Let's go passthrough
Added 2024-12-03 16:17:21 +0000 UTCHello you guys,
I’m excited to share a mechanism for creating a fantastic passthrough solution for those of you with compatible headsets! This setup works perfectly for the Quest 3 and Quest 3S. As for the Quest 2, I’m not entirely sure since I don’t own one—if any of you test it, please share your results.
Here’s how to get your play area represented in VaM so that characters can move around, avoid obstacles, and interact seamlessly with your real-life environment:
Scan Your Play Area
Use this video as a guide: Scanning Tutorial.
After extensive testing, here are some tips for the best results:Light: Ensure your space is well-lit—daylight works best.
Avoid Mirrors: Remove or cover reflective surfaces, as they cause holes in the model.
Furniture Cubes: Decide if recognized furniture cubes should stay or be removed based on what works best for your setup.
Check and Adjust: Take your time and review the scan multiple times. Over-scanning can lead to diminishing returns.
Ceilings: Scanning the ceiling is optional. If space is limited, prioritize walls and objects.
Size Limitation: While there is a limit, I scanned my entire 70m² flat without issues, though the headset did warn me near the limit.
Dump the Model from the Headset
The tool I've developed makes this process possible. Visit the Membership section of my Patreon for a link. Open it in the Quest browser and click "DUMP ROOM MESH". This launches AR mode and saves the mesh file on your device.Choose One of These Options to get the VaM CUA:
Option 1: Email the file to me at vaan20.voxta@gmail.com, and I’ll create the VaM CUA file reflecting your play area. I’ll send it back as soon as possible for all my Patreons.
Option 2: Transfer the file to your PC via email or USB (from the Downloads folder). Follow MacGruber’s guide to create the assetbundle yourself.
Set Up in VaM
Place the CUA file in the Custom/Assets folder.
Create a new Custom Unity Asset atom and add the Chroma Keyifier Plugin to match your passthrough key color.
Use the plugin’s wireframe view to align the CUA with your play space.
In VaM’s VR1 Menu, disable "Thumbstick Navigation" to prevent accidental misalignment. This ensures the CUA remains in place even after restarting VaM.
Select your CUA as the floor in the Places Module.
Important Note:
If using the Wizard, avoid selecting "FULL + FLOOR"—this creates an additional floor, which is unnecessary.
Comments
You can do Cost Management with Voxta by getting an account at OpenRouter.ai - there's a list of 260 different models there, and you can set up OpenRouter multiple times so you can use different models (costing different amounts) based on what you need it for. For instance, you can use an expensive, but excellent model for Summarization (OpenAI GPT 4o - $2.50 per million tokens) since that happens infrequently, and then use a cheaper model for Text Generation (Llama-3.1-70b-instruct - $0.12 per million tokens).
TomahawkJackson
2025-01-10 23:02:37 +0000 UTCOh, now I've read that you've already tried that. Well, the credits are actually a value thst translates to tokens used by LLMs. Each model costs differently. They can also be used for speach generation so it's difficult to say that they will last for said amount of time. They top the balance every month so if you run out of them you either need to buy a higher tier or wait.
vaan20
2024-12-12 14:01:20 +0000 UTCOh, that's not a question to me. Can you please post on Voxta discord on the help channel. You'll get and answer quickly.
vaan20
2024-12-12 13:58:36 +0000 UTCQuestion: I'm very interested in your, and Voxta's content/potential. I have a 2080Ti card, which runs VAM fine, and other local AIs, but I think it would struggle with Voxta and VAM. That being said, how do Voxta cloud credits work? What does '1500', or '5000' get someone? Time on server? Words processed? Both? How long do these credits last? I wanted to ask this question on Voxta's patreon, but in order to post questions, you need to upgrade to a paid tier. I've been following your posts for a few months, and I must say, you do amazing work. TIA for your help.
LivFreOrDie
2024-12-12 13:56:58 +0000 UTCThanks
Clayfacer
2024-12-03 22:38:06 +0000 UTCAmazing 😛
Stalker
2024-12-03 22:13:25 +0000 UTCHaving a comprehensive tutorial is essential to achieve what I'm striving at. The plugin should simple and understandable for everybody.
vaan20
2024-12-03 21:50:04 +0000 UTCThanks Vaan20, I'm sure the criteria you used to make the updated tutorial will be excellent to see the potential and practical application of alm/bundles in action.
Rolo
2024-12-03 21:47:42 +0000 UTCI'm looking at my ALM 101 tutorial script and I'm actually missing one thing in the implementation to say that I've managed to achieve the CORE functionality. I don't want to describe the basics if there will be one more way of defining actions in the scene that would be using states rather than routing. The same way that is used in scenarios. So then you'll be able to define a huge set of actions and don't link them at all and still have a functional scene. They will be called Global Action and you won't be able to route them but you'll be able to control them using flags.
vaan20
2024-12-03 21:36:20 +0000 UTCI did not forget. It's very important for me to make this as several users asked for something similar!
vaan20
2024-12-03 17:47:41 +0000 UTCExcellent update, friend Vaan20! Please don't forget to make a tutorial from scratch to understand and see the potential of ALM / BUNDLES in action. Thank you very much.
Rolo
2024-12-03 17:45:45 +0000 UTCThanks again for your efforts with this project - it’s amazing! I recommend disabling the thumb stick moment and player height adjustment before creating your scene. This will prevent you having to re align every time. And yes, make sure to use the wireframe option in the Chroma Keyifier plug in during setup. (Looks like you’re in the Matrix - it’s pretty cool!)
Linchpin636
2024-12-03 16:41:38 +0000 UTC
