[Official] 🎉 Welcome to AI Render Open Beta – Start Here!

[jason_RL]

Hi all,

Welcome aboard! AI Render is a powerful free plugin that connects real-time 3D tools like Character Creator and iClone to AI-powered rendering through the advanced ComfyUI workflow.

This is where AI generation meets 3D precision. Whether you’re an experienced ComfyUI creator or just starting with AI rendering, this forum is your go-to resource hub for tips, workflows, and community support.

---

Installation & Setup Guide

Environment Requirements

Operating System: Windows 10 or above

Graphics Card: NVIDIA 30 series 8GB (minimum), AMD GPUs are not supported

Program Version:

• Character Creator 5, Character Creator 4.6 or above
• iClone 8.6 or above

Full versions only — trial versions are not supported

Get full version here

Recommended Storage Space: 80 GB

ComfyUI & Models

• The AI Render installer will also set up ComfyUI automatically for you — no separate installation required.

• Default models included: Stable Diffusion 1.5 for image generation, Wan2.1 Fun 1.3B Control, and Wan2.1 VACE 1.3B for video generation.

• You don’t need to install all models in advance. Instead, simply download the ones required for the style presets you want to use from the AI Render Settings, ComfyUI tab.

Download the Plugin

Character Creator 5:
https://cdn4.reallusion.com/AI_Render_Beta_0901_for_CC5_Enu_Full.zip

Character Creator 4.6:
https://cdn4.reallusion.com/AI_Render_Beta_0901_for_CC4_Enu_Full.zip

iClone 8.6:
https://cdn4.reallusion.com/AI_Render_Beta_0901_for_IC8_Enu_Full.zip

Check out the latest version

---

Quick Start — Install Plugin + Pre-built ComfyUI

If you’re new to ComfyUI or just want to start creating with ease, download the plugin and the Reallusion-built ComfyUI package.

• Includes ready-to-use Style Presets (2D, 3D, and photorealistic looks).

• No need to set up complex workflows — just plug in and start rendering.

• Presets can all run locally as long as your system meets the environment requirements.

Learn more about the style presets and how to fine-tune in:

[Official] Fully Customizable 2D, 3D, and Photorealistic Rendering Styles

AI Render Installation Guide for iC & CC | iClone & CC Tutorial

Advanced Setup — Integrate with Your Own ComfyUI

If you already have ComfyUI installed and are familiar with workflows, you can connect the plugin to your existing environment. This lets you:

• Achieve pro-level character consistency and precision for multi-frame AI videos.
• Make the plugin compatible with your existing ComfyUI workflows, so you can continue working with setups you already use.
• Build custom workflows and experiment with high-end models (e.g., Flux, WAN FusionX).

Learn more about how to achieve production level controllability and quality in:

[Official] Consistent Characters and Precise Control for AI Films and Commercial Production

[Official] Maximize AI Render Performance: Custom Workflow Setup & Tips

AI Render Custom Workflow with FLUX | iClone & CC Tutorial

Scaling Up with Cloud-GPU

Running heavy workflows with large models can be demanding on local hardware. To break through these limitations, you can leverage Cloud-GPU solutions:

• We’re providing RunPod & RunComfy templates.
• Skip complex local setup and access powerful GPUs instantly.
• Great for rendering advanced workflows that require high VRAM.

Learn more about how to set up cloud GPU and get templates in:

[Official] Cloud-GPU Templates for AI Render — Powered by RunComfy & RunPod

---

Estimated Render Time

Based on default models: Stable Diffusion 1.5 for image generation, and Wan2.1 Fun 1.3B Control and Wan2.1 VACE 1.3B for video generation.

• Image (512×512): ~5–10 seconds per render
• Video (512×512, 5 seconds): ~5–10 minutes per render

These estimates do not include model loading time.
Actual performance may vary depending on your system specs (GPU, CPU, RAM) and the complexity of your scene, prompt, and style settings.

Performance Tips

• For GPUs with less than 16 GB VRAM, we recommend:

• Use 512×512 or 832×480 resolution for video.

• Keep video length under 5 seconds for best stability and speed. (If the project frame rate is 60 FPS, 5 seconds equates to 300 frames.)

