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
© 2026 Daz Productions Inc. All Rights Reserved.
Comments
Neat, thanks for sharing! I'm only now starting to look at installing ComfyUI locally, I have no idea how that's going to go, but the work you're doing makes it awfully tempting...
If this is to me I am quite flattered considering what a huge fan I am of your work!
In my video entitled Using Daz Studio with ComfyUI 2026A the very beginning shows a glimpse of a tour of my workflow.
All of those colored nodes all over the place is a modular animation studio that I've created. They're Note nodes containing blocks of prompt text for various portions of my modular template that I use as my positive and negative prompt texts. Some are for identity-locking, some for facial performance, some overall actions... and I have them for several of Rosie's personas as well.
I'm really quite a newbie when it comes to ComfyUI - spending all of my time (currently a month - now going back to Studio) developing this modular system within a single workflow.
Now I'll be using results from ComfyUI that I really like to go back into Studio and make more new custom facial dials for the idea of using these ComfyUI performances to drive better work for my Iray animated renders.
It was meant to be to you, my apologies for not pointing to you directly; sometimes I forget that while I *feel* like I'm chatting with friends over a table in a café, I'm *actually* shouting over a crowd across a large hall at a convention.
My goal is to see how closely I can mimic my experience using online front-ends like OpenArt.ai and Elevenlabs for animation with a local ComfyUI installation and one of available engines like LTX-video.
Last week, I dipped my toe in attempting an animation of a one-page comic I'd done earlier this month and was impressed. A day ago, I spent about three hours testing lip sync for the very first time in AI and I was blown away. Totally.
Over the course of that three hours I did totally newbie tests using almost random voice clips I generated to fit a "story" I hadn't actually worked out, so it was all done on the fly (not the best way to do, but I was only testing).
I took 90 seconds of the tests and plugged them into iMovie and made this:
If clicking on the image doesn't bring up the animation, you can use this link to the animation:
https://sterdan.com/wp-content/uploads/2026/04/Captain-Knight-on-Death-Station-Lip-Sync-Test-01.mp4
Now, there's a couple of glitches and it's far from perfect, but I was still blown away for a first attempt. This was made from three of my stills (DAZ's TheGirl, Sal.A.Manda and CheetahGirl, all tooned and modified, below), four audio clips I generated on the fly and four prompts (that, looking back, I could have done much better).
The only thing stopping me from really ramping up on animating again is cost; my current, pre-paid subscriptions would probaby allow me to do about 10-15 minutes of 720p lip-synced animation a month.
I know that my M4 Pro laptop would be much slower locally, and that while an RTX 5090 could do an animation in one or two minutes that might take my Mac 10-20 minutes, my electricity cost for running the M4 Pro full-bore, 24/7 for a full month where I live would cost me about three to four dollars, quite affordable.
I don't mind longer rendering times, especially for a lip-synced animation, but my main question is about how my experience working with a local ComfyUI and, say, an LTZ-Video installation compare to doing the same thing with OpenArt's or Elevenlabs' versions?
I mean, online, I chose my engine, dropped my image in, dropped my audio file in, wrote a prompt and a few minutes later I had a video. Could it be set up to be so simple locally (albeit with much longer render times), or is that a pipedream at this point?
Naughty, eh?!!!
Bravo, sir... Bravo!!!
Thanks, much appreciated.
RE: "naughty" – Well, the Talus corporation has long been suspected of illegal bio-weapon research, and that space station has *something* loose running amok...
As I said, I was impressed at how stupidly easy it was to make it, starting with – essentially – three unrelated, previously-made renders. I'm excited to think about what could be done when someone approaches it with an actual plot. The three characters used were toonified versions of the original Girl, Sal. A. Manda and Cheetah Girl and various DAZ clothing items; three older (and much older) DAZ characters from my library, there are so very many more charactes and assets waiting to be brought to life, each with their own stories.
Now if I can figure out how to drop the online services to open up the floodgates...
that is pretty impressive
Thanks very much, but you can see it's more a tribute to the tech and software available to us now than to any real talent on my part. Now that I can see what's there, I do plan on taking the time to do a real animation with an actual story.
I glanced briefly at ComfyUI, *that's* gonna take some time and probably more brainpower or patience than I have...
Yeah, I tried a couple of online services but didn't like what they did to my IP character Rosie. So my pal, Tugpsx, turned me onto ComfyUI Portable. Just recently Copilot told me that the full install of ComfyUI is better and still doesn't use the internet unless we ask it to.
I made a custom LoRA of Rosie using 256 renders, letting each image train for 100 steps - so it was 25,600 steps.
