News about the DAZ Studio and Poser software

Category Archives: The Animation Industry

Success – Minimax H3 video at 1376 x 768px, on a 3060 12Gb card

Posted on by Posted in Companion software, Freebies, The Animation Industry, Tutorials Leave a comment

Hurrah, success with Minimax text-to-video on my RTX 3060 card! I can now generate video to Minimax’s 1.0 resolution. The sweet spot seems to be 12 minutes to get a seven second video at 0.9 resolution.

 

Windows 11, RTX 3060 12Gb with just 24Gb DDR3 system RAM (most people run 32Gb of DDR4). It’s amazing I can do large coherent videos on this, in a few minutes, on an old server. How far we’ve come since the days when Stable Diffusion 1.5 was mostly gloopy mis-fires at 512px, and took ages.

Minimax’s prompt adherence is now much better, now I’ve moved on from using a supercrushed Q2 GGUF for the prompt-processing. Next step is to see how well it works in combination with Poser / natural media emulation. But for now, a full setup tutorial.

TUTORIAL:

Here’s how to do it on this humble entry-level card, with the latest software, file-types and turbo boosters. The workflow is linked at the bottom of the post. All free, as is the way with generative AI.

First, install a new ComfyUI Portable, then use the Manager to update ComfyUI to the latest 0.32.x or higher (needed!). If the ComfyUI console throws a polars error from Python when starting up, then also pip install polars-lts-cpu (it’s polars for older CPUs, which lack the latest whizz-bang multimedia processing extensions).

Second, clear 40Gb of space on the SSD for the files you need, and also leave plenty of headroom for the Windows swap-file etc. I have my swap-file capped at 24Gb to match the system RAM.

Models: At HF get minimax_h3_fl2va_pruned-w4a8_convrot_pruned.safetensors (11.6Gb) and minimax_h3_ref2va_pruned-w4a8_convrot_pruned.safetensors (11.6Gb). The first does text-to-video and first frame or first frame/last frame. The second takes character / environment reference images (e.g. Poser character renders), and combines the character(s) and scenes into the video, if you have the prompting done correctly. Correct prompting is everything with Minimax H3. Put the files in ..\ComfyUI\models\diffusion_models\

Clip: At HF get qwen3vl_32b_minimax_h3-w4a8_convrot.safetensors (14.6Gb). Put the file in ..\models\text_encoders\

The above are Winnougan’s highly efficient ‘pruned’ blends of the FP4 and Int8 format, suitable for the 3060 12Gb card and matched with the very latest ComfyUI features.

VAE video: At HF get Kijai’s minimax_h3_video_vae_int8_convrot.safetensors (3.7Gb). This decodes the video frames into a video, and is far faster than the original. Place the file in ..\ComfyUI\models\vae\

VAE audio: At HF get minimax_h3_audio_vae_fp32.safetensors (577Mb). This is the standard from ComfyUI, and works so quickly it’s not worth speeding up. Place the file in ..\ComfyUI\models\vae\

Turbo LoRA: At HF get Abiray’s minimax_h3_turbo_4step_ckpt600_V4.safetensors turbo LoRA (590Mb) and place it in ..\ComfyUI\models\loras\

Having ComfyUI at version 0.32.0 or higher allows you to run the turbo video VAE and also to pair the turbo LoRA with their new ‘Sage Attention replacement’ speed-up called Comfy Kitchen Attention. Simply add their new ModelAttentionBackend node between the model and the LoRA, and toggle it to use ‘Kitchen Attention’. That’s it, though note it’s best for lower-end cards and that 50-series cards may not see much difference. I think I’m getting a 20-30% speed-up from it.

With this setup I get good results at 5 steps with the turbo LoRA at 1.2, Euler / beta. Image quality is still slightly crispy and granulated, but better than smushed and gloopy. The crispiness could be because I’m over-forcing the turbo LoRA, of course.

7 mins = 7 seconds, at 0.6 (1056 x 608px).
12 mins = 7 seconds, at 0.9 (1280 x 736px).
18 mins = 8 seconds, at 1.0 (1376 x 768px) (the model’s native trained resolution) and using 6 steps.