• You can upscale videos later after previewing, to balance speed and quality.

---

Navigating the Forum – How Official Threads Work

All official threads are labeled with [Official] in the title. These threads are monitored by the Reallusion team and contain the most accurate, up-to-date information.

Product Features:

[Official] Fully Customizable 2D, 3D, and Photorealistic Rendering Styles
Explore unique style presets and customize AI outputs.

[Official] Consistent Characters and Precise Control for AI Films and Commercial Production
Learn advanced workflows for multi-frame AI videos and character consistency.

[Official] Maximize AI Render Performance: Custom Workflow Setup & Tips
Get the best results by optimizing performance for your system.cl3

[Official] Cloud GPU Templates for AI Render – Powered by RunComfy & RunPod
Access cloud GPUs for heavy workflows without local hardware limits.

Educational Resources:

[Official] First AI Render Webinar on Aug 22 – Register Now!
Learn how to set up Flux 1 Dev on cloud GPU and connect it with iClone for production-level AI image generation with precise 3D guidance.

[Official] Second AI Render Webinar on Aug 29 - Register NOW!
Master the full LoRA training workflow—from preparing CC/iClone characters to training with FluxGym on RunPod—for consistent character results in AI Render.

[Official] Third AI Render Webinar on Sep 5 - Register NOW!
Explore WAN video workflows in iClone, combining 3D animation control with advanced video-to-video rendering for dynamic, cinematic results.

Version Update:

[Official] AI Render Plugin Update — Coming September 1
Announcement of the upcoming AI Render plugin update

[Official] AI Render Plugin 0901 Version Release Note

Events:

[Official] AI Makeover Community Challenge – Free CC5 Awaits!
Showcase your creations and participate in the community challenge.

Get Help:

[Official] Bug Report & Technical Support
Central hub for troubleshooting and bug reporting.

---

Community Guidelines

Be Respectful

• Treat all members with courtesy and professionalism.
• Respect different skill levels, backgrounds, and creative styles.
• No hate speech, harassment, or personal attacks.

Stay On Topic

• Post in the correct forum threads (e.g., bug reports, style guides, challenge entries).
• Keep replies relevant to the thread topic to help others find information easily.

Share Constructively

• Offer helpful, respectful feedback when commenting on others’ work.
• Ask questions, share tips, and support fellow creators — we’re all learning together!

Report Bugs Responsibly

• Use the official [Bug Report & Technical Support] thread.
• Include clear details, screenshots, and system specs where possible.

Please avoid posting offensive, inappropriate, or infringing content.
Reallusion is not responsible for any user-submitted content that violates legal, ethical, or community standards.

Get the Full Version to Use AI Render

AI Render requires the Full Version of iClone 8 or Character Creator 4.

Buy Now
[Purchase iClone 8 & Character Creator 5]

Try Before You Buy — 30-Day Free Trial:
[iClone 8 Trial Download]
[Character Creator 5 Trial Download]

[StyleMarshal]

First

CC4 - Camila - AI Render

[eljofa]

installing right now
this is game changer

Thank you RL

[eljofa]

I am entering in the prompt – new york city street in the background – and it doesn’t show the street… what to do?

[linus10111]

People, myself included, have always been saying “RL, upgrade your renderer,” but with the path Reallusion has taken toward AI-assisted rendering, IMHO there’s no need anymore. With AI stepping in, it’s a whole different ball game. Exciting times up ahead!

[Room101Studios]

I was still salty that you guys deprecated iRay as a renderer and as much as I LOATHE AI, I’m thoroughly impressed this time. I typically do my final renders in UE, after I run a low test through iRay, but who knows where this will go, especially after CC5. No LoRA, as I didn’t have time to set it up before I left for work, so this was quick and dirty, but I tip hat to you guys.

You did good.

[True_Vision]

ok this will great when it works correctly. Right now veins sink in and make it look like an alien and totally changed the face doesnt even look like the same character and totally change the skin color. But i see good things it does clean it up and make it look better but if i wanted different i would make it that way. So far its making changes not making better. I would be glad to work with you guys on it. I have used iclone from the first version. I am an tester if it can break i will find it.

