NEW — ILLUMINATE AI LTX 2.3 AUTONOMOUS VISION DIRECTOR v1.2
COMPLETE CIVITAI PACKAGE — WORKFLOW, CUSTOM NODES AND INSTALLATION
ILLUMINATE AI v1.2 is a major quality and prompting update for the LTX 2.3 Autonomous Vision Director.
The workflow analyzes the supplied reference image, understands the visible starting situation, creates a physically achievable chronological scene, writes the complete LTX 2.3 production prompt and validates it against the original image before video generation begins.
The final validated prompt is displayed directly inside the workflow before it is sent to LTXDirector.
NEW IN v1.2
Improved Image Quality
The complete render and refinement path has been reworked for noticeably cleaner output.
Improvements include:
cleaner hair structure during movement
finer skin, fabric and facial detail
reduced coarse and blocky motion textures
improved moving-edge quality
stronger Stage 2 detail reconstruction
native FP16 output without the previous downscale and bicubic enlargement chain
Current quality configuration:
Stage 1: 8 steps
Stage 1 CFG: 1.25
Spatial latent upscale: enabled
Stage 2: 4 steps
Stage 2 sigma endpoint: 0.40
Native FP16 output: 1.00
no third sampler required
More Accurate Autonomous Prompting
The autonomous Prompt Director has been expanded and refined.
It now handles:
duration-aware scene planning
more accurate participant and anatomy tracking
improved position and action recognition
better hand, limb and contact ownership
persistent anatomy consistency during occlusion
improved multi-participant continuity
primary and secondary action validation
more stable movement planning
stronger protection against morphing, anatomy loss and role changes
The system remains image-guided and preserves the original scene whenever a complex transition cannot be completed safely.
FOUR PROMPT MODES
Manual Prompt
Uses your own finished prompt without an Ollama request.
Standard Vision
Qwen analyzes the reference image and creates a compatible cinematic or character-animation prompt.
Adult Assisted
Uses the reference image together with a short written idea and expands it into a complete chronological scene.
Adult Full Auto
Creates the complete scene automatically from the reference image, selected duration and visible starting situation.
Adult modes require confirmed adult and consenting participants.
INCLUDED VERSIONS
STANDARD — RECOMMENDED
The main release version.
Use this version first for the best balance of:
image quality
motion quality
anatomy consistency
render time
workflow stability
ANTIGHOST — EXPERIMENTAL
An additional experimental version with optional temporal Anti-Ghost processing.
This version is included for testing and comparison. It may improve ghosting or duplicated motion in some scenes, but it will not necessarily improve every video.
The Anti-Ghost processing is disabled by default and can be activated for individual tests.
IMPORTANT — NEW CUSTOM NODES ARE REQUIRED
The updated ComfyUI-ILLUMINATE-AI-LTX node pack included with v1.2 is mandatory.
The newest workflow will not work correctly with an older ILLUMINATE node pack.
The installation method has not changed.
Installation
Close ComfyUI completely.
Delete the existing folder:
ComfyUI\custom_nodes\ComfyUI-ILLUMINATE-AI-LTXAlso remove the old standalone cleanup package if it is still installed:
ComfyUI\custom_nodes\ComfyUI-Illuminate-Ollama-CleanupCopy the included folder:
02_CUSTOM_NODES\ComfyUI-ILLUMINATE-AI-LTXinto:
ComfyUI\custom_nodes\The final path must be:
ComfyUI\custom_nodes\ComfyUI-ILLUMINATE-AI-LTX\__init__.pyDo not create an additional nested folder.
Load the included workflow and use:
ComfyUI Manager → Install Missing Custom NodesRestart ComfyUI completely.
Load the new v1.2 workflow JSON.
A browser refresh with Ctrl + F5 is recommended after updating the Custom Nodes.
OLLAMA AND QWEN
The autonomous Vision modes use:
fredrezones55/Qwen3.5-Uncensored-HauhauCS-Aggressive:9bInstall Ollama and run:
ollama pull fredrezones55/Qwen3.5-Uncensored-HauhauCS-Aggressive:9bOllama must remain running whenever Standard Vision, Adult Assisted or Adult Full Auto is used.
Manual Prompt mode does not require Ollama.
REQUIRED BAT FLAGS
The included system check expects:
--fp8_e4m3fn-text-enc --fast-diskExample:
.\python_embeded\python.exe -s ComfyUI\main.py --windows-standalone-build --fp8_e4m3fn-text-enc --fast-diskA ready-made BAT example is included in the package.
FIRST TEST
For the first successful generation:
load the same starting image into the LTXDirector and the Vision Reference Image node
use the STANDARD version first
leave optional LoRAs disabled
keep Anti-Ghost disabled
generate at 24 FPS
enable RIFE only after the normal 24 FPS output works
review the final generated prompt before LTX generation begins
PACKAGE CONTENT
The download includes:
v1.2 STANDARD workflow
v1.2 experimental ANTIGHOST workflow
updated mandatory ILLUMINATE Custom Nodes
Windows Portable BAT examples
Ollama/Qwen installation BAT
English installation guide
German quick-start guide
model and node checklist
troubleshooting information
release notes
Models are not included in the ZIP.
