Jump to content

ValKalAstra

Resident
  • Posts

    508
  • Joined

  • Last visited

Posts posted by ValKalAstra

  1. I think moderation can have a chilling effect going on. For example I don't think I've ever gotten a warning. As far as I know I have not gotten close to one either but I've seen a few threads I took part in just vanish lately. Not a peeve, just... kind of makes me wary of posting. But this probably gets uncomfortably close to discussing moderation so I'll zip it.

    Peeve time then.

    On a personal level. F this year. Never in my entire life have I experienced such a non stop storm of calamity wreak havoc on my life, my health, really just everything. It's just one thing after another after another after another. It peeves me how much strength it saps out of me to just get up, NOT turn cynical, NOT turn to apathy. It's been enough. No exaggeration, not a single week has gone by without some catastrophic event necessiting repairs, doctor visits, intervention, long gruelinghours of emotional talks, navigating past angry people, dealing with hate IRL, just no more energy.

    Argh!

    Okay.

    Yah.

    Alright.

    No.

    F this year.

    • Like 8
  2. At first I wanted to protest that it really isn't that hard to find modest clothing in Second Life. When I saw bareshoulders counting as oversexualized, it began making sense though. So let me think.

    Addams has got a wide variety of casual looks and generally works really well for mixing and matching. Even if one outfit might show cleavage, you can usually slot in a top from a different outfit. It also tends to play nice with Blueberry but those outfits would probably be too risky for your tastes.

    There's the already mentioned Ison and... gosh, I keep forgetting the name - a smaller store for french fashion that always escapes my mind. Either way, if you're okay with cute clothing (think korean drama fashion) then you might enjoy some of the side rooms of Zenith. Hilly Haalan then offers such a wide variety of clothes, that you're bound to find something fitting, even say - a business suit for female avatars.

    Oh yeah - a bit of a hint what you can do:

    You can mix and match with BOM clothing. I've got a completely black tanktop as BOM that goes under everything and it's a neat trick to "modest up" otherwise risky outfits. Might be worth grabbing a BOM top with mod permissions so you can colourmatch it and further cover it with accesories. Let me take an example image quick!

    Left image is without the BOM top, belt and smartphone that I use to cover clipping on the belt. The right is... well the same but with an added top that blends in really well to the point it might even appear to be one piece (no pirate treasure, sorry :P), in my opinion.

    bomxample1.png.caec0732db90c05367d4918d53af72cd.pngbomxample2.png.5a3e64aa3be3f959fafecc56660aa2a6.png

     

    Edit after editing the edit:

    Obviously the example is not what OP would probably consider modest - but it should illustrate the usefulness of blending mesh with BOM to cover up parts you want covered up.

    • Like 2
    • Thanks 1
  3. 3 minutes ago, Love Zhaoying said:

    I had seen some articles saying AI is bad at getting "human hands" right (# of fingers, positions, etc.) so you'd think that a cat's legs would be easier (or maube not).

    Hands need some heavy lifting to get right. It's a definitive weakness for the reason you've mentioned. It can be done though. As for why the cat is a bit of a mess - that happens with multiple subjects in an image. You'll need to either do so called regional prompting or inpainting. To compare, here is a majestic chonker with the right amount of legs.

    00018-87301247.thumb.png.834c6d5b658cee9547b81f42000e83db.png

    • Like 6
  4. If I may - a few words on the various settings and what they do:

    CFG or Guidance Scale - Governs how much you let the AI off the leash. Low CFG values means it all but ignores your prompt while high CFG means it will strictly adhere to the prompt but neglect any lessons it knows about image composition and such. A value of 6.5 - 7 is usually ideal unless you explicitly want it to go wild or strict.

    Steps and Sampler - These need to be configured in unison. A sampler is what turns an image from noise into the result. Each sampler needs at least a certain amount of steps to work but anything past that is just pointlessly blowing out GPU power. Exception so called ancestral samplers (usually marked with an a in their name like "Euler a"), these can run forever, continously adding new noise. As a rule of thumb: Most will do fine with 25 Steps. DDIM will work with 10, Euler is older and may need 80 or so.