All 24fps at 16:9 widescreen ratio. 0.9 seems the sweet spot for me re: the balance of time / quality / size. That’s the video which heads this tutorial.

Workflow as a .JSON file for ComfyUI: MiniMax_H3_RTX3060-12Gb-workflow.ZIP (225Kb).

Minimax H3 on a 3060 12Gb card

Posted on by Posted in Companion software, Freebies, Spotted in the News, The Animation Industry Leave a comment

How times change. I’d resigned myself to never generating any worth-having generative video on a humble entry-level 3060 12Gb graphics card. Indeed, the card was starting to look ‘entry-level’ no longer. Then about a week ago came the open release of the incredible Minimax H3. Free, fully local, unrestricted, and fully supported on release day by the free ComfyUI. And it will run on my card.

It was worrying to continually read that one needed 32Gb of fast DDR4 system RAM on the PC. I only have 24Gb of DDR3 RAM.

But after a little tinkering I can now generate video on my humble 3060 12Gb VRAM card. 9 minutes to generate 8 seconds of 16:9 video (736 x 412px), with excellent prompt adherence. At 24 frames per second. GGUFs and a turbo LoRA mean 22Gb of files are required, all told, beyond the installation of a fresh ComfyUI Portable 0.30.2.

The largest video I can get with a quick GGUF workflow is a text-to-video 0.3 (736 x 412px) video output, generated using 8 steps with the turbo LoRA and Euler / beta for the settings. Generation times are similar when inputting a first-frame for the video to start with, and offering this in combination with a more basic text prompt.

The drawback of such a ‘smallest possible’ GGUF workflow is that the auto-matching sound (ambient + SFX + music) is not good, but I don’t much care about sound. But I’m surprised that small supercrushed GGUFs can even generate video without producing visual gloop and glitches. Just for the record here’s what the GGUF workflow runs, 22Gb in total…

Model: H3-FLVA-Pruned-Q3_K-M.gguf (8.2Gb).
VAE: Standard safetensors (5.4Mb for both together).
Clip: qwen3vl-32B-MiniMax-H3-Q2_K.gguf (7.9Gb).
Turbo LoRA: minimax_h3_turbo_v4_step600_ema_pruned_comfyui.safetensors with Euler / beta, 8 steps (600mb).

After this initial success of the first basic “can it even work” test, the next steps are to: clear space on the hard-drive; work out which is the best much-larger model and clip file in the mysterious new Int8 ConvRot format (the replacement for the GGUF format); and then keep a look for better turbo LoRAs (new ones are being cooked as we speak). After that, have a look at the more sophisticated references-to-video model (also free and open). It seems the latter can do a “makeover” of video input, suggesting obvious uses for enhancing Poser and DAZ animation videos.

So, I hope this initial success may encourage those with a similar old PC / card, and only 24Gb of older RAM. It can be done.

Oh, and the model doesn’t know H.P. Lovecraft, other than that (on a good day) he should look something like this…

This screenshot gives you a flavour of the sort of “crunched” quality at 0.2 (lowest, 608 x 352px) and with just 6 steps on the Turbo LoRA. A bit granulated, but at 24 frames per second it’s acceptable for drafts or for pushing through Klein 4B for a comic-book panel re-style. Obviously 0.3 and 8 steps produces a larger and better video. I can’t do 0.4, as 0.3 is the limit on this quick lashed-together GGUF workflow. Update: Now I can, with a new VAE video decoder, new turbo LoRA and a few tweaked settings.

Release: Stable Audio 3, with ComfyUI checkpoints and encoder

Posted on by Posted in Companion software, Freebies, Spotted in the News, The Animation Industry 1 Comment

Stable Audio 3 has been released. Free, offline and local. Feed it a text prompt and it generates field-recordings, movie-style sound-effects, mixed-effect soundscapes… and now also music. It’s of obvious use for those creating animations, visual novels, motion comics, YouTube illustrated audiobooks, Ken Burns style documentaries etc.

Very fast, small, and with outputs free for commercial use. Full 44.1 kHz stereo quality. Even the very small models for version 3 (2.2Gb + 1.1Gb text encoder) can produce up to two minutes and can apparently generate this in a few seconds even on a CPU.

