Thank you

Again; nope. Vague crap definition by whom? Somebody on TicTok? Wasn't me so don't try and put words in my mouth.

"But most people would not be able understanding the signification of this text ! same like placing a mathematical Formula to explain the algorithms"

Huh? the signification of this text. Really? Reminds me of the Homer Simpson quote. "smrt"

Diffusion models are built by repeatedly adding noise to an image until the training image is beyond recognition. The reverse generates an image.

Example is a photo of a cat. The cat image has a predetermined amount of random noise added and the label ( the text part of the model) says it a cat. There is more noise added and the label is cat. More noise is added and the image is labeled cat. Soon there is only static noise in the training image and it is labeled cat. That is your "model". A set of rules for denoising an image. To get an image you type c a t and a random noise is generated and the denoising starts. The operator (you) determines the strength of the denoising logarithm and the number of iterations it makes in an attempt to create the request. The result, for better or worse; is "cat"

That's it in a simplified nutshell as diffusion models now stand. There is a massive legal mess happening around this method as we have all heard. Well the method is changing. GANs are being used to train now. General Adversarial Networks. This is closer to autonomous driving training than current diffusion models, but that is beside the point and a different conversation yet again.

I'm not responding any more to this. The subject is drifting off topic into personal attack which is against the TOS.

I have watched and held my breath. I only speak up because there is a lot of misinformation here. JBC you have no idea of which you speak. You have no idea how a diffusion model is produced or deployed. The waters get muddied by disinformation. Portraying yourself as knowledgeable in this field is laughable at best and quite a dis-service to those of us who are data scientists and imaginative artists.

"A good AI collects Information then it builds up a so called model, you can compare it with a mathematical function of your brain, but they call that file a Model witch is one of many learning algorithm's comparable to a .Json file Not a Image it self, the learning process with given Information generates a sort of Intelligent Filter, just like you use in Photoshop just that it is a learning filter. You can also compare it with a encyclopedia of Information."

What kind of BS is this? Really? Would you really like to know how a diffusion model is built? Not your way. An "A.I." doesn't "collect information" ;and exactly how you equate this to "many learning algorithm's (not a possessive btw) comparable to a .Json file." makes no sense what so ever.

We've had this conversation before. More than once. It has been shut down for various reasons each time. Usually the misinformation and fear of the future arguments win the day. I understand that. There are unanswered in the court of law arguments currently underway. They will fall on both sides of the fight as there is more than one way to build a diffusion argument and many companies are building their own using GANs and not the addition/subtraction of noise method currently in use by Stable Diffusion and others.

So stop with the BS and try reading a white paper or two. I have linked to them in the past and got in sh!t over it so I will not do so again. There are reasons that Ken and others have stepped out of these conversations and folk with knowledge like myself fear to tread.

Thank you. Smooth as glass.

HartyBart you may want to look at Kel1171's scripts for Poser 13 to find what you want. I know he had such a script ready to go to the marketplace, but there was a lot of AI hate running around here. It was submitted, but Rendo stomped on AI so I am not sure if it ever made it to the marketplace here. He may have submitted a version to DAZ as well. Haven't talked to him in a bit. In either circumstance, such a beast exists already and I know people using it so it's available. Try site mail here if nothing else.

Ken; I posted above about people using 3d software like Blender, Poser and Studio output as input for AI. Sounds like a useful tool you have there and a fine addition to my Poser 13. Let me know if it ever gets released.

There is one awful big irony here.  This week's RR newsletter features an AI GENERATED IMAGE as it's cover.

I suspect prolific output to be a big part of this. Freestuff is full of AI generated material, not just the galleries. Rendo rewarding for that could be part of the issue. They are trying to sell assets after all not give them away. Servers, storage, and bandwidth are not free.

I do believe that this is slightly ill-considered. I have no issue with AI. Just the way some of the folk use it. Low effort type and go may be fun for some, but it just becomes so much eyestrain after a bit. NVIATWAS anyone?

I've been a member here since Metacreations owned Poser, Kai made Powergoo; and Renderosity and Renderotica were the same site. I changed username some 15 years ago or so. I visit here every day. On rare occasions I post to the forms. I have never posted to the gallery here in all that time. Does that mean I'm not any good or afraid to show my work? Nope. I was a professional for 35 years until I retired. I do show work elsewhere as well as post images to galleries. Some of them even {gasp} have AI components.

Texture creation, background removal and replacement, lighting changes, extraneous detail inclusion or removal, not to mention masking for editing and so much more. Just ask Adobe or any of the others mentioned. "AI" is not just still image creation. AI is being used with Poser and DS as basic input for AI effects. Blender and Vue both have AI plugins. Photographs get the AI treatment. Are you going to forbid the use of AI "assistance" in those galleries? If so; how do you tell?

Invisible watermarking? It's there in the models themselves and no; stripping the metadata does not remove the watermark. It's embedded at pixel level so AI doesn't train AI. It can be removed but not easily as it is weight (model) level and not application level. It will appear even if a non-ai generated piece is manipulated thus you can actually make a Poser render, photograph, or any other image identify as AI even if it isn't. The point is that well done work wont appear as "AI" and until you figure out a reasonable method of detection. AI artwork is appearing in every gallery due to the tools we use. Time to hire termites on staff as the wooden shoes are falling everywhere.

Glad to have helped.

Are the morph packages in the same directory as the figure?

Did you run the .bat file after the installation of each package?

Both are necessary to get the older DAZ figures working, not just V4.

I broke down and tried to update my install. I must say that it went quite fine. I hope. This leads me to ask; hornet could this update possible trigger something in your anti-virus similar to the last time? Not sure how you would check.

I did not even realize there was an update. Poser is set to check for updates on launch. It never sees any. Manually checking via the in program dialogue also maintains my software is up to date. Obviously I am not. My version number is 296 and has been since installation. Guess I wait till I see the outcome here before I download the latest patch. I am pretty sure I allowed Poser through my firewall when the installation occurred as my purchases old and new show up in the library but now I am nervous about the updates that my installation never sees.

I rechecked and I was correct the first time. Bonnie the Construct is blocking your ez-dome. I started with an empty scene and everything worked correctly; 2d or 3d object as a catcher worked as expected. As I go to post this I see Y-Phil has posted a prettier image using the background node. Disk on the right is shadow catch only. Sorry about Ref_lite_Mult. checked. You'd think that would be off by default by now

Actually, with further experimentation maybe it's not working correctly. I can make a 2d object a shadow catcher; such as a cloth plane in Superfly, but not a 3d object such as the Hires Disk. Back to the experiments.