    Denoise or Denoising - How much the image is meant to change from the source image you give it, ranging from 0 to 1,0. A value of 1 will all but throw out the image and a 0 will do no changes at all. This is what you will need to find a decent balance with.

    The choice of the model is also an important one and it helps to understand how and why things came to be. There are three major revisions of Stable Diffusion. 1.5, 2.x and SDXL. Out of these you can outright dismiss 2.x, it was a failed attempt. 2.x tried to remove the nudity from the model and very quickly realised that nudity was what made the model understand human anatomy. The resulting images were a mess of uncanny horrors.

    If you've got a good PC, use SDXL and 1.5 If you have got a less good PC, use 1.5. That leaves the choice of the model and here it helps to understand that there were for the most part three major development paths that have all converged by now. Anime, based on leaked Anime models (usually NovalAI), base stable diffusion and... asian girls. There was a lot of hype from korea early on. These have all merged into one by now just with various weights attributed to this or that and maybe some additional training in certain aspects.

    Newer models generally work much better than older models. NSFW models will do better with anatomy (including hands) but need heavy counter prompting to put on clothes. The merge with anime models means that unless you want massive tracts of land, you'll need to actually prompt OUT certain anatomic sizes. Furthermore, it helps to understand that the Anime models used a different style of prompting, based on booru tags (no, that is not safe for work to google).

    If you want certain clothes, you need to prompt for them.

    Last but not least, keep prompt length and image resolution in mind. SD 1.5 was trained on 512x512, you can go up to 768 but anything beyond that goes kaputt. Use upscaling. SDXL was trained even more rigidly on a handful of resolutions around 1024x1024 and can also go +-256 in either direction (bit more but eh, find a cheatsheet yourself :P). Prompt length then, each prompt is split into one or several brackets of 75 tokens. This is important to know because it will govern things pretty strongly. Let's say you've got 76 tokens in your prompt, then the first 75 tokens will form one bracket - and the remaining token will form a bracket on its own. Then both brackets will be weighed equally - resulting in that one token coming on super strong. So stay below 75 tokens ideally or look into "BREAK"ing prompts.

    Alright, enough crash course SD. Took some time to try a more direct conversion without my LORA. This was with a denoising value of 0.6, some prompting and a lot of cherry picking on SDXL. SDXL tends to make faces come out a bit strong on bone structure and none of the stable diffusion models can deal with my eye colour in any reliable shape or form.

    sltoai.thumb.png.bc91638c1ba6a3a142a1c1c3058a92f0.pngsltoaires.thumb.png.4c0f4d28b94f31e2beb7044fd175c12b.png

    lf you want more consistency than that with Stable Diffusion, you either want to make heavy use of controlnets or train your own LORA. Something I've done together with a friend of mine. That's also what allows you to translate images into different styles more easily. Above result is cherry-picked, below result is not (but could be better):

    sltoaires2.thumb.png.5d73271ffe51f438fdda3c56a4bf1549.png

    Footnote: Talking about bias in training data, there's a funny social one and that's age. Most of us want to appear younger than we are, as we get older. Some won't answer the age question truthfully. Look, long story short, the training data thinks a fourty year old person will look like an actually sixty year old person, because so many sixty year olds lied about their age being fourty - and furthermore, so did the fourty year olds, which you can find in thirty year old category and if you want thirty year olds, you're gonna find them in the 20-25 age bracket. I found that funny :D.
     

    Edit:

    Oh, oh, an interesting use case for sl photography! You can push a sl screenshot through image to image, mid denoise and then take cues on lighting your scene more believably. If you compare the three shots, the biggest flaw of my sl one is the flat frontal lighting.

    Heck, you can even use it as a little indicator where to manually add highlights in post.

    • Like 5
    • Thanks 5
  5. 1 hour ago, Istelathis said:

    I think it is a puma, it surprised me when it showed up though 🤣  I thought you might notice that one.