Very importantly for creatives, the new version offers iterative editing of outputs. You can regenerate just the bits you don’t like (audio ‘inpainting’) and keep the rest. And the speed should make that a relatively easy matter. You can also auto-extend your audio in the same style (audio ‘outpainting’).

Trained on legally clean sources, so the anti-AI mob can’t wail about ‘stealing’.

The official release is gated behind a Huggingface sign-in, presumably to guard against vexatious lawsuits re: misuse by miscreants. But the ComfyUI models and encoder are here and freely available: ../Comfy-Org/stable-audio-3. I’m still downloading, but I guess it might then need an update of ComfyUI, re: getting the text encoder recognised? Update: needs to be the latest ComfyUI 0.22 or higher.

What’s currently missing is i) an example prompt structure to generate a complex but coherent sequential soundscape (can JSON be used, with timings?); ii) a working example of how the iterative editing is done; and iii) an optimized ComfyUI ‘studio’ workflow (e.g. for optimised stereo separation and movement, and for running the Small SFX with the Small Music model alongside each other for a basic multitrack in the workflow).

For the previous Stable Audio, note that one could multitrack simply via the prompt. e.g. “A balanced mix between a good field recording of a man walking through dry leaves in winter, and a recording of small birds calling plaintively in the surrounding Canadian boreal forest.” The word mixdown would also work. I assume this will work on the new version 3, once I get it installed and working.

I’m less interested in the music than in the SFX and soundscapes, but if the music interests you then note the new Stable Audio Prompt Guide for Music at their site.

Preview: iClone AI Studio

Posted on by Posted in Companion software, Spotted in the News, The Animation Industry Leave a comment

iClone is now promoting its AI Studio generative-video rendering feature as ‘coming-soon’, a feature with was trailed last month in a 2026 roadmap post. Looks interesting, if they really have cracked character/scene consistency and temporal stability. But it’s Cloud-based video generation, and presumably credits-based, so you’re likely to need a big pot of cash to enjoy it.

Contest: $3.5m Future Vision XPrize film-making contest

Posted on by Posted in Competitions, Spotted in the News, The Animation Industry Leave a comment

The Future Vision XPrize is now live, a $3.5 million film-making contest. Submit a three minute trailer for a film that tells a story about an optimistic future. Specifically, show how good things will become real in a technologically-advanced future that is happening because of us (not being done ‘to us’). Animation and AI tools can be used. Deadline: 15th August 2026.

2,600 Carnegie Mellon .BVH files for Poser, as animated .GIF previews

Posted on by Posted in Companion software, Freebies, Poser, Spotted in the News, The Animation Industry 1 Comment

I see that Shriinivas has an animated directory for the 2,600 Carnegie Mellon .BVH motion capture files. Released November 2023, you get little munchkins in .GIF animations which show each motion-capture.

The .BVH files are linked with a named link alongside each animation, but note that these don’t lead to the Poser versions. For Poser you want the cmu-ecstasy-motion-bvh-poser-friendly-2012 freebie archive. But the names in each set are the same. Thus Shriinivas’s “90_09” means that in the Poser files you need to look in folder “90” and there find the file “90_09.bvh”.

Scroll down his page to see the link to sub-pages for named ‘Animation Categories’, e.g.

Instructions for loading .BVH to a Poser figure, here. Sadly there’s still no drag-and-drop of .BVH in Poser.

Qwen3 TTS – install and test in ComfyUI

Posted on by Posted in Companion software, Freebies, Spotted in the News, The Animation Industry, Tutorials 2 Comments

Qwen3 TSS has been released, and it allows local ‘prompt to custom character’ voices. This adds a whole new dimension to local text-to-speech (TTS). It’s also a pleasingly small model at around 5Gb total (if you already have many TTS Python requirements), so is very feasible for those with older graphics cards and slower Internet connections. It has an Apache 2.0 license, so is fully open-source and available for commercial use. All the below requirements are free, as is the way with local AI.

As you can see, you can describe your exact voice and the audio generated conforms to the description. Voices can be described with great detail, far more than shown above, and their modulation over time also (e.g. “rising excitement”). There are obvious uses here for unusual character voices for animation, games, audio drama, vocal additions to audio soundscapes, etc.

