That's so freaking stupid. When a company known for producing 3D models of human characters announces AI features, people expect "We're going to let you quickly generate and refine full 3D characters using natural language with a custom AI model that's been trained using our extensive knowledge of 3D character creation" not "We got StabilityAI to use DreamBooth with an outdated version of Stable Diffusion so you can generate bad images of our characters or whatever, please give money"
Right? I'd expect a "woman in her 30s, long, brown, curly hair, medium height, wearing a jumpsuit" to move the sliders around and generate a 3D character matching the description. Not another Stable Diffusion wrapper.
Of course the furry is seething over tokenized self custody. Let me guess AI is taking yur jerb as well? It's hilarious when wets take construction jobs and lower the demand for labor, but its the end of the world when AI makes your non work less valuable?
If anything this adds more to my theory that AI is just another hyped thing by companies, just like crypto, NFTs, and the Metaverse… I mean think about it, all these projects came out of nowhere, causing millions to lay their eyes on it, all these projects promised that ‘it’ll change the world’ and ‘the next big thing’, they all have a strict fanbase that doesn’t take criticism, has a bunch of RU-vidrs and popular users constantly talk about it, and they all are involved in some sort of scam almost on a daily basis
@@ethanissupercool7168 I mean, it definitely has uses. But I agree the hype surrounding it is _ridiculously_ overstated. Midjourney, for instance, is far from useless, but at the end of the day, it's mostly just a stock image generator. Unlike NFTs in my opinion, these have the potential to be extremely useful, but calling it "the next internet" or "the AI revolution" or something similar is stupid
NFTs a few years back, a big announcement about Cascadeur support, that new funny animation library with a subscription model. And a few other things. Daz is jumping on every "current thing", nothing really works and we still wait for working IK/FK chains, softbodies, a working cloth sim and a usable timeline editor. And that's for years now. It seems like they just don't care about their customers.
and meanwhile they had announced DAZ Studio 5 a few years ago that was not even promising new features and some sort of staggered release, and even this they have not done yet. these guys are incompetent. they struck gold when they started out, had some good ideas in the beginning but probably could not keep the developers that had initially, and then turned to shiny new things every six months or so, just to keep the people entertained. been using daz since 2010 for work, and my frustration level is maxed out to the point I do not even care to post any suggestions on the forums anymore. they simply suck, and I am sure they are already dying a slow death. doubt they will have any sort of a resurgence in the next few years unless they drastically change course and hire proper developers.
Daz has a cool base of tools for creating characters and clothes library and they need to become competitors with CC4 and iclone and improve export to game engines, but they have been constantly going wrong for several years
The negative reactions are valid IMO. I gave this a quick go and even for a first, initial release it was terrible and incredibly lacking. There's certainly nothing that makes it worth the sub price. If this was to use the figures you already own, including custom characters, all within Daz Studio itself then it could've been really useful. Yet it appears to me that you'll need to re-buy characters you already own (not sure of the price) to use them with this. Mike: There's an option for 1-4 images per generation on the left.
Just Why? There are some genuinely cool things that they could have done with AI. Like they could have made some sort of prompt based AI system for DAZ studio where users could put in a prompt and it would try to make a 3d model for them using DAZ's models and morphs. Or a system that attempts to pose a 3d model based on prompts. But I guess those would take actual effort, this looks like something that one guy with access to DAZ's stuff and a reasonable knowledge of SD could make in a day.
absolutely right. i doubt the staff at daz have anyone with actual ml engineering skills so they're effectively banging rocks together with off the shelf models to try to cash in on the hype. there are tons of ways that a tool like daz could be enhanced with generative software, but as you say that would require effort.
Who would of thought companies would use ethically dubious machine learning technology to attempt to extract more money from users in an extremely lazy way?
When I saw this, I was expecting them using text prompts to create fully 3d modeled possibly animated models, but this.... no. I would love a tool that can do like, "create a female cowgirl character with a large cowboy hat and spurs, include animations for whipping, running, and dancing." and create a 3d model like it, even if it is low poly.
Yeah, we're a long way away from that. I'd say over 5 years, if it comes at all. Even 2d image generation technology isn't nearly there yet, 3d is lagging far behind that.
That would require effort on their part, and they only have very few competent people who ACTUALLY work. They are simply understaffed when it comes to developers and software designers. They do not even create the base figures by themselves, they usually hire a freelancer for it. Daz has been stale for well over 10 years now and unless they hire some proper development team that brings in a wave of fresh air, they will slowly die off...
