ValKalAstra

I've had some success with an - ugh - flat ass deformer (https://marketplace.secondlife.com/p/Avatar-Clinic-Ass-Enhancers-Flat-Ass-DEMO/19361119). It needed some overall tweaking of the lower body shape still but it wasn't too round anymore. Oddly enough I also needed to increase my butt size to 30 because anything below just saw it balloon again. Can't take a picture right now but maybe give the demo a try? /edit: As for the dimple discussion, not gonna lie - me as a total blender noob looked at that and went "yup, that's a vertice gone bad while mirror editing, better smooth that out". With how pronounced it is in anything but super flat light, it seems more like an accident than intent.

This kind of echoes my sentiment. The shoulders to me seem like an improvement but the butt has turned comically large when compared to my overall physique. I'm not judging for those that prefer this, you do you - but to me this makes me rather uncomfortable for what is supposed to be an expression of myself. I've then tried to adjust the shape and for some reason the slider for butt size seems to behave strangely around 30. If I go below 30, the waist becomes larger in the left <-> right direction and if I go above 30, it enlarges in the front <-> back direction. I'll give it another try some time, tweakt he shape more and maybe see if some deformers can "fix" the issue. Til then I've got my old 5.3 body and my legacy body.

Any explanation I can give is going to be a gross oversimplification because it's all a lot more complex than that. However, let me give it a try. The issue is multi-facetted. First. Calling it AI is giving the wrong impression, it doesn't understand. To the "AI", a hand is not a hand. It has got no idea what it is. It only knows noise. To train it, it was shown how images were turned to noise and was then told to learn to predict the noise. What we do when we make these images is to tell the AI to create a noise image and then substract the predicted noise from it. Then we let it do that for several "Steps" until we get out an image. Note: an image, not the image. Noise and randomness are inherent and the image is not stored anywhere in the model, only what it learned from predicting the noise added to it. A model is an amalgamation of this process over several billion of images over multiple epoches. It has done this again and again and again until it got pretty good at predicting what noise would be added based on what fed into it. To get out an image in a direction we want, we use tokens (prompts) which are turned into vectors via embeddings and which are then fed into the noise predictor to guide. Very simplified: We put a finger on the noise prediction it has learned and tell it to predict more in THAT direction. So why does it struggle so much with some concepts and not so much with others? Three reasons of the top of my head. Complexity Quantity of available data Consumer Hardware Complexity If you look at guides on how to draw, you will often see human bodies broken down into simple shapes. For example: A head becomes a circle and an oval. An arm becomes a sphere (shoulder), cylinder (upper arm), sphere (ellbow), cylinder (lower arm). Now using just simple shapes, how would you describe a hand? Maybe sphere (hand) going into a smaller sphere (knuckle), cylinder (index finger), sphere (2nd knuckle), cylinder (index finger continued), sphere (3rd knuckle), cone (finger tip). It's immediately more complex to describe and some say, even famous comic book artists never quite learned how to draw feet (hi Rob Leifeld <3). While the AI doesn't understand what a hand is, it does however run into issues because of the complexity of hands. This somewhat leads into 2. Quantity of available data Models were trained on ginormous datasets of publicly hosted images. We're talking billions upon billions of images. From here, you can do some funny experiments to figure out why certain aspects work much better in AI than others. For example, if you do an image search for woman... you're going to find a lot of portraits, usually in a certain type of posture. A mid ranged shot is going to be rarer and a full body shot even more rare. As a result you're going to have the most success in getting an image that is a portrait of a woman. Those work really well with little mistakes. You're going to run into some struggles with mid range shots and full body shots - oh boy, good luck. Now action shots? Oh hell naw, you're in for a world of trouble now. Very little data combined with high complexity of motions and even skin deformation (stretch, compression, etc). Now hands. Just do an image search for hands and you're going to find hands going all directions, disembodied, drawn, fingers curled or straightened, fingers locked, hands touching things. It's an incredible complex data set that is hard to describe and doesn't have a lot of training data either. There are also follow up issues. For example breast size. Look at drawings of women on the internet. If you were to say that a lot of them come out rather busty, you'd be correct. If you were to add that anime drawings especially tend to defy gravity, you'd get bonus points. Now consider that these are also part of the training data and you get the reason why the model is so dang thirsty. Version 1.5 especially because it was blended with a leak of an anime model early on. So if you want better hands, you would need a lot more images of hands in the dataset. Something that quickly runs into a different problem. Consumer Hardware Right now, the technology needs a beast of a computer to create images. It just so is within the range of consumer computers. There's a bit of a problem with that, aside from excluding large amounts of people. The current implementation on 1.5 isn't using 512x512, the size of the images it was trained on. It's using a latent space of 64x64 times 4 channels. It's going to work out okay with portraits, because the subject matter is big enough. It'll work with buildings and other things. Now take eyes. They're going to be extremely tiny at the range of a portrait and even smaller in a mid range shot. At some point, they're not even blips anymore and hands suffer the same fate. They're comparatively small while also being way more complex than a face and having way less training in the model because there were less images available. You can see this quite clearly if you toy around with Dalle3 or midjourney which can throw a lot more power at the problem, using models with more training data and (probably) larger latent spaces. That's why the AI does weird stuff. All it does is predict noise and we put a finger on the prediction. Sometimes what we want is too complex or too small in the composition to come out well. Workarounds Not all is lost. There ARE hands in the dataset. There ARE closeups of eyes in the dataset. What we can do is use further guidance, higher resolutions or targeted inpainting. Further guidance is achieved by using Controlnets. These weigh on the prediction and guide it towards a certain pose or concept. It's not perfect but much more reliable than not using it. Higher resolution increases the space it can work and generate with, to a degree (these models have upper limits). More than that, use targeted inpainting. Same logic really. If the hand is coming out too small and thus scuffed, we inpaint just the hand and then tell it to try that area again. To the model, it only needs to look at the hand now and can draw from the many close up sketches of it. If your computer crashes out at that step, you could always cut out the part you want to try again in an image editor and only put that back into Image 2 Image. Then when done, paste it back in. If you're feeling brave, a1111 has got a multitude of extensions that make the VRAM issue non existent. There's the multidiffusion extension with the combined Tiled VAE. The latter allows you to essentially "split" part of the process, letting you go much further than you would otherwise. There's also the Ultimate SD Upscale script or the option of using a tile controlnet to upscale an image. These are pretty powerful and allow you to upscale a 512x512 image to whatever resolution you want. Other work arounds: Use other models like Dalle3 for a source image, then blend with control net on your SL image.