Tested and working, after a lot of work. Here’s how to manually install for ComfyUI portable:

1. In ..\ComfyUI\models\ create the new local folders ..\ComfyUI\models\qwen-tts\Qwen3-TTS-12Hz-1.7B-VoiceDesign\ and its subfolder ..\speech_tokenizer\

2. Download the required models Hugging Space at Qwen3-TTS-12Hz-1.7B-VoiceDesign and speech_tokenizer.

Put the downloaded files into their locally pre-prepared folder and sub-folder.

3. Now get FlybirdXX’s ComfyUI-Qwen-TTS custom nodes to run these models. Windows Start button, CMD, cd into the ComfyUI custom nodes directory, then…

git clone https://github.com/flybirdxx/ComfyUI-Qwen-TTS

4. Install the requirements for the new custom nodes. Start, CMD, cd to the ComfyUI embedded Python directory, then…

C:\ComfyUI_portable\python_standalone\python.exe -s -m pip install -r C:\ComfyUI_portable\ComfyUI\custom_nodes\ComfyUI-Qwen-TTS\requirements.txt

(Replace ComfyUI_portable with whatever your local path is).

There should be no conflicts, as yesterday’s patch for these custom nodes fixed the official Qwen TTS demanding transformers==4.57.3 which could have killed Nunchaku (which requires a lower version).

5. These Custom Nodes require a download of SoX which is an .EXE installer. Sox is a venerable freeware sound-exchange code library, kind of like ImageMagik… but for sound. After install you must add it to your Windows PATH. Thanks to Promethean Dante for the fix here…

Looking at the node code it seems SOX is only needed if you try to generate on CPU rather than GPU, but the lack of it prevents the nodes from loading in ComfyUI. It seems you need both the Python sox module installed (it installed along with the requirements.txt – see above), and its Windows framework via the .EXE installer.

6. Start ComfyUI, and set up a simple workflow thus with the new nodes…

Time: 70 seconds for a five second clip, on a 3060 12Gb card. Reasonable, not super-turbo but workable.

The basic requirements of Qwen3 TTS are compatible with a ComfyUI portable install — Python 3.8 or higher, PyTorch 2.0 or higher, so the above custom node set won’t bjork your PyTorch by trying to upgrade it. Beware others similar custom nodes for Qwen3 TTS in ComfyUI that will try to upgrade Pytorch to 2.9 (not good, for a portable Comfy).

Vital: Prompting and multi-voice guide for VoiceDesign.

Contest: Internet Archive’s Public Domain Film Contest 2026

Posted on by Posted in Competitions, Spotted in the News, The Animation Industry Leave a comment

The copyright release season approaches. The Internet Archive is running a contest for creative short films that use public domain material, especially the releases due on 1st January 2026.

Make a 2-3 minute short film, and add an equally open soundtrack. No ban on the use of AI, which is especially notable since good-quality local AI Video2Video generation has become accessible/affordable in the last year. But I’d image that the original footage is what the judges will be looking out for. The 1930 date suggests obvious linkages with early pulp science-fiction. Deadline: 7th January 2026.

Here’s a survey of what’s entering the public domain in 2026, with a focus on fantasy, science-fiction and horror.

Stable Audio 1.0 in ComfyUI

Posted on by Posted in Companion software, Freebies, The Animation Industry, Tutorials 2 Comments

Stable Audio 1.0 is a free “prompt to sound-fx” generator, based on an ingestion of the well-tagged sound FX files at the huge public-domain Freesound website. As such it produces royalty-free sound FX clips, of up to 47 seconds in length. Here it is, working in ComfyUI portable.

Workflow:

1. Copy the 4.7Gb model.safetensors from the Archive.org standalone’s ..\Stable-audio-open-1.0-webui-portable-win\stable-audio-tools\models folder. The standalone’s .torrent file is the easiest and most hassle-free / re-startable way to get this huge file. Then you put it in ../ComfyUI/models/checkpoints and after that you rename it as stable_audio_open_1.0.safetensors

2. Download the 800Mb Google T5 encoder model.safetensors file from HuggingFace, rename it t5_base.safetensors and copy that into ../ComfyUI/models/text_encoders/

That’s all you need. Just set up the workflow as seen above, and you’re ready to generate. All the other guff in the Archive.org Stable Audio portable is now taken care of by ComfyUI. ComfyUI audio generation also feels faster than the standalone.