I mostly use Daz for drawing reference. I've tried feeding its output into AI to produce a realistic "photo" from the 3D render, and I have to say the results are quite impressive (but uneven). My concern is not that it will get better than human artists, but that nobody cares if it's better, so long as it's cheaper and faster. We've already been flooded with horrible AI imagery where the user did not even make the minimal effort to fix mistakes. And they're trying to monetize this. And for all I know they're succeeding.
I'm a Daz user and member at the site. Nothing wrong with the program, it's the company that is way out of touch. It won't do any good going there to find out what people think, cause the Mod's are deleting negative comments (something the site is famous for). I can tell you the majority are not happy and some vendors are considering leaving Daz, some already have, some are even giving up on modelling all together cause they're just so tired of the BS.
I figured they would have where the ai could reference the 3d models look and pose so you could get more coherent results from the ai. Kinda like a last step to stylize it. but didn't expect them to do nothing and plop some generic ai slop. But nevertheless I'm happy theirs push back I'm really sick of this ai nonsense.
afaik they have an store where people can sell content for their program, have they trained their IA stuff with thirdy party content as well? if so, then they WILL lose their core business.
As someone with the "Return RU-vid Dislikes" addon, I can confirm that as of writing this comment, that video has 137 likes and 320 dislikes. It's not 100% accurate obviously, but it is one of the world's most popular addon so the data should be reasonably accurate.
Yeah of course it got messed up. These guys have NO IDEA what to do. Their history of releasing new ideas is less than stellar. The potential is always there but they are mostly too lazy to release stuff with the proper polish and QA. As if they just sit around a campfire throwing ideas around, then something sticks but only one guy gets put on the task while everyone else is busy talking about it. Then they release it and nothing works, or just some things work and then they let it simmer in this state and abandon any progress on it without ever talking about it ever again :D
I use DAZ as realistic figure reference. Unless you jump thru hoops and exert a LOT of time and effort, it's VERY hard to get clothing to behave naturally. Here, at least, they have done a good job of making images look good and natural. If the new AI feature could take CUSTOM figures and pose them as needed, it would be a godsend. I don't think that's on the horizon, however. DAZ wants to promote their pre-configured characters, that you HAVE TO BUY. Currently, it all seems VERY clunky.
As a long time DAZ user, I initially thought the idea of training AI on 3D assets definitely sounded better than all the stuff that currently just scrapes copyrighted content... But this ... Definitely not worth it.
The problem with that of course is that they are just using the same Stability AI data laundered through training with their own material, so it doesn't even do anything to really address the ethical concerns.
They have an AI generator but their rig doesn't even have IK controls in their own software. You still cant even export hair that you create in the software. But hey atleast we are given an AI image generator. Their daztoBlender bridge is completely busted and half their export options straight up dont work but who cares right? atleast we have an image generator...
I'm not a DAZ user, but I did check it out a while ago... The thing that I really liked is the ability to pose your character exactly as you picture it in your mind. And this removes all that control, and now you have to pray for something usable. Apart from that, what's the incentive to not just use stable diffusion? People use those models because with enough work, they can look realistic... Stable diffusion can do that by itself. This might be the dumbest thing I've seen in a while.
The funny part is that there was literally no way for this to work out for the. Daz3d based modle. Already exists on Civitai. Use it to pose. Blender, Cascadeur and Comfy do it better. If they wanted to go down this route they should have used it as a SD3 tech demo at least that would have been somewhat useful.
I wish I wasn't a Daz user, but I spent too much on their shop. I am ashamed to even know what it is. I'm surprised you know what it is because I'm surprised anyone knows what it is.
yup. None of the decent Daz users wants this. some vendors are even closing up shop at Daz, and now Daz is fishing in the Renderosity pond to attract new vendors.
all the bandwagoning companies seem to be operating on the assumption that they have a magic secret formula that puts them ahead of the competition....by putting in the same AI slop as everyone else. a thing being impressive is not the same as a thing being useful. anything that's actually professionally valuable like upscaling will probably work fine no matter whose algo you use, and will have open source alternatives everywhere pretty quickly. there are definitely applications for LLMs but natural language prompts targeting non-artists is a complete nonstarter both artistically and economically. that's without even getting into the IP rights nightmare.
I understand that AI art has developed a rather poor reputation, but even by the standards of AI art this is utterly abysmal. Beyond the fact that I don't know why anybody would want to generate what amount to generic stock photos of 3D chatacters, the results are significantly worse than anything that's been released for Stable Diffusion in the past 2 years (likely due to the incredibly limited data set), the generation time is insanely slow, there is almost no fine tuning, and it is rather expensive for what you're getting compared to it's competitors.