Meh, the second someone tries to talk mad ***** about you, there's a sudden dramatic overabundance of flickering candle light in the room. A stranger is sitting on a chair that wasn't there before and chuckles, his face masked in a cloak of shadows. "Are you sure you want to cross that threshold, dear?"

*raises hand* That happened to me as well on the PBR Alpha Viewer - I contributed to a Jira about it but before I could go in depth, my old computer died on me. Haven't had it happen on the new one but there was also a version update for the Alpha Viewer meanwhile so might have been fixed as well.

I'm the resident techie adjacent gal and I give this my seal of approval. Good explanation! Woah - you've made leaps and bounds with these already! Yah, having the correct resolution is very important. SD 1.5 works best in the range of 512-768. Deviate too strongly and you're going to lose a considerable amount of quality and cohesion. The crux with that is that you really want a higher resolution for things like faces and hands. There's a little trick you can do and that is head on over to inpaint, set it at a low denoise of give or take 0.4, switch to "only masked", mask out the face and then run the same prompt at the highest resolution you can. This will give that small area more resolution to work with (and for those that would like to automate that: adetailer extension does this for you under a1111).

In my experience, there's no dedicated group of product testers. However don't underestimate the reach of your social networks. You might not have a regular friend group to test stuff but you might know someone that does. Failing that you can check out the Wanted Forum but depending on how much testing you're looking at, you might need to talk compensation. As for places where you can test: There are plenty of sandboxes to try things at.

Okay - genuine question. What's the thread even about at this point anymore?

Okay but... in my experience of trying to break into clothing creation, that's not partial to GenX. In my efforts for example, I have been getting flat out ignored at all of the major body brands. ---- As to the thread itself: Creating and fitting clothes is hours of work. Let's do some fun conversions! Average salary for a self employed 3d artist was give or take $33 an hour (glass door). Let's say they've got a decent workflow. Going through all the extreme ends of a body's shape sliders, making all the adjustments, rigging it perfectly, testing it inside of SL on the betagrid, adding and configuring the scripts, setting up proper packaging, testing the packaging, create listings, update vendors, make ad pictures... Let's say they're a shiny lightning god of work and finish in two hours **somehow**. Two hours at $33. Give or take 15.000L$. That is not even getting into how you're getting double tapped by various conversion fees in SL. Let's say you sell at 250L$ a colour. That's sixty items you need to sell for that body type just to break even. Now, with all the setup and fiddliness involved, with transaction fees and market fees and technical issues, you're not only looking at probably needing way, way more than two hours to go from mesh and texturing done to rigging and packaging it in SL and being sale ready, but you're also gonna need to offset the fees too. Now how realistic is it that a body with a distribution of aforementioned 700 users as the stats mentioned above put it at, will buy your item sixty times? Now consider that I have put the needed time investment at a very optimistic (read: unrealistic) estimate and it's going to need way more than 60 items just to break even.

