Hi,

Is Poser going to evolve with AI ?

But that would be perfect, creating an image with Poser and have a photo realistic render.

That would be a dream come true.

If this subject was already discussed... Sorry !

Pierre

Not sure when the dev team may get to this, but there is already an script add-on in the marketplace to do this.
There's a nice video in the promo images explaining how it works.

https://www.renderosity.com/marketplace/products/161157/openpose-for-poser-12

Thank you but I see that you must have Stable Diffusion installed. Not for amateurs !

Please,

Stable Diffusion is all about amateurs.

Maybe but I understand nothing about it especially to install it on my M3 mac.

I think AI could be of use for Poser in Poses. Poses are vectorized information, and there is plenty of it. It would only need a qualification that could link it to terms in the prompt. Problem, as ever with AI, is ownership of data.

As far as image enhancement is concerned I think it is better if Poser stays true to it's roots: provide template imagery/geometry of posed/animated figures and leave the final workout to other apps dedicated to that task. There is no way a small company like Bondware could ever stay par with development in the sector. 

In another thread on this subject I showed the below AI enhanced image:

Guess the Poser figures used.

Little black dress one is clearly LF2 - I can recognize even every asset used, as I made the original LOL - I'm just concerned about what happened to the poor girl's hands XD

That being said, the moment AI (in its current model of machine learning from images without proper permission) is included within Poser is the moment I'll drop Poser for good.

Correct about the dress one. But who is the beauty on the left?

Posette for sure. 

Ooops ... one Poser generation off. I just looked for the clothes. It's P2 Casual Woman. WOW!  I NEVER would have guessed that! 

Sorry but we are moving away from the question

Sorry but we are moving away from the question

Okay but it’s just a script to use with stable diffusion. Not an integrated solution to help us create with Poser. 

The official answer was "not sure when the dev team may get to this". Meaning that, IF it's ever considered, it's not priority.

He was offering you an alternative.

Okay okay but what you don't seem to know is that I know nothing about Stable Diffusion. It's hard to install especially on a M3 Mac like I have and I'm not sure that I could consider Stable Diffusion one day. So The answer is simple to me it's no The Poser team have no intention to integrate AI in their software which is nevertheless a pretty good idea in my opinion. They could boost their software a great deal.

As an artist with much better artist friends, I don't particularly want the integration of a development that's resulting in artists not getting work or laid off because ' AI does it for free'. Also, as an artist, i don't want the integration of a development that was trained on stolen art from artists.  I know the other software did it. Imo it was a mistake the same way NFTs were. We will see.

Because they borrow the model from someone else. Here you will create it yourself. I just want AI to give real texture to scenes. Nothing else. You don’t seems to understand what I envision. I would like to give my Poser man or woman a real photographic look and the same for nature. Turn your Poser scene into a real picture. But you have created the scene. 

We just have different visions I suppose. 

I suppose that you know of the face swap feature, you upload a picture and you want to replace the face of someone on the picture. You also upload the replacement face. There’s nothing stolen from anybody. If Poser can give life to our dead characters nothing need to be stolen to anybody. 

You cannot afford to risk mixing software like Poser with Stable Diffusion unless you own both. If later on Stable Diffusion is outlawed or it is shut down for any reason, Poser would be left useless.

FaceApp ended their face swap feature because of legal issues. Their rating dropped to the point where they don't have much left before they have to close up shop. Lot of AI apps are dropping like flies.

I think what you are looking for is post-processing. Take your Poser render and go to a A.I. website and ask them to improve the render. They'll charge you for it, but you got your amazing textures and everything.

I've used AI to generate render images, but the results are not what I expected. The images look similar to the typical AI-generated images that are all over the internet. The outcome can vary a lot without extensive text input from you. There's only so much data available to train the AI render engines. In about 5 years, the AI might run out of new external data to learn from and start using its existing knowledge, which could lead to data corruption.

They are ai site that still use face swap. I just wanted to make a point about fixing some scenes without having to steal someone’s art. I didn’t realise that they were some much people against it. I sense a lot of aggressive feelings towards me because I dare to pronounce AI with POSER. But I know that if one day there will be a ai feature well implemented in Poser to make your work more beautiful you will be all happy. 

As an artist with much better artist friends, I don't particularly want the integration of a development that's resulting in artists not getting work or laid off because ' AI does it for free'. Also, as an artist, i don't want the integration of a development that was trained on stolen art from artists.  I know the other software did it. Imo it was a mistake the same way NFTs were. We will see.

Because they borrow the model from someone else. Here you will create it yourself. I just want AI to give real texture to scenes. Nothing else. You don’t seems to understand what I envision. I would like to give my Poser man or woman a real photographic look and the same for nature. Turn your Poser scene into a real picture. But you have created the scene. 

Adobe trained their AI with the consent of artists as I read. So it can be done the right way. 

what i ask of the AI ​​is to mix the qualities of the firefly rendering with the superfly rendering data

what i ask of the AI ​​is to mix the qualities of the firefly rendering with the superfly rendering data

No one's being aggressive here, focus on the topic and try to explain what you had in mind when you said incorporate AI in Poser because I'm clueless, I know what AI can and cannot do.

What is it that AI can do that you cannot already do in Poser?

AI works at a 2D level, creates a mediocre low quality 2D image, with a lot of artifacts and errors, with a questionable origin database, database made entirely of images (2D).

Poser is a 3D program, using a database of crafted 3d models (assets) in combination with high quality images (textures), located on your computer, it's called Runtime.

