VaM2 September and Early October Progress Update Video
This video shows some of what has been worked on in the last couple of months. There is a lot more to report on beyond what is shown, so I'll be typing up a longer text-only progress update and get that out soon. I didn't want to delay any further, so the video is coming first this time. I apologize for this taking longer than I promised. I ran into some bugs and usability problems while making the video. I decided to fix these immediately to make it easier to make the video, but also to produce a video mostly free of awkward issues.
The video highlights improvements and new features beyond what VaM was capable of. Please note that nothing shown is in final form.
Here are the caveats of this video you should be aware of:
Physics realism will improve further from what is shown
Rendering will improve further. Not all of the HDRP features are turned on or set to maximum quality, and the materials and textures and not final.
The collision model here is very simple with just one simple collider per joint, which is far less accurate than VaM. This is not the new planned skin-accurate collision and soft-body system. This simple collision system is just enough to make basic posing easier.
The clothing and hair shown is a stand-in for the real thing coming later. Some of the delay in getting the video out was getting this type of skinned clothing working to sync up with the underlying model skin. The video could not go on Youtube without putting some clothing on the model.
Fingers and toes are not yet rigged up to physics so are stiff in this video. The pin posing and muscle system shown in the video will make creating finger poses very easy and natural without fooling with individual joint controls. Coming soon!
Here are the improvements from VaM we are trying to show off:
All of the VaM2 character model work previously reported on along with some additional improvements since the last report. It shows a high-resolution Daz3d Genesis 8 model using dual quaternion skinning (DQS), high definition (HD) morphs, and joint controlled morphs (JCM) all actively working with low performance overhead and rendered in the new HDRP system from Unity. All of these are vast improvements over VaM.
Improved body physics and dynamics. The VaM character body was quite bouncy and it took some wrangling to get it to react in a natural way. The new VaM2 physics model is still a work in progress, but it has much better stability than the VaM model, which will ultimately lead to better realism. The physics in VaM2 is more easily tunable thanks to the physics rig tools that have been worked on. These tools are not shown in the video. More info coming on this in the text update.
New pin posing system. This pin system allows grabbing the model in the exact spot touched rather than being limited to fixed controls at joint origins like VaM had. Pins can be quickly added and removed as needed without the hassle of opening the UI. This makes posing orders of magnitude faster and easier than VaM.
New muscle joint drive and posing system. This system can have "muscles" try to hold joints in specific positions with adjustable muscle power. This will make it easier to reuse poses as the character tries to hold a shape and not just hold body parts to specific points in space. These muscle based poses also react in a more realistic and less glitch-prone manner when other objects collide with the model.
VaM 2.X September and Early October Progress Update (YouTube)
Comments (103)
Interesting, I can't say much about it to be honest
Projects of this nature are prone to timeline ambiguity. We are entering unknown territory together with the devs.
I really understand and that certain things simply take time is absolutely clear, what I find a shame and not so customer-friendly is very little information to know when to go next. Everyone has been waiting for months for news and these come very little I hope that a first version will come soon before it was said very big there will be some alpha phases until the end of the year the year is .almost over unfortunately nothing has come up until now: - (
I'm no longer giving out estimated dates or plans because I not good at making them and I always miss, leading to even bigger disappointment. We do have internal plans and goals, but I cannot share because we often change direction and timelines. If you want more visibility into the project than what is provided here, please join the Discord server. I have created a new 2_x (2.X) channel there where I am very active and provide a lot more information. I'll just repeat what I said in a reply above a couple days back because that is the best I can do right now. There will be more coming from us to you before 2021 is over. All I can say right now is Soon™️!
Should we expect an final update for 2021? A state of the development post would be great, given it's been around a year waiting for something on 2.0
That’d be cool. I don’t even mind that we don’t have an alpha test yet. Another progress update would be great!
Happy December!
Hope the meshed vr team can make vam create CG level contents and i am going to support you continuously if that is within your future plan!