Guess I know my next Blender Project *glances at the skirt turned hotpants with flower ribbons*

Some random Band-Maid.

Aye - that's where the negative prompts come in handy. Put "Nude" or "Naked" in the negative prompt field (I think you need to toggle that to active on playgroundai). If it still comes out undressed, you can increase the weight of those words like so: (nude:1.5).

Oh, sorry - I kind of totally missed your post there. There are somewhat easier ways but at the end of the day, the technology is still pretty early and user unfriendly. For local installs, Some do swear by EasyDiffusion, with the focus of it being a "one click installer" that takes care of the technical aspects. I can't vouch for it as I never used it but @Istelathis seems to be using it https://github.com/easydiffusion/easydiffusion But... yah. It's still user unfriendly. There are also some websites that aim to provide a more user friendly alternative. I haven't kept up with them but most offer some free generations per day and then ask you to pay but as long as you stay below the threshold, it should be fine. One example might be https://playgroundai.com/ (caveat: You need to remember to mark your session private there and they tend to label some things differently, such as denoising to image strength). Good enough to toy around with though. /edit: PlaygroundAi seems to be 500 free images per day

I've only now seen the "little" spat in the picture thread. To summarize: *sighs*. Still in light of that whole thing, I'll still reiterate that your picture very much speaks "Second Life" to me. It could come up on my Flickr feed, I'd look at it, go "woah, that's amazing" and not even think that it might not be SL. It would fit in a long line of SL pictures I've seen from various people that awed me and inspired me to do better myself. In general - yah I think Scylla has got the gist of it there. The sentiment doesn't seem to be against editing as much as it is about the current AI war shenanigans. Case in point, those apps that put smiles and teeth on avatars a while back were actually based on machine learning (aka AI) too but those didn't warrant a new thread - so yah, it's just trying to get ahead of the AI battlefield.

You're doing the same thing I often tend to do, which is to put yourself down. Stop it. These are absolutely lovely, creative and on top of that really well made. Be proud of what you have achieved here, they look amazing and I am happy to see you experiment with greenscreen! You're always going to run into some numbskull purist trying to gatekeep a hobby. Heck there's a prominent forum member that will absoutely jump down your throat for the mere suggestion of using Black Dragon. I find it best to just kind of acknowledge them and then not engage with them on that subject further. Let them rage. I'd rather be over here, enjoying my time with what I want to make. So, is your picture still SL? For me, absolutely. The subject matter is very much your avatar and it just works. It's blended together nicely too. Something like that could easily be a backdrop with baked lighting, it's nothing I would find unfeasible in SL. Don't mind the grumpies. Easier said than done, I know but - you do good.

I mean, even popular fiction has the monkey king scrawl graffiti on buddha's finger - so I'll say, there's room to go and the other half of what Sun Wukong did kind of fits with the overall theme of, "yo my dude, that's some <insert reference> judgement".

Pet Peeve: When you make Pasta chips for yourself, let other people try some. Come back a bit later and find your Pasta Chips not only completely gone but also get a review along the lines of "too salty". You've got to be...

Please don't let that be... *looks it up* Awwwwwww. Okay. So, I'm one of those that enjoy the boyfriend type shirts. Something about crashing people with a bit of non conforming gendered clothing speaks to me. Perfect for when I get my Tomboy phases. As such, it was obviously my duty to check out this garment. Review: I think it's... okay? Especially for the current prize point, you are getting a decent amount of stuff and even with some quality compromises, it's still better than a lot of other offerings on the grid. However, it gives me the feeling of being an older work of theirs. Some notable mesh and texture issues and yah, the whole fit doesn't really make sense. That said, it won't become part of my wardrobe and I'm the target audience

I'm kind of hypocritical in how I use it. I like seeing who looks at my avatar. It's a bit of vanity and like it was mentioned, you can just tell when there's an incoming instant message. However I've turned my own look at off. There are some moderately unhinged people on the grid. The only drama I ever got into was over a look at indicator. One example was that I was at a shopping event, cammed somewhere (an ad, not a person), went afk and came back to someone blowing up something nuclear in my instant messages over gawking at their partner. It has been turned off since. Too much of a drama magnet.

You, my dear fluffy lion, need our lord and saviour Rebecca Black, patron saint of the most hated music video about this fine day of the week. May the bubbly lyrics soothe your peeve like a lightning rod.

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.

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. 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.

Sadly: the lack of rez rights. The bane of photographers on the grid. OP mentioned doing it at a competition with no rez permissions.

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.

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. 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): 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.