a "Enable post on Wall" flag when replying that is disabled by default. It can be disabled by default. When people want to show some interesting results, they can keep it disabled. If people want to show their best results that turned out really good and are proud of, they can enable the "Enable post on Wall" checkbox.
I think that MM is pretty far down the road of enabling people to choose to omit AI images and, once he's finished, people can be pointed at the "Block Synthetic Content" setting in the Content Filters section of Preferences. I would imagine this would include images on the Wall, so I'm not sure that another setting is needed? I'm proud of everything I share, otherwise I wouldn't waste my time posting it! On that basis, I wouldn't choose not to display my content on the Wall.
The only issue I can foresee is for people who like some AI images but not most of them, so they might be reluctant to block all AI content. I'm not sure what the solution to that is though....
SmushingTin said: I'm spending way too much time doing this, I think its a kind of addiction
Same - I've been generating explicitly 'fetish' images of things that only exist in my head, but these have crossed over into non-fetish things like fantastic landscapes and cities. Eg this afternoon putting women in various places evolved into wonderful gothic fantasy railway station-cum-castles.
A couple of attempts to get women in 'puffy wear' wet and muddy...
SmushingTin said: I'm spending way too much time doing this, I think its a kind of addiction
Same - I've been generating explicitly 'fetish' images of things that only exist in my head, but these have crossed over into non-fetish things like fantastic landscapes and cities. Eg this afternoon putting women in various places evolved into wonderful gothic fantasy railway station-cum-castles.
A couple of attempts to get women in 'puffy wear' wet and muddy...
These are absolutely amazing, well done, thank you.
I took some of the pie shots from the Bing Creations here and made a LoRA from them and added them to my current LoRAs and I reckon they have pretty much nailed it. Best pie shots I've made so far anyways, definitely more dynamic.
Wetmaxiskirts said: I think that MM is pretty far down the road of enabling people to choose to omit AI images and, once he's finished, people can be pointed at the "Block Synthetic Content" setting in the Content Filters section of Preferences. I would imagine this would include images on the Wall, so I'm not sure that another setting is needed? I'm proud of everything I share, otherwise I wouldn't waste my time posting it! On that basis, I wouldn't choose not to display my content on the Wall.
The only issue I can foresee is for people who like some AI images but not most of them, so they might be reluctant to block all AI content. I'm not sure what the solution to that is though....
I think the wall is an issue, I didn't even know it was a thing until I started posting in this forum. Personally I don't like that my pics go on there as I feel like it takes something away from the people who are making what we really all come here for. I mean from a quick glance the AI pics in the main stand out because of the quality of the image (even if they likely have 7 fingers or cross-eyed on closer inspection), they look like they are taken with really expensive cameras with perfect lighting etc...
Personally I want to be able to share my creations with you guys but I dont want to take anything away from the people who put the real effort in and I do feel like my images on the wall may do that.
kemistry said: I took some of the pie shots from the Bing Creations here and made a LoRA from them and added them to my current LoRAs and I reckon they have pretty much nailed it. Best pie shots I've made so far anyways, definitely more dynamic.
This is a brilliant idea!
I was thinking of taking the good images that I have produced which I like to make a LoRA to make it easier to produce similar images. It is annoying to have to spend so much time tweaking a prompt to get it to look right. It would be nice to make a LoRA for getting the coverage that I want without having to increase weights too much which introduces defects like missing hands.
Taking the DALLE-3 images and using them to train Stable Diffusion is genius. It looks like it worked well from your results.
kemistry said: I took some of the pie shots from the Bing Creations here and made a LoRA from them and added them to my current LoRAs and I reckon they have pretty much nailed it. Best pie shots I've made so far anyways, definitely more dynamic.
fantastic idea, I have only briefly dabbled with SDXL, its incredible, i'd probably use it more if I had a better PC, but I have 2 issues with it:
1. the mess it creates is quite restrained, its hard to create a really "engulfed" feeling like Dalle 2. the mess often feels like its oddly floating above the subjects, its almost like a photoshop
I appreciate you guys putting the effort into making models and stuff though, that shits beyond me, let me know if you need anything from me, Dalle is absolutely fantastic at mess coverage, and that "stain" that happens when you get pied in the face or slimed, the main problem is the censor, obviously.
but like, here's 2 images, my first image is an SDXL creation, I used seaart seeing as my PC sucks, that probably doesn't help with the image creation, and then the second is from Dalle, I just feel like right now, dalle is better at that overall "mess", like SDXL has this weird habit of having like, slime covering someone, then super clean next to it, thats not really how slime works right? the persons face should be sort of stained with the mess, idk. SDXL always feels like its got this weird solid feeling to it, rather than a liquid
I moved to SD after Dalle 2s censorship got too much for me, and I can imagine the same happening here after Dalle 3 gets old and SD gets better, I feel pretty good about my images, but I cant help but feel disheartened when I change 1 tiny detail and get blocked, it feels very arbitrary and random, frustrating.