Hey meshed! Got a question for you... How do you recieve your old morphs and textures files from the current VAM build?
very excited for the alpha! Any idea when it will be released?
i have been subscribed to this project since September 2017 and ive gotta say, watching you evolve and develop this has been an absolutely fascinating experience. setting aside the sexually explicit nature of the project, what youve been doing, from a technical perspective, is nothing short of amazing. i dont know enough about physics simulations or coding or game development to properly convey my opinions in an organized manner, but ive been a gamer for my entire 30 years on this planet so im no stranger to interacting with physics engines and all the wonkyness that takes place when a game tries to emulate reality in terms of animation or physics. and what youve created far eclipses every other project of this type. No where else have i seen anything approaching the complexity of simulating soft body, bones, skin or muscle physics and doing it as well as you have. oh and a second thing: i make mods for games, and ive made a few for this one as well. and the fact that youve made all of this (as far as 1.X, anyway) so easy for other content creators to modify, and the way that you worked WITH the community to make it that way, AND how models, skins, physics, bone structures, animations, visual effects, etc are essentially drag-n-drop is outstanding. im excited for 2.X and what you have in store for us, great job. and all for only $8 a month
Thank you! I really appreciate everything you said here and for your long term support. It hasn't been easy, and 2.X brings a whole new set of challenges, but they are challenges I'm excited to tackle! We're currently working very hard to get the first 2.X alpha released which will serve mostly as a tech demo so everyone can see for themselves how it is evolving.
You don't want to be a billionaire? I get it but I can't wait to visit your little corner of the metaverse
do you have any IPO plans?
I prefer to keep this a small private project that I am fully in control of, so no to that one! :)
I hope you know games arent developed overnight. These guys put in a lot of work to make this happen but we just need a little patience ;)
I will cancel my support in January, think this is dead....
Please join the official Discord if you haven't (you can link up to it here: https://www.patreon.com/settings/apps). I am more active there. The 1st 2.X alpha is very close to release.
This project has not been updated for 9 months, how long do we have to wait?
As long as it takes! As mentioned multiple times in the last progress updates, 2.X is being built from the ground up. 1.X was in development for a couple of years before this Patreon project even launched so it had a leg up on development. 2.X started from scratch while the Patreon project was ongoing and 1.X was still being worked on. 1.X is now on the shelf, and we are fully on 2.X. 2.X will take a long time to develop. The 1st alpha is coming soon. It will have limited capabilities, but will show off a lot of the new technology and potential.
looks gretat
Will 2.x have some sort of keyframe animation tool (like Acidbubbles' Timeline) built-in or will we still need to rely on plugins for that? If so how would the new pins system work with keyframed animation?
Yeah 2.X will have built-in hybrid key frame and motion capture animation system. Pins are meant for making joint angle poses. Animation will directly drive controls and/or joint rotations. The system will be flexible because you might want to animate muscle/joint action and not fixed points. Animated muscle/joint system will allow a lot more dynamic interaction with the animation. But if you want to just run a regular animation you would have all the pins off and all the controls on. Or you could use pins to slightly modify the animation. edit: in case it wasn't clear, pins won't be animated edit 2: keyframes could capture traditional controls or capture joint angle poses or combination of both. You can also directly set joint angles without using the pin or pose tools, and use keyframe capture of that.
Ive had to step away from using VAM because performance is so poor on my older system. As we reach vam 2 users are pushing scenes to their limits and as a result for the first time many scenes are just too much for a 1060. I do have a 3080 on the way but just incase that never arrives haha I my request would be optimization. Also anyway to keep those penises in the pussies is a big win haha All I seem to do is make scenes and use scenes where it will go anywhere but where its meant to LOL :-D
In case you're unaware, here are two plugins that might help address both your problems: https://hub.virtamate.com/resources/diviningrod.8992/ https://hub.virtamate.com/resources/givemefps.1367/ Either way, Meshed has made it clear that performance is one of the primary goals for 2.0. Can't wait!
Request for V2: - Please switch the button and the trigger! It's really uncomfortable to work with menu when editing. Maybe an option to switch? Or to map the controller? - Another request: Could we have a atom to display and interact with our desktop? Like in SteamVR and Occulus Home? That would be great to use VaM instead of Home to chill. Thank you!
I think there is already a way to switch the button and trigger controls but options to map everything seperately could be interesting.
Just got lucky and bought a rtx 3080ti excited so see how it runs on VAM.
Will you put in future a DLSS option? Schould be possible.