[Mauritius_Strato]

Finally! I stopped to use Character Creator because after a long work to have the character I want, the rendering was not so real. So I moved in Fooocus or ComfUI . But now I can restart with my beloved Character Creator!

[Mythcons]

I’m very excited about this. I love how it samples existing 3d data, combines it with the render and then creates an improved version. Exactly the way this technology should work. Not as a replacement for creativity, but a means to enhance it.

[Mythcons]

You can dial down the AI effect so that the changes to the render are more modest. I think many fill find a good balance so that the AI doesn’t take over the render result.

[Mythcons]

It depends on your scene set up. Remember that this is designed to sample existing 3D Data. Try adding some prefab asset or geometry in your scene. The AI will sample that geometry data when it applies the AI effect.

For example. if you put a giant square in the background, the AI should sample the contours and depth of that square, which can be used to generate a building. At least that’s my understanding.

[Digital_Puppets]

yeah i think this will be the best approach, just use AI to get the render quality looking better, but it think for the best results you still need to do the work to get it close.

[4u2ges]

Yes, looks like an interesting toy to create pretty stills… but that’s about it.

Video mode is practically unusable. To generate a descent details quality (such as face, fingers), you need at least 2K render, for anything less or for a bit distant objects you either do not get much improvements (with low AI Creativity), or details are all smudged.
I was not even able to get a 720p video render without getting insufficient memory errors (that is given I have 24GB VRAM).
Not worse the effort at all to go though all that for 5-15 sec videos (which is what you will get most I think - if you are lucky).
That’s my first impression, maybe someone will have more luck with it.

Anyway, if you are after stills, then you may have plenty of fun experimenting.
But here is an AI Creativity progression I’ve made (just to say been there… done that )

[linus10111]

@4u2ges

I was not even able to get a 720p video render without getting insufficient memory errors (that is given I have 24GB VRAM).

Interesting observation. I had hoped to get video out of mine, which is 10GB VRAM. What are your full computer specs?

[4u2ges]

I have an older generation i7 processor, 32GB of RAM and 3090 with 24GB VRAM.

[linus10111]

Im still downloading setup files.
Using an i9-10th gen processor, 32GB ram and RTX3080 (10GB VRAM).
Just read the system requirements. It seems this AI is dependent on GPU power more than anything else, so I’m guessing that my renders will be slower than yours. I was really excited to jump into this AI stuff, but that excitement’s starting to dip a little. Looks like video rendering could end up being more of a headache than I hoped

[SCE]

Very happy about this. Thanks for staying current with the times.

[pwnz7321]

This is very cool. I could use some more tutorials to make it so my characters come out looking the same way each time. I looked at the LoRA training video but I need more hand holding to really understand. I’d love to setup my 2nd pc to run renders also. I see there is an option to connect via lan to other pcs. I need tutorials though because I’m dumb.

[Mythcons]

It’s on the cusp of being an actual alternative to a render engine. Give it half a year. The technology is getting there, and this integration is RL planting a flag in the ground and being on board with the future.

How I’d probably use it for now: Render at a low resolution, with a low frame rate, and use either the integrated video upscaler of a dedicated video upscaling model.

Character Creator AI Render Experiment 01

[Sophus]

@4u2ges

Video mode is practically unusable. To generate a descent details quality (such as face, fingers), you need at least 2K render, for anything less or for a bit distant objects you either do not get much improvements (with low AI Creativity), or details are all smudged.
I was not even able to get a 720p video render without getting insufficient memory errors (that is given I have 24GB VRAM).
Not worse the effort at all to go though all that for 5-15 sec videos (which is what you will get most I think - if you are lucky).
That’s my first impression, maybe someone will have more luck with it.

This is how it works. Video AI models must store all frames of the video uncompressed in VRAM to get a consistent result. If a model would be trained on rendering 2K Videos you would need an 80 GB VRAM GPU minimum for five seconds. With current techniques you always need to upscale the output video.

You need something like Framepack to generate videos progressively with low VRAM. It tries to predict the next frame based on the current frames. FramePack