In the above workflow, setting “batch” higher than “1” seems not to work.

I found it is possible to multitrack/mix more than one FX, via the prompt rather than nodes…

A balanced mix between a good field recording of a man walking through dry leaves in winter, and a recording of small birds calling plaintively in the surrounding Canadian boreal forest.

Update: This simple batch workflow works for batching different prompts and/or getting multiples ‘takes’ of the same prompt…

Local Microsoft Azure AI voices

Posted on by Posted in Automation, Companion software, Freebies, The Animation Industry Leave a comment

The original Msty (1.9.2, not the flaky new Studio version) is a fine free desktop host for running local offline LLMs (‘AIs’). But it has no offline text-to-speech. Your AIs can’t talk, unless you get an API key and are always online.

One offline solution is the freeware Simple TTS Reader 2.0, which reads whatever gets sent to the Windows clipboard. Very very simple, it just does the job — anything copied to the clipboard gets read aloud. Great, but sadly it can only use Microsoft Speech voices, which are rather robotic and limited. It appears Microsoft has moved on from these, to its new and far better Azure TTS voices.

However, there’s a hack to get these Azure voices locally. There’s a handy NaturalVoiceSAPIAdapter from GitHub, with a straightforward Windows installer. This freeware makes Microsoft’s Azure natural (aka ‘neural’) TTS voices accessible locally to some of the SAPI5 compatible TTS desktop software packages. On Windows 11, these more advanced voices are otherwise locked to the Windows Narrator for local use, and no other software can use them (booo…). But now Simple TTS Reader can use them too.

As well as NaturalVoiceSAPIAdapter you also need to get your target Azure voice from this selection of free voice downloads. Unzip it as directed and put it somewhere sensible. When you install NaturalVoiceSAPIAdapter, you need to tell the software where the voice is on your PC, e.g. C:\TTSVoices\ Remember: Un-zip, don’t install in the usual installer manner.

After that, restart Simple TTS Reader 2.0 and you have a good local Azure TTS voice for automatically reading whatever is sent to the clipboard. Now when you hit ‘Copy to Clipboard’ at the end of a Msty LLM response, the text will be read in a reasonably good AI voice.

Regrettably Msty can’t automatically ‘Copy to Clipboard’ at the end of each LLM response. It has to be done manually, by clicking the icon. Ideally, Msty would add a “copy each new sentence to the clipboard, on completion” option.

Having Azure voices locally on your PC may also interest animators who’d like to have such quality voices without going online. Of course, there are also dedicated TTS AIs now… but they can be very fiddly to set up, require many Gbs of downloads and disk space, and also a good graphics card to run them. The above fast Windows 11 solution requires a mere 80Mb in total and no graphics card.

The Balabolka TTS freeware can also see the Azure local voices, but note that it’s 32-bit software and thus you will need to install both 32-bit and 64-bit versions of NaturalVoiceSAPIAdapter.

Update: the voices appeared to have totally stopped working. But it is possible to get them working again. I got them working by first re-installing both 32-bit and 64-bit versions (Balabolka is old 32-bit software) and the Ryan and Sonia voice packs, then selecting: Windows Start menu search | Speech | Speech Properties | Text to Speech | Setting Ryan to the PC’s default voice | Apply and exit.

Update: They support pause tags in Balabolka…

… and also pitch and speed shift.

iClone goes full AI, with robust ComfyUI integration

Posted on by Posted in Companion software, Real-time animation, Spotted in the News, The Animation Industry 1 Comment

Reallusion’s iClone embraces AI image rendering, with a new official ComfyUI integration. It looks like they’re going all-in for AI, regardless of the AI doom-moaners. Good for them.

We’re excited to launch the Open Beta of AI Render plugin, a powerful and completely free tool that bridges real-time 3D animation with AI-powered rendering, seamlessly integrated into the ComfyUI workflow. … AI Render uses custom nodes that connect Reallusion’s 3D environments directly into the ComfyUI ecosystem.