I've learned quite a bit since then so my next custom LoRA will be better, but this one does a pretty good job at keeping Rosie looking like Rosie
Agreed 100%!!! :)
The part where she's walking and starts walking backwards is one of the issues where I prefer to keep many animations rendered from Daz Studio, where I have a lot more control and flexibility.
I believe that these anomalies come from the AI model attempting to get back to the original source image - but I'm not entirely sure. Like I said, I'm still a noob at generative AI
When it comes to Rosie's Cyberpunk Street Warrior persona, I've all but given up on using generative AI for her animations due to incosistencies of what it does with her gear from one shot to the next. Most of her other personas behave pretty well though.
I love what the AI does with pouring Life into her - so that's where I'm using the AI results to guide me on timeline work using my custom dials. I'll also be trying my hand with Studio Premier's Geometry Sculptor to make special shaping morphs for things like expressive suprasternal notch (base of throat by collarbone) movements, for example, along with very specific custom expressive shapes.
I just assumed she was backing up as the Xenomorph changed direction
HA! That's what *I* thought!
I'd actually thought I'd add either little "ping" sounds when she slowed down, stopped or sped up that would signify instructions from the Professor, leaning into the error
My thoughts on it are that Kling Avatar 2.0 "runs out of ideas" after 5 seconds of generation. It's starting with that single frame to build the not-so-endless corridor on and after five seconds of creation, it backs up a bit because it "ran out of room". You can see when she stops moving forward that the continuous corridor wall changes to a jutting joint and what looks to be a cross-corridor to the left. I used another starting clip with a different character and it hits the same spot in the same way as far as the walking or running goes.
This happens again, in the second Freya clip: at the 5 second spot, she stops and starts walking back.
Different engines are interpreting things differently – no surprise – and I'm both impressed in some and disappointed in others.
I ran a few more quick tests with a different voice clip and it was interesting how different some of the results were from what I expected.
I kept the audio down to 5 seconds, with the simple prompt of either the "The character is walking warily while she speaks" or "The character is running while she speaks".
Kling Avatar 2.0 did this (you can see that by stopping at the 5 second mark she doesn't pause or back up):
https://sterdan.com/wp-content/uploads/2026/04/OA-Kling-Test-Signs-of-Life.mp4
The default "OpenArt Lipsync" has her walking, but it *really* doesn't want to imagine what the unseen parts of the corridor look like so it does very litlte background movement and concentrates on the character instead:
https://sterdan.com/wp-content/uploads/2026/04/OA-OA-Lipsync.mp4
And finally, a more interesting test; I animated it in Seedance 1.5 Pro, which doesn't do lip synching. It generated a sweet animation with sound effects and a sound track, which I then added lip synching to with Video Sync Lipsync 2.0 Pro. The face is mildly distorted while she runs with the lip synching (it wasn't before) but I think it's good enough for what I want.
https://sterdan.com/wp-content/uploads/2026/04/Freya-Test-02-Lips-and-explosions.mp4
Wow! you are all going great case projects with varing solutions. Love it. @wsterdan, Kinda like the cheekiness of that short piece.
You're going after a much higher-qualtiy product than I am, of course. I have an easier road with simple, cartoonish characters that allow for a litlte less rigour... okay, a lot less rigour. Your stuff is much more mind-boggling, and I'm eager to see how much more you keep doing with Rosie. It's hard to imagine more life being breatherd into her. Totally awesome stuff.
For the AI animation stuff I'm just now wading into, I think that after the last couple of days of testing I have a solid idea of what I might do with it in the near future. A lot of the way I work tends to find where the limitations are and instead of spending too much time trying to find workarounds to do what I want, whenever possible I redesign the project instead to fit the limitations. In this instance, for example, one of the limitations seems to be animations more than five seconds, so I'll plan on either keeping the clips under five seconds or split a longer squence in two so that the final frame of the first animation becomes the starting frame for the second part. We'll see if it works, or if I wind up with a lot of jarring hiccups. I think I'm getting enough character continuity between the different engines I'm using that I should be able to mix and match a bit, too, using different engines for their strengths. We'll see just how delusional I am.
Of course, one of the other "issues" is that any workflow I figure out now might be obsolete next month when the engines I'm using today will be updated by then, and some of the limitations I'm hitting wil be gone.
Finally, I downloaded ComfyUI, but I thought I'd test it first with their cloud version. I think it might be easier for me to spend another $15 a month and keep using what I'm using. Yes, I'm *that* lazy and *that* cowardly. I'm not proud, but "I ams what I am".
Thanks, it's a lot of fun to play with this stuff. I'm really looking forward to tucking into it with an actual plan and see what happens.