Apologies for the ramble, I normally dont post on the forum at all, yet here I am, weird feeling for me, hello? I guess? lmao
kemistry said: I took some of the pie shots from the Bing Creations here and made a LoRA from them and added them to my current LoRAs and I reckon they have pretty much nailed it. Best pie shots I've made so far anyways, definitely more dynamic.
fantastic idea, I have only briefly dabbled with SDXL, its incredible, i'd probably use it more if I had a better PC, but I have 2 issues with it:
1. the mess it creates is quite restrained, its hard to create a really "engulfed" feeling like Dalle 2. the mess often feels like its oddly floating above the subjects, its almost like a photoshop
I appreciate you guys putting the effort into making models and stuff though, that shits beyond me, let me know if you need anything from me, Dalle is absolutely fantastic at mess coverage, and that "stain" that happens when you get pied in the face or slimed, the main problem is the censor, obviously.
but like, here's 2 images, my first image is an SDXL creation, I used seaart seeing as my PC sucks, that probably doesn't help with the image creation, and then the second is from Dalle, I just feel like right now, dalle is better at that overall "mess", like SDXL has this weird habit of having like, slime covering someone, then super clean next to it, thats not really how slime works right? the persons face should be sort of stained with the mess, idk. SDXL always feels like its got this weird solid feeling to it, rather than a liquid
I moved to SD after Dalle 2s censorship got too much for me, and I can imagine the same happening here after Dalle 3 gets old and SD gets better, I feel pretty good about my images, but I cant help but feel disheartened when I change 1 tiny detail and get blocked, it feels very arbitrary and random, frustrating.
Apologies for the ramble, I normally dont post on the forum at all, yet here I am, weird feeling for me, hello? I guess? lmao
Not rambling at all, love to hear your thoughts, thank you.
I'm waiting for your other forum thread with bangs bump of everything you've created. I've just learned from MM you can do it as a zip, so no fiddling around.
Finally started to make some headway getting mess, in this case mud, into pictures. A breakthrough was putting football and rugby into the prompt as it has lots of RW material of muddy players in these sports. The following are all Perchance:
I'm going to take a few day's break from this now. This weekend has been a big eye-opener and spurred my creative needs, but I can see this becoming a rabbit-hole that hoovers up a lot of my time and becomes addicting, like some kind of slot machine. Need to 'detox' for a bit.
chryslerfire said: princesses in the morning morning
Sorry I keep bugging you with this question, but what prompts did you use for all these in parts 1 and 2? I'd like to fiddle with them a bit
the main prompt i started with for this was "muddy lady in muddy prairie princess ballgown sits on muddy bed in muddy flooded pink bedroom" i did change it around a little bit.
sorry i will try to include prompts when i post in future. sometimes i just end up using so many.
i do apologise if i am posting to many will have a break after this maybe
No need to apologise, we all get very enthusiastic about playing with these tools and opening more and more doors to what they can do. These 'office scenes' are simultaneously titilating, surreal, anarchic, and hilarious.
Rainman_666 said: AI generated images from various prompts asking for winter clothing, oceans, newly bathed, fighting with pastry, etc...
I've struggled getting the AI to get puffy clothing wet as easily as rainwear. I don't think it's a censor, I think it's because it doesn't have the source material from RW images so readily. The holy grail is puffy jackets and vests splattered with cream and cake batter.
i do apologise if i am posting to many will have a break after this maybe
I stopped posting my image gens last week, given the volume I was creating, it felt I was cheapening the Forum. I'm not sure the forum is the place to dump a load of images, I started getting 100s of notifications and whilst the site is being reworked, I can imagine others are getting the same and the Wall being swamped. Hoepfully the site changes will make things a bit better all round