Now in Open Beta. Officially supports only Stable Diffusion 1.5 and WAN 1.3b video, though node-wranglers may find information about other types of AI image-gen on the Reallusion forums. Likely the new Wan 2.2 relatively lightweight 5B model will be of most interest there.

Seems to have a standard approach. The Comfy nodes appear to take Depth, Pose, Normals and Edge preprocessor data simultaneously from iClone’s real-time viewport, also marrying them with one-click style presets (it’s an IP-adapter) for refinement of output. The two comics presets are manga style, so there’s no western comics style as yet.

Interestingly also… “AI Video Generation models optimized for consistent frame-by-frame results”

How consistent? That would be the worry there. Still images are one thing but, even with four Controlnets working at once, are we still going to get slightly “wobbly” faces? One would also have to worry about shifting colours.

Microsoft’s New Ray Tracing AI – now in ComfyUI

Posted on by Posted in 3D Utilities, Companion software, Poser, Spotted in the News, The Animation Industry Leave a comment

Life moves fast in AI-land. Last month I blogged here about Microsoft’s New Ray Tracing AI. This month — courtesy of Paul Hansen of Germany — Microsoft’s new tech is now free in ComfyUI. Along with an outstanding install guide and documentation. All free. Currently, 2 seconds of finished raytraced animation takes 22 seconds on a 4060 card. Import of .FBX is coming soon.

Microsoft’s New Ray Tracing AI, now at 16 fps

Posted on by Posted in Poser, Spotted in the News, The Animation Industry 1 Comment

Microsoft’s New Ray Tracing AI (YouTube Video, six minutes with hardcoded ad at the end). They ingested 16 million ray traced 3D images, to make an AI that simply infers (from its past knowledge) what the play of real light in a 3D scene should be. Then the AI applies it and ‘renders’ the scene in a microsecond. You can tweak materials, and it updates instantly.

Animated? Yup, their pseudo-raytracing currently clocks in at 16 frames per second (on MS’s research labs hardware, admittedly). Quite respectable, and there are also frame-interpolation AIs out there that might boost it to 30FPS.

Physics? Yup, they even added that too. Even dynamic water.

Generative AI overlay? Not yet. But if this gets a general open-source release and isn’t locked away as an exclusive for Microsoft Flight Simulator, someone will add generative AI imaging to the mix. Imagine not only AI raytracing, but AI raytracing + a layer of SD ‘style change’ based off the 3D scene (but still faithful to it). The ‘Hollywood-real look’ for your 3D scene, in near real-time and beautifully lit.

What a time to be alive. Indeed, what a time to have a huge Poser runtime. Poser has such massive possibilities ahead, if only it can shrug off the AI-haters. The devs don’t even have to develop for it, I would imagine. Just open up some general hooks in Python, to let users hook into whatever local AI they choose to run.

Release: IK Studio for After Effects

Posted on by Posted in Companion software, Spotted in the News, The Animation Industry Leave a comment

New to me, Richard Rosenman’s $50 IK Studio – Inverse kinematics plugin for After Effects (February 2025). Apply IK to the joints of your 2D cutout cartoon characters. He also has a survey of The Best Rigging Tools for After Effects for 2D characters and props. Probably far easier to do such animation in Moho or Cartoon Animator. But if you have to use After Effects, this appears to fill a gap.

Release: Pencil Pro, for Blender

Posted on by Posted in Companion software, The Animation Industry Leave a comment

I’m all for natural media emulation from 3D scenes. So I was pleased to see the new $15 Pencil Pro add-on for Blender, allowing your 3D scene to emulate a pencil drawing. The result is obviously not great at long distance (e.g. a city scene), where it looks like a point-cloud with the points mapped to tiny graphite strokes. Better for medium and close animation shots, though there’s still a distinct Rhubarb & Custard-style wobble. Which is charmingly old-school in its way, and would be acceptable to young kids watching shorts.

Suitably rough and sketchy and believable, if you overlook the polygonal angles from the 3D. No per-frame autocolour, but something that auto-colours greyscale (e.g. Akvis Coloriage AI) might give colourisation that is not too wobbly, when run frame by frame.