@wsterdan, I can have you up and running in no time since you have done the hard part and downloaded ComfyUI portable. The whole setup should have you up and running in under 10 minutes.
Thanks! I've PM'd you.
I remain unsure that there is any advantage to dealing with the complexity of comfyUI for AI animation as I can create and save characters, sets and props, and reuse/animate them with consistency
just as I used to do with 3D content
I use Mage.space to set up my scenes and animate them with grok Imagine from Elon Musk’s X.AI
Here is the “content library” of my saved character props and locations I am using fro my current graphic novel
I already have 59 complete pages after two weeks of production.
No complexity at all. All of those nodes that we see all over the place is my own doing - a modular prompt block animation suite specifically designed for Rosie and her personas.
What are you using?
This is great. I've gotten some crazy results with LTX-2 myself.
For what it's worth, I truly don't feel that the workflows I've been using in ComfyUI could come close to replacing what I can achieve directly in Daz Studio - especially in giving me what I want as a final output.
It's still a great tool, don't get me wrong. Many of the motions, particularly the close-ups at the very beginning, were created using ComfyUI:
In my experiments I've seen many limitations to animating in this way, but also a baziliion possibilities.
The time taken for ComfyUI to spit out many versions of the same inputs is quite fast, making for a fairly fluid workflow.
When it comes to doing an actual production, however, I prefer to use the AI results as a form of previsualization data that I can take back into Daz Studio and continue performing these shots by hand along the timeline. In some (perhaps many?) cases, it may also be prudent to include some of the AI results into an edit of production, since they do turn out looking very much like my Iray results. So I'll continue to hone my skills with these AI models that I've been using successfully, but without aborting what I really love to do - sit in the Daz Studio Director's Chair and work with my actors personally.
I definitely like the results that I get from my hand-key-framed eye gestures, motions, blinks, and other involuntary actions over that of the AI, while the AI has also definitely taught me more about involuntary motions for other parts of the actor that I can now strive to achieve myself - which, to me is a Huge Win!
There are also times when I prefer what I can get from dForce over what I can get from AI, but that is also true in the reverse. The magic of being a Premier member authors me the Geometry Sculptor, which allows me to edit dForce simulations using custom morphs created on the fly that can be turned off, on, as well as exceeding limits in either direction, which gives me similar results to what I often dreamed of using if I ever got my hands on Chrono-Sculpt, which never happened. All I can say is... it's Magic!!!
I have to add to all of this one very important fact that drives much of what was just said:
I never came into "Animation" looking for automagic solutions. I came into Animation because I love doing it!
…
An excellent point. I think it comes down to it being another tool in the box that different people will use in different ways. I totally get the idea of using it for previsual work that you can use as an aid to keyframing your 3D. Ironically, I was able to use the first Poser beta before release, it being a digital mannikin tool to help artists previsualize their art. It's a huge circle, but it does keep going around and around...
I didn't have a lot of time to "play" today, but in the spirit of previsualization, I thought I'd see what some of my characters would look like in a more simplistic "Saturday Morning Cartoon" style would look like, so I spent about an hour redoing the crew from my first test in such a style. I'm going to go back to my initial style, but this was fun to knock it out quickly, just to see what it would look like:
If clicking on the image doesn't open the animation, here's a link to the video:
http://sterdan.com/wp-content/uploads/2026/04/Freya-Knight-Death-Station-Simple-Style-01.mp4
Walt, that's one amazing Test Run!!! I always love your visual style and attention to each character having their very own expressive silhouette. I think this was even More obvious and enjoyable in those earlier Hrimfaxi episodes. I always told my friends how impressive it is that, if all we could see were facial movements, each character would be immediately recognizable - which is total directorial GOLD!!!
I love the new Toon look and always look forward to more adventures!!! :)
Thanks for the feedback, it's very much appreciated, especially from you.
The second thing I was testing (besides a simpler style) was a way to use the AI lipsync/animation tools ot get around what I think is my biggest issue. I have a very difficult time lipsyncing a character when there are multiple characters onscreen. Despite trying to get the AI to focus on lipsyncing a single character with very specific details (I mean, how hard is it to tell AI to lip sync the human female and not the lizard?) I rarely get a single person "talking". I either get multiple people talking a the same time, or the lip sync being applied to people in turns, first one, first the other... for the same single voice.
I need to look at masking out "extra" characters while lip syncing, or some other method.
Thanks again!
Vace Multitalk witn Wan is supposed to be good for defining which character out of 2 says what
you do a prompt for left and another one for the right character
but not for me
So hopefully before 2027 I can have an all new adveture episode to reveal - but I won't rush it.