    I decided to see what would happen if I changed the model to the default, using realistic and keeping all of the other settings the same and this is what it created:

    Model: sd-v1-4

    Image Modifier: Realistic

    Realistic_S0_St25_G7.5(1).jpeg.c073fa30e5b9553dd814de53c0b92529.jpeg

     

    I don't like the model that easy diffusion comes with by default it makes pretty scary images 😭😨🤣

     

    Using the model deliberate_v3 with the same settings just created a room, which looked nice but had no avatar in it.  I switched the model to  dreamshaper_8 and this is what I get for Realistic:

    Realistic_S0_St25_G7.5.jpeg.9bf5761f7f0dbd1a9beb3efd82f34462.jpeg

     

    I have a bunch of other models, but the HDD I have them stored on is slow as molasses and it takes forever for them to load into memory so I can't try out too many of them for now, I will goof with a few other settings though and see what happens.

    Changing the Guidance Scale from 7.5 to 1.1 resulted in this:

    Realistic_S0_St25_G1.1.jpeg.a74c6d4b2ea1910abb6b12ba50b9030b.jpeg

    I'm not sure what that creature is, I think it is friendly though.

    Changing the prompt strength in Easy Diffusion seems to be pretty important:

    Realistic_S0_St25_G14.1.jpeg.b97eb7db41b36e94dd827eb785d8f0e3.jpeg

     

    Bleh, so many things to fiddle around with.  I could spend hours trying to figure it out, and I most likely will.  I'll do one more before going outside and touching some grass 🤣

    Watercolor_S0_St25_G1.1.jpeg.579dda6d94a41f44fed56eed2e8e2b6d.jpeg

     

    Seed: 0, Dimensions: 968x512, Sampler: dpmpp_2m, Inference Steps: 25, Guidance Scale: 1.1, Model: dreamshaper_8, Prompt Strength: 0.5, Preserve Color Profile: true
    Processed 1 images in 1 minute 13 seconds

    These are cool though. Did you try using a Canny and Depth Control.net, perhaps? Although seeing the inference times, it might be a bit taxing on the system? I haven't checked EasyDiffusion, I'm more of an automatic1111 girl.

    And yeah, totally - the sd 1.4 model is absolutely ancient at this point, basically to the point of being no longer useful. Deliberate is a good general purpose model, dreamshaper works well for fantasy images. Another decent model is RevAnimated - although you absolutely WILL want to add nude to the negative prompts there. It's thirsty - but produces great clothed images too. Just needs convincing. :D

     

    • Like 3
  6. Oh, me, me! I've been dabbling in AI stuff too. My first tries were with Midjourney more or less a year ago and when I could no longer afford that - I stumbled accross Stable Diffusion. That has been my go-to and I have spent a lot of time experimenting and playing with it. I've since experimented a lot in trying to bring my looks over into other concepts, such as drawings, paintings etc. So it's not going to be a direct Image 2 Image but rather something created with probably the... 8th or 9th generation of LORA and custom blend models.

    Either way, some of my experiments with it. As comparison my Avi from SL first.

    Yet another portrait practice

    In case someone warrants it, my signature stare in custom sticker form - I need to create more of these!

    kal_stare.png.847af29a73064eb1544897c1c5843e04.png

    A shot of Anime Nina and vibrant water colour!

    animekal.thumb.jpg.a1425d31e4dc9dff11ddcfe7f55e579d.jpg

    Nina has played too much Eldenring!

    drawingkal.thumb.jpg.7b4bba6124f7fc2a6dac27643ccd9818.jpg

    And Nina got style - at least in the eyes of the all spooky AI.

    fashionstylekal.thumb.jpg.f31ccffcc79ac7c861cdd93979f4ff6f.jpg

    • Like 12
  7. 5 hours ago, Orwar said:

       I think the MP clean-up routine they have sounds quite reasonable. There's a whole lot of old junk still but either people must be buying it or the merchant is still kicking around - in which case it's their prerogative to have stuff on the MP.

       In-world .. Well as long as the land is being paid for, why should it be treated any differently?

    Aye. If it still sells or the merchant is still around? There's in my opinion no reason to yank it. Sure there might be some ancient, half broken mess in there but sometimes there's some charm in that. Also I have seen people buy said mess and dance with joy over their purchase. I didn't quite understand it but heck, joy is infectious so let them have it! I've had my fair share of fun in going into obviously old builds and almost looking at it like a virtual archeologist. What could this have been? Which weird religions must this have been made for?

    Personally, I'll toss a coin to the witcher coder that implements a better way for users to find what they want. Both in world and on the marketplace and let's just say the latest marketplace changes aren't it. I've gone from a frequent marketplace enjoyer to someone that looks at it, groans under a flood of gacha that refuse to be filtered and heads back out.

     

    • Like 2
  8. 2 hours ago, BilliJo Aldrin said:

    one more for the block list

    Well then, feel free to add me as well but you need to hear this and ideally understand this: A graphics card overheating is not only not working okay, it's a suicidal firehazard to not fix it. That's neither a joke, exaggeration or me being an evul gamer grrrrl sneering down at you - that's a genuine safety warning.

    A GPU is built to sustain heavy loads for an extended amount of time. It is also build with tolerances that some people use to overclock them. Your GPU, by your own words, is both underperforming compared to cards of similar makeup and still punching way past the tolerances by overheating anyway.

    Block me, ignore me, insult me if you must but bloody hell:

    Fix your GPU. Do it, please. Clean out the dust. Check the thermal paste. Buy an old cheap one on Amazon. Switch out the fans (probably costs more than just switching the GPU). Do any of these things. If you don't know how, there's people that will help. But hell - A GPU overheating is running so hot that your computer goes "holy hell, fire hazard!!!" and emergency shuts down everything to prevent a quite literal fire. Don't bloody chance it. Period.

    Don't be like a friend of mine that lost nearly everything because he didn't bother to clean up his dust caked up notebook despite constantly overheating and it caught fire one day.

    • Like 13
  9. 5 hours ago, Scylla Rhiadra said:

    If ever there was a good case for the use of ChatGPT or some other form of automated response, responding to this kind of suggestion (which pops up here, what? once every couple of months?) is it.

    Curious Nina is being curious:

    Quote

    Dear [Name of person],

    Thank you for sharing your thoughts on removing free users from our service. While we appreciate your input, we respectfully disagree with this suggestion.

    At our company, we believe in providing a wide range of options for all users, including those who may not be in a position to pay for premium services. We offer free access as a way to introduce people to our product and provide them with an opportunity to learn more about what we have to offer.

    Additionally, our free users are an essential part of our community. They help us to gain valuable insights for improving our service, provide feedback on new features, and share their experiences with others.

    We understand that it can be frustrating to feel like some users are not contributing to the financial success of our company. However, we believe that the benefits of having a diverse group of users outweigh any financial concerns.

    Thank you again for sharing your perspective, and we value your continued support of our service.

    Sincerely, [Your name]

    Yah, that'll do. Perfect amount of corpo nothing speak while telling them to knock it off. As for the topic: Fun fact! Doing that is a popular solution for club owners who want out of the biz, actually. It's this super cool trick they can use. Just build a landing spot, mandatory group and group join fee going four digits. That generally kills a majority of visitors and the rest leaves in due time because "great now folks are still standing around afk but the amount of readable profiles has gone way down, why am I even here?"

     

    • Like 1
    • Thanks 2
  10. I'm reminded how it works on reddit. A user you have blocked can't reply to your posts anymore. It's one of those ideas that's good on paper until you've seen it in action. That function is abused for so many actions done in bad faith it's ridiculous. Two people are having an argument? One blocks the other and gets to do a victory lap, mic dropping the last shot. Want to spread misinformation - just block any prominent member that could debunk you and let the FUD fly until someone else corrects it hours later. By that point, it's been spread far and wide.

    However not once have I seen it used as a way to stop a fight between two people. It was always a tool to get one up on someone else. No surprise: I'm against the functionality of giving users a tool to dictate how others can interact with a public forum. That's a power reserved for site owners and moderators, not people that overextended in an argument.

    • Like 5
    • Thanks 2
  11. 13 minutes ago, Lewis Nerd said:

    Then I have no option but to reluctantly make you the first person I add to the forum ban list so I no longer have to interact with you.

    I don't respond to threats.