Please read 00_READ_ME_FIRST.txt and the included installation guide before the first run.
Description
Version Changes / Notes
ILLUMINATE AI LTX 2.3 v1.2 is a major quality and prompting update.
What changed
Improved overall video quality with cleaner hair, skin, fabric and moving details.
Refined Stage 2 reconstruction for a better balance between sharpness, stability and reduced coarse texture.
Stage 2 now uses 4 steps with denoise 0.36.
More accurate autonomous prompting and better interpretation of the reference image.
Improved participant, anatomy, position, contact and action consistency.
Better handling of partially hidden hands and temporarily occluded anatomy.
Stronger protection against morphing, disappearing anatomy and incorrect role changes.
The final validated prompt is still shown before it is sent to LTXDirector.
Important installation notice
The updated ComfyUI-ILLUMINATE-AI-LTX Custom Node Pack included with v1.2 is mandatory.
Older versions of the Custom Nodes are not compatible with the newest workflow and may cause missing functions or incorrect behavior.
Installation has not changed:
Close ComfyUI.
Delete the old
ComfyUI-ILLUMINATE-AI-LTXfolder.Copy the new included folder into
ComfyUI/custom_nodes/.Restart ComfyUI.
Load the new v1.2 workflow.
Included versions
STANDARD — recommended main version.
ANTIGHOST — EXPERIMENTAL — optional test version with additional temporal anti-ghost processing. Results may vary depending on the scene, so STANDARD should be tested first.
FAQ
Comments (7)
Thanks for sharing. This is a very interesting WF. I did not tinker much with LLMs myself and I really enjoy exploring this. I got some questions though if you do not mind. I could not find an answer in the txt file or other instructions for:
1. The very bottom left is the EDITABLE DEFAULTS for the full auto settings. I assume editable means I am allowed to change things, and while some make sense to me there seem to be keywords used, is there a list of what words do work for the different settings?
2. I often got cumshot in the end, almost always. I cannot find an option to not get it. Reading through the different stage prompts I found a line that said something like "respect NO_CLIMAX and LOOP". Is that maybe keywords I can put in the EDITABLE DEFAULTS somewhere? Because I did not find these two options somewhere in the workflow.
Thanks, these are very good questions — and you actually found something I need to improve.
The Editable Defaults can be changed, but there is currently no documented list of supported instructions. I will add one so it is clear which settings and values the Director understands.
You are also correct about NO_CLIMAX and LOOP. These controls are referenced internally, but I did not expose them properly in the workflow interface. That is why the Director may frequently choose a cumshot as the ending.
I will add a clear climax control with options such as AUTO, NO CLIMAX, CLIMAX and LOOP / CONTINUOUS ACTION. Until then, I do not want to give you an unreliable workaround without verifying exactly where the instruction must be entered. Thanks for catching this.
@IlluminateAI Alright then I will be waiting patiently (impatiently xD) Thanks for letting me know. Can't wait.
@bowiba1265909 I think I've fixed it just give it a try.
@IlluminateAI I finally had some time available to try it out. Will still need to test more later but my first impressions are very good. The "no fluids" option definitely works so that's great news. Also thanks for providing a list with the commands. Putting the LLM prompts in txt files was also a good decision in my opinion. The WF looks much cleaner now, and less overwhelming for beginners.
What I like the most so far though is that it runs much faster for me than the last one. I assume it was the memory leaks mentioned. I have had to wait minutes for the LLM to finish when I changed images or prompts, longer than the actual generation of my videos. Now it works really fast even when it has to analyze something new again.
Can't wait to play around with it more. Will definitely check for more updates on this later, too.
And if you call it good enough and move on to something new I say thank you for the WF once again.
@bowiba1265909 Thank you again, I really appreciate the detailed feedback!
I’m not completely done with the LTX workflow yet. I’m currently developing a custom node that is intended to reduce the ghosting issue and, in the best case, eliminate it as much as possible. It is still highly experimental though, so I don’t want to promise too much before it has been properly tested.
The next goal is to push the image quality even further while also improving the overall performance and efficiency of the workflow wherever possible.
My Krea 2 workflow has also received a huge update recently. It now includes an auto-prompter, a major user-interface rework, cleaner organization, and several improvements to make it easier to understand and use.
The same kind of rework will gradually happen with my other workflows as well, including Pony, Illustrious, and everything else I offer. My goal is to bring them all closer to the same cleaner, more optimized, and beginner-friendly standard.
I’m also always happy to receive more feedback and new ideas from users. Many of the improvements in these workflows came directly from people testing them, pointing out problems, or suggesting features I had not considered before. In the end, everyone benefits from sharing their experiences and ideas.
So there is definitely more to come. Thanks again for testing the workflow and supporting the project!
Hi.
Just wondering: Is there a reason why you use old checkpoint 10Eros_v1_bf16.safetensors over the newer ones?
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.