This software has so much potential with their models for animation, instead of working on that, they are releasing some garbage quick buck grabbing addons. On the other hand 3rd party software's like Reallusion getting the market on quick animated game ready characters. I guess Daz3d will never learn from this and instead going to stay as a lazy company that relies on still image rendering and let their model uses on steam porn games. Disappointed.
07:55 Reminds me of that docu where mr Grumpy Anime Grandpah himself was shown a demo where AI was used to animate ragdoll humans. "Well... Every morning... not recent days, but I see my friend who has a disability... It's so hard for him just to do a high five, his arm with stiff muscle reaching out to my hand. Now, thinking of him, I can’t watch this stuff and find [it] interesting. Whoever creates this stuff has no idea what pain is whatsoever. I am utterly disgusted. If you really want to make creepy stuff, you can go ahead and do it. I would never wish to incorporate this technology into my work at all. I strongly feel that this is an insult to life itself.” - Hayao Miyazaki
Not really an AI supporter per say... but honestly...I feel like you can make it work for you with a little thinking outside the box.. It can help with render block. You could use it for background creations or ideas. etc. I spend a lot on content at the Daz store every month and dont see that going away for me personally just because AI came about.. But yea, I feel like a person can get the best of both worlds if they allow themselves to.
I saw someone in a Discord server mess with using Daz renders through an AI re-render and I thought it was a good way to make TTRPG art for a book you would publish. You still need a good eye for it though or it comes out a bit uncanny because of the lighting or the clothing looking too much like a Daz model.
Daz have a great ability of misreading the room. Look at the early nft threads. There are so many things Daz users want them to work on, AI generated images is not one of them. Native linux support would be good, as would a render engine that can use amd gpus.unless you want to wait hours for a cpu render you have to use Nvidia.
So cursed 7:58. Edit: Not sure if anyone mentioned it. But you only had one image selected, so it only generated one. It seems to be under Prompt Strictness.
I always like prototyping reverse engineered software and tools, of course, I can only do it legally, aka a clean room… For a while now I always wanted to do the same for Daz, but I always didn’t want to do it cause I thought they weren’t a big company and it would be kinda scummy This video made me change my mind lol Now to only figure out a way to reverse it legally… only issue is that all models are based on genesis… a model… how can I recreate it without braking laws… I’ll figure it out, if I found a solution for Unreal Engine I’m sure I’ll figure out some loop hole… or maybe I’ll make an original character creator myself lol. Not trying to be mean but makehuman isn’t the most realistic 😂
2:58 I'm watching this on phone at 480p. Even before you mentioned it, I already noticed the hands are very problematic even on a small screen and low res 🤣 Anyways, what's the point of this app? It seems to me that they're killing their users now that people can just generate without having to pose, set materials etc with this new service.
There is some merit to this idea. Having the same character in your game and in your AI art seem useful. The issue is that the results are so bad. Also, Steam does not allow games to be published on their store that use AI art. Who is going to use this?
I maintain that AI generation is a party trick and not actually useful for creating content with real world requirements. I do think there is room for tooling that uses it to modify/work with the artist as they create, smarter editing tools, but everyone with more money or greed than sense is falling for the party trick and chasing an unattainable goal.
How about AI to help you pose the models you already have? Or to stylize your model’s hair? Or generate a skin texture? All of these so you can modify your own models before you render. Or how about using AI to help set up your lights? That’s how you use AI, not this raw image generation BS that’s already been made and is already old.
Every creative person absolutely loves idea of taking away whole fun of creating something and replacing it with "generate" button. Right? RIGHT? :D Some really clueless people are in head of these companies.
I was initially completely negative about AI, but I've found one valid use case for it. It is legit a great way to easily identify people who are lazy and/or have no taste, so they can be permanently frozen out of all collaboration.
So like... this is one of the first generative AI where you can morally generate artwork. If it's ONLY trained off of Daz's art and being licensed out, then yeah it's genuinely an AI product tackling the biggest issue of AI, that being the source images being used and copyright. However, it runs into the issue of WHY most models use copyright art work on a massive scale; without it the AI can't properly learn what things are. So I can genuinely give praise to Daz where it's due, I think the people attacking it are being outright idiots for only seeing flaws, but I can't say it's quite ready for use, nor is it the best product to make. A company based on providing tools for artists should make something which can help aid an artist, not replace them.
All they had to do is make a suite of tools making it easier to work with SD. Control net stuff would’ve been epic! Instead, I’ll just have to make a product…
Daz3D must hire better AI prompt engineers in order to utilize and connect the best of both worlds, Daz3D and the AI. They must not confine themeselves into the previous "just free" and just give only free level technology mentality or they will cease to exist in the near future as the AI development is very fast in developing this technology.
You know, for just about 2 hours worth of work and research you can have something significantly better than this on your home computer. Like I can get an AI image generator up and running with 6 gigs of vram (Less vram just means slower). All you need is Python, git, comfy ui, 2 checkpoint models (one to make the average the other to fix it || unless you're using SDXL), and a more details LoRA and with a midrange computer you can make something significantly better is there anything daz can put out. Now I know some of you are thinking this is a lazy cash grab that Daz did and you're right, I just wanted you to know how lazy.
DAZ should get it together like Iclone and Character Creator. They have better models and tech than Reallusion does. Why can't they become the prememium character creator software?
it's at 134 likes and 248 dislikes...you really need to use firefox and an addon to appreciate the ratios that a lot of these videos have....hey @youtube, hiding how disliked something does not make it likeable :)
1:48 Just checked and at least according to Return to Dislike to RU-vid its at 135 likes to 329 dislikes. About what you might expect, perhaps even a little better than for them than what you expect. Haha.
3:46 One reason I stop supporting Daz after ten thousands of $ of purchase. Almost occultic mindset in my humble opinion and if they don't like your post, they make excuse to delete it.
And the products are NOT game engine friendly out of the box. Will have to tweak heavily, like decimating, texture atlas(ing), etc. Don't even think about using them for mobile development.
So automatic1111 does it better as it is a tool made to make merged AIs, train them but also use txt2img, img2img and accept whatever extensions you want it to. For DAZ it looks like one of the models on civtai you can get for free but the DAZ version is was as you cannot adjust it for you in the prompts and other AI stuff Now onto the txt-2-3D AIs well I guess that is what DAZ went with not txt2img. txt-2-3D is a newer thing with more bugs and really their brand is they allow for 3D modelers to model in their software having an AI do it for you gets rid of your job which for management is most likely what they want as less people to pay but for people who use DAZ3D is not what they want as DAZ3D is already not the industry standard. Maya and other Autodesk stuff is. So i guess DAZ is looking for a way to get in where Autodesk has not as Adobe tried that already provided the tool and made it part of the product they already are paying for instead of a separate one. From what i can tell almost no one uses it. I hope DAZ will run into what Adobe did and their consumers will just not buy it. For Adobe they baked it into other products so the consumers had to get it just almost none of them use it. For DAZ making it another product well they are competing with free. Just the free version came before and works better then theirs right now. The 6 fingers problem is already solved, the objects in the wrong location problem is also already solved, DAZ is very behind in the AI game and should stick with what they are already doing instead of pissing off who pays for their software.
Garbage in, garbage out. You feed the AI generic, pre-made, out-of-the-box assets in generic, pre-made, out-of-the-box poses, rendered at mediocre, stuck-in-2010-CGI quality with default pre-made, out-of-the-box lights and materials, you'll get mediocre, generic looking images... but with AI jank. Big L.
Yeah its a bit less than beta to be charging for this yet and it may not even be ready to be released for free in its present state. Maybe they should hire someone with a clue to plan how this is to be released rather than the sales drones. The problem is that Daz is rather late to the AI game, all the groundbreaking features they are touting have been released by everyone else already.
I know this has been said a million times, but AI needs to be bullied out of anything creative when its not directly helping it. like I could imaging how insulting it is that you're using AI in that matter to market to people who know what they're doing. you're a tool developer first, act like it, stop using AI as a marketing buzzword, consult with the pros that are using your software, and adjust, if something really needs AI, then use it for its purpose. you know how cool it would be if AI could help me figure out where I should put my linework in like very obscure angels? maybe help me adjust to certain composition standards? allow me to name layers accordingly faster so I can focus on the important bits?, maybe notify me if the perspective on some parts are off? because believe it or not even if you thumbnail and get enough references, if working enough on something complex enough, you'll lose the spark of what made you wanna draw that piece to begin with, ending in it being duller than what you wouldve imagined in your mind. and allow me to turn it off when I am done with the sketch and basic line work. because soon enough I would need to do so. instead its making generic, dull, wax figure, softbrushed images. circumventing the entire process, ruining what art is to begin with.
Man, I haven't been as critical of AI art as some. I see stable diffusion as almost its own art style (even with VAEs and etc) the way people usually use it. So all of that is fine. But this product extension is so lame that it broke my normal "meh and move on" threshold lol. This is a product for people with an actual goal that has constraints. I know how to use ComfyUI and UNet too. This is like someone said "we need an AI product out yesterday." Watch though, in a year we'll be eating our words :p