What can AI do for Poser? 1) create a low quality texture 2) create a low quality background 3) skip all the 3D stuff entirely and create a final AI image. See, it's option number 3 that most Poser users don't agree with, because we paid for Poser, we paid for 3D assets, we paid for high quality textures and backgrounds, and step number 3 skips Poser entirely, you don't need Poser to create an image with AI. There's no 3D in AI. AI cannot help Poser with anything, but it sure can lower the quality standard of a render. Let that sink in.

vince5 posted at 11:35 AM Fri, 27 September 2024 - #4489795

what i ask of the AI ​​is to mix the qualities of the firefly rendering with the superfly rendering data

How so? They use two entirely different methods of rendering. FireFly "fakes" reflections. SuperFly's are realistic raytraced reflections..

Crystalis posted at 2:26 PM Fri, 27 September 2024 - #4489800

DeeceyArt posted at 11:54 AM Fri, 27 September 2024 - #4489796

vince5 posted at 11:35 AM Fri, 27 September 2024 - #4489795

what i ask of the AI ​​is to mix the qualities of the firefly rendering with the superfly rendering data

How so? They use two entirely different methods of rendering. FireFly "fakes" reflections. SuperFly's are realistic raytraced reflections..

What he meant is he wants the definition and clarity of Firefly style render but without the Superfly style noise, fuzziness. Reflections are raytraced starting with Poser 5. Poser 4 was the last one to use the fake reflection maps.

As far as Firefly reflections, I stand corrected. My bad! I haven't used FireFly in so long that I forgot 8-)

You had to "occasionally" show us your 800 dollar video card lol anyways, what I was trying to say is that his demands can only be achieved in post-process, after an image is generated. That's all AI can do, take an image and make something else out of it.

How do you do a face swap in Poser: you change the morph and the texture of the character, it's easy as that. And regarding the photorealistic looking renders, don't we all strive to achieve that? You explain to him that's only possible in postwork and he gets mad at us for explaining how things work in a 3D program. Go ask AI if AI can be integrated in Poser lol Ask AI which websites it used to harvest the images collected too

There are 3d AI creators. 3d AI Studio and Meshy are 2 such programs. I can't speak to quality since I've never used them. I'm just pointing out they exist and AI is not limited to 2d.

pierremeu posted at 9:03 AM Fri, 27 September 2024 - #4489791

Adobe trained their AI with the consent of artists as I read. So it can be done the right way. 

Nobody actually consented. It was just a "from now on if you're using our program you consent, but also we're not gonna warn you about that change - oh, we're gonna hide is as an option just to avoid legal trouble, but the option is on unless you know to look ant turn it off".

Go look, thousands of artists have been leaving Adobe lately because of those bad practices.

RedPhantom posted at 4:12 PM Fri, 27 September 2024 - #4489803

There are 3d AI creators. 3d AI Studio and Meshy are 2 such programs. I can't speak to quality since I've never used them. I'm just pointing out they exist and AI is not limited to 2d.

There's a program that allows you to scan an object from different angles and creates a 3d model based from the capture images. You can't exactly call that AI.

What these two websites claim is just not possible. How you gonna generate the back part of the 3D character when you only input the image of the front part? Use common sense. Their demos are reversed... they used a 3D model to create the image, then they show you the image turning into a 3D model. It's a scam website which is trying to obtain your credit card information. Keep that in mind with any website AI or not, that demands your cc info.

Crystalis posted at 5:07 PM Fri, 27 September 2024 - #4489807

Thank you. I just tried to use them, they both demand payment before doing anything 3D, they won't talk to you unless you put a credit card in. I don't trust any of them. When I tried to access the Meshy website the laptop fans started racing and the adblocker just went crazy.

There's a program that allows you to scan an object from different angles and creates a 3d model based from the capture images. You can't exactly call that AI.

What these two websites claim is just not possible. How you gonna generate the back part of the 3D character when you only input the image of the front part? Use common sense. Their demos are reversed... they used a 3D model to create the image, then they show you the image turning into a 3D model. It's a scam website which is trying to obtain your credit card information. Keep that in mind with any website AI or not, that demands your cc info.

Re: Superfly renders: if you use the correct render settings they are anything but fuzzy. It took me a while but with the help of lovely people here I think I at least have the render settings down . Everything else is still in development, but that's the way art goes.

You had to "occasionally" show us your 800 dollar video card lol 

LOL not intentional. I have since retired and there's no way I can afford that kind of hardware anymore. 

I was able to get Meshy to work for free without any credit card. It's nothing I'd use regularly. The UV map is a mess. But here's what I get in blender.

For those of you who do want to integrate 3D models with AI, I saw a video on YouTube earlier this evening made by Olivio Sarikas that shows you a website where you can do just that and very easily. 

https://www.youtube.com/watch?v=ERgZ1xh54ys&t=563s

The website is called CGDream. He gets into the details in the video.

pierremeu posted at 9:03 AM Fri, 27 September 2024 - #4489791

Adobe trained their AI with the consent of artists as I read. So it can be done the right way. 

Adobe hid their "consent" as a non-advertised, no-warning-before-adding, tiny option that was on by default.

Nobody actually consented. It was just a "from now on if you're using our program you consent, but also we're not gonna warn you about that change - oh, we're gonna hide is as an option just to avoid legal trouble, but the option is on unless you know to look ant turn it off".

Go look, thousands of artists have been leaving Adobe lately because of those bad practices.

So many companies have done this too. So shady of them opting users in, and then they find out later after the fact.  Even LinkedIn did this (what do they even need it for there?). 

If I were to guess, I'd wager they got paid by AI-pushing companies to do that. Guh, I wasn't aware of LinkedIn doing this too.