    I mean if you took what Zalificent said as a threat - then you quite literally just did in fact respond to a threat. Yes I'm fun at parties but you probably wouldn't know, they're for... grown ups.

    image.gif.17a0e5995dae794036de158b4a6fa40e.gif

    Jokes aside - I wanted to point out that seeing adult content is actually an option that you have to opt in. At least that's how it was for me when I joined give or take two years ago. It was turned off by default, both on the marketplace and in my viewer of choice (Firestorm - I never even touched SL official). Also maybe it's just a bug on my end but when I head to the marketplace not logged in, I can't even select anything above General so - yah. The default does seem to be for adult stuff to be opt-in, rather than opt-out.

    • Like 5
    • Confused 1
  12. 16 hours ago, Marigold Devin said:

    I had no sense of smell then little sense of smell after having covid several weeks ago. I can taste maybe as third as well as I used to be able to and just enjoy food mostly for the sensation of salt and sweet or texturally.

    For me that was give or take a year ago. Someone in the family combined the trifecta of being easily infected (heavy smoker), not giving a single damn and constantly heading out to socialise.

    He's fine now!

    MY sense of taste however noped out after the fifth infection, which funny enough I didn't even feel sick foranymore. Took me til now to get it somewhat back. Can only really taste salt and texture or very strong sweetness. On the upside, it finally forced me to learn cooking after having lived a life of everyone comparing my cooking to flamethrowers.

    What I meant to say is that I empathise with you. We don't quite notice something properly til it's gone aye?

    • Like 1
    • Thanks 1
  13. 1 minute ago, Love Zhaoying said:

    I fold, otherwise I'll be calling your bluff and all I really have in my hand is J-Rock (not J-Pop).

    ..staying with Poker metaphors, that is.

     

    And here I was ready to counter with this glorious loreal commercial:

    I'd chip in for poker puns but I'm genuinely blind to what any two cards in the game even mean, so before i accidentally shorthand you, I happily fold!

    • Thanks 1
  14. 25 minutes ago, Love Zhaoying said:

    I see your Bubbly K-Pop and raise you some Bubbly J-Pop.

     

    Well, well - I'll raise you some bubbly ... cpop? Is that actually a thing? Dunno. I raise you with a Hong Kong musician, first debuted in Korea, here singing in English about a Chinese ghost love story. Also most dramatic clothing scene ever.

     

    • Like 1
  15. Confession. Sometimes when I am semi zoning to some editing or forum reading, I throw on some video game soundtrack with oomph behind it. Something like the Theme for the Imposter King can turn everything epic.

    Click.

    Click.

    *DA DA DA DAAAAAa - ANGELS START SINGING, THE WORLD ACOUSTICALLY ENDS IN A FINAL STAND*

    "Yup. That about sums up reddit today."

    Click.

    ... Click.

     

     

    • Like 1
  16. Nina's spicy hot takes, aka unpopular opinions:

    • A lot of threads serve only one purpose: Social Drama Fireworks. Everybody knows this. Everybody lights the fuse anyways.
    • Some just want to see the Forum burn and see every thread as an opportunity for the first.
    • They would never agree but - Second Life is actually a fun place with fun people of many different walks.
    • Like 6
  17. 1 minute ago, Orwar said:

       But can it mesh and rig the character model for an SL avi/NPC? 

    The tech is getting there but it ain't close yet. Most companies learned from the whole Stable Diffusion mess and are keeping things behind closed doors so it's hard to say how good things are but what's out in the open is middling so far. You can generate mesh, yes. However the mesh is a complete and utter mess. You could probably rig it but at the current state of things, it would set SL on fire.

    https://github.com/threestudio-project/threestudio

    What works really well is to bake textures onto mesh objects. You could create cool photo backdrops that way. I won't say easily because nothing about manual installs of the tech is easy - but it can be done. You could also create skyboxes rather easily (with say, https://skybox.blockadelabs.com/ and modifying it into the proper projection). There's pleny of AI on the grid already, usually in the form of low effort images though. As someone that does learn and experiment with AI - I get ticked off when I see typical AI mistakes someone couldn't be bothered to fix and then sells it anyway. But that's just me.

    • Like 2
×
×
  • Create New...