I've been working towards this stuff since 2010, but back then I was chiseling stone by hand 13-16 hours a day in the hot summer sun - only working on CG in winter - and even then I did snow removal. So it wouldn't be at all fair to say that I've "been at it" for 15 years. Much of that time was learning how to optimize materials and geometry for faster renders.
Now I'm mostly working on things that drive me wild from the likes of Lightstorm/WETA Digital, ILM, Blur... etc., with the understanding that I am only me, and I'm using Daz Studio. So my goals are never to completely replicate what they do, but to pay close attention to words from folks like Richard Baneham, who really inspire me to make Rosie 8 Great!!!
These are the sorts of things that I watch as my own, personal enjoyment - except that I buy the actual digital deluxe versions so that I can watch all of it
I always accidentally get caught up in explaining myself - I just can't help it. So I apologize. But then I duck back out and let the conversation continue on without all of this... okay? Peachy!
Something that I view as an important part of utilizing Daz Studio and Daz 3D - even though I am aware that there are Many other options:
Rosie is my IP Character. Bam. She's not born from an image of someone that I stole from the internet, she wasn't a selection of random images generated from an AI prompt... she's mine. I sculpted her using Daz 3D.
This, to me, is a very important aspect of this whole craft - designing our own digital actors to portray what we want shown on the screen. Sure, I get criticized for not telling stories. But that's not where I am yet. I'm still honing stuff. Important (to me) stuff. Wendy knows that all of these things that I do are story-driven - I have plenty of that brewing as I animate Rosie. Those actual story moments just happen to disappear in my fun musical edits in most cases - simply because my videos aren't yet meant to tell stories. But they will! And in the meantime, they're all still My Characters!
With the invention of my Motion Capture Editing techniques that I've developed in my Dynamic Character Animation course, I have found Daz Studio to be my hands-down favorite means of animating - Period.
While this may not be for everyone, it's truly erasing the misunderstandings created over decades that we cannot animate in Daz Studio. Quite the opposite - we can have an easier time directing our digital actors in Daz Studio than nearly any other system!
Unlike a professional animation studio,we get to animate our 'finished product' actor right from the start - and render it out when we're done playing with it. This is incredibly different than what we'd be learning to do for most other systems. We buy from Daz 3D - it Works in Daz Studio. Easy!
I am heartbroken that Reissormocap is no longer part of the Daz 3D store - not sure what happened there, but I'm glad for me that I had already collected the entire Reissormocap store. Whew!!! But I feel for those who had not. They were highly original and very well made - and they came as individual poses from throughout the animations, animated pose files, as well as aniBlocks - my preferred method.
Being able to render using Iray and/or Octane and/or Filament in any resolution we want and any format... we are basically given the entire flow within one piece of software except for the final stage movie editing process - and I'm glad that they don't spend time including that into Studio since DaVinci Resolve already exists - Hollywood-class production for free? Um... Yeah!!!
Having Daz 3D and their PAs as our art department is Magic! It's Powerful! It's also Incredibly Affordable! When I was being urged to try another popular 3D animation solution from another company, I had determined that, even after spending $1,000.00 USD I would still be left incomplete for making someone like Rosie and being able to place her in the environments I do. And that would just be my startup!
Daz 3D is Much different than that. I've seen several criticisms about the company and, while some folks might be able to twist a solid point against them, in every case I've seen they don't continue and show all of the benefits that nullify their gripes. Daz Studio comes with such a plethora to get started - before ever spending anything. And then, spending tid bits here and there on specific needs can take you to producing a cool video of a story you want to tell - just like that!
Powerful, Friendly, Cost-Effective, and loaded with personal choice.
Next month or so I'll be delivering a Creative Cart webinar that discusses many of these things and much more - because I truly believe in the power of each artist having their own, unique character that drives them to make cool art - now matter whether it's animated or not.
Agreed. Lip sync has been an issue for me as well.
When I animate using Daz Studio, I can animate each character on their own in their own, otherwise empty scene - then bring them together for the render if I do it that way. My AI doesn't like rendering a character to alpha, which is a huge downfall.
That's one of those areas where I'm using the AI as Pre-Viz.
For very specific source images I can get excellent lip syncing. So I can then use that to drive me edits to an Anilip 2 result - or even sculpt the lip sync animation myself, by hand. Because what I'll be building as custom dials are special phoneme enhancements to make speach look a lot more real - or stylized - and in Studio, I can achieve both - in theory. Actual results are yet to be seen because I haven't actually started that process yet.
Anyone interested in hearing a master at animating believability as a keynote speaker - ladies and gentlemen, Mr. Richard Baneham speaking to a room full of Animation Students: