Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Yes, I agree. Wendy productions are innovative, well made and inspiring.
All the best for you, @WendyLuvsCatz
Agreed Wendy is very creative !
nah, I just reuse other's stuff be it 3D assets or AI
I used to draw when I was younger but I wouldn't call my scribbles art
but that is what this thread is supposed to be about, a mash up
This started as a 3d render.
Latey I have been using a service called voice.ai to narrate some of my videos.
Everthing in the 3 minute clip is AI generated except the female speaker who was animated
in MOHO 13.5
mixed media video
Fascinating. It's almost like a movie camera was set to record each frame as it passed thorugh an infinite number of similar dimensions. The music is a perfect fit, all in all fun to sit back and relax while watching.
-- Walt Sterdan
That final zoom in, with cats materializing while approaching the infinite limit is eerily creepy.
well it is a batch img2img conversion of the 2 DAZ renders that follow it with the same prompt and seed
ones after are Deforum using just the prompt
This are some tests I did in SD, and while the realism goes up exponentialy, I think I need more tests before I could use it for actual production.
Has the question of whether or not it's okay under the new EULA terms to use our Daz-made renders forAI remixing been addressed anywhere? I did check that other thread but found nothing helpful and it's long since closed. I've a few things in mind to work with as remixes -- on my own computer, not an online service like MidJourney, etc. -- but very hesitant to use anything until I have a better idea what Daz policy is.
Once upon a time, I'd have no worries about turning stuff from my old renders into brushes and tubes for personal or even commercial use in Photoshop or Paintshop Pro. These days I'm not sure doing that won't get me into trouble, especially if the new image I'll be using said brush/PSP tube on was fully or partly generated using an AI.
photo of my late cat (forum avatar ) with a selfie video using Runway
@eheugalde
Those look great hopefully more coherant processing of animation frames will be possible.
it is also possible to animate the face of your Daz character still renders with AI
Fascinating additions since I last visited. Great stuff, all contributors. Wendy, you are an inspiration!
Super impressive.
-- Walt Sterdan
Started as a Daz render
Gorgeous.
-- Walt Sterdan
rigging AI images as blimps using Transfer Utility
Interesting (and cute); where do the blimps come in? I'm obviously missing something regarding using the AI images with Transfer Utility.
-- Walt Sterdan
I should have explained but there ae a few ways to create them, you need the background removed
this one is a cutout with volume created using a Carrara plugin by Philemo
I have also used Alpha to Mesh in Zbrush and masking and extract under subtools to make them
basically any modelling program that can cut out a silhouette/billboard and give it some volume like a cushion
the Carrara one the UV matches the image
initially so does the Zbrush ones but I end up Zremeshing them and then creating a planar UV and projecting the image on using Polypaint and Spotlight
or you can do that in a UV mapping program as the shape of the island still the same if planar it just needs scaling if image not square
Ah, got it, thanks! Carrara blimps! Thanks for taking the time to explain, much appreciated.
-- Walt Sterdan
This one also started as a Daz render/
I really, really love the "illustrated" look you've got here. I find it far more beautiful than anything simply photorealistic.
-- Walt Sterdan
+1
Wendy, your skill set is amazing!
the blimps too weird so back to DAZ characters but with one AI background and AI generated soundtrack
Cool. What do you use to create the AI-generated soundtrack?
https://www.songr.ai/
it is actually in the description
an AI version (images batch processed in Stable Diffusion and interpolated)
I know I'm late to this thread and that it's spawned its share of heartburn on the topic of AI, but I wanted to acknowledge the many wonderful examples of its use here as a tool in a workflow, and to share a few of my early tests. For the longest time, I have been plagued by my inability to push my 3D renders over that frustrating, though increasingly permeable, line into reality. I'm hardly the first to do this, but using a locally installed instance of Stable Diffusion as a sort of "post-production filter" in my workflow, I've wound up with a result that's at least somewhat aesthetically pleasing and promising to me. Starting out with DAZ renders of a morph-tweaked iteration of Krashwerks' Anuschka, I'm using an SD model called Paragon (in this instance), one of several that are very good at photorealism. Having found a balance between the Denoising and CFG factors in SD that respect both my text prompts and the input image (the raw 3d render), the net result is a hybrid of the original 3D render and the AI output that allows me to preserve the integrity of hands, feet, clothing, props, and some other features in postwork. Anuschka's face, as seen in the examples, is the AI's interpretation of her using just a breath of SD's Face Restore. Since the same person can look slightly different in the real world depending upon lighting and other factors, I find the facial fidelity across the output images to be satisfactorily believable, overcoming one of my biggest AI frustrations, next to hands and feet. Again, my goal is not to make a viewer commit to a perception that my images are photographs of real people, but only to cause them to believe it long enough for them to ask themselves whether the subjects might be real, and ultimately to allow them to fiind some beauty in the content. If I can do that, then - for now - my efforts are successful.
NOTE: One huge benefit to AI post-processing I've noticed is that the AI tends to identify and correct several notorious shortcomings of 3D rendering, such as telltale 3D facial characteristics, unconvincing interactions between soft bodyparts, as well as unrealistic collisions between soft bodyparts and solid or soft surfaces.
EDIT: I seem unable to upload images at the moment, so I'll come back to this.
Yeah, I've been trying for the past couple of days over at the "NPR" thread to try and get images uploaded. No go. I was able to post some images but that's only because they had been previously upload on the forum about a year ago, so was able to grab those images from a previous post. But for some reason, the forum won't let us post any new images. It's a real bummer and hopefully it's just a glitch that will be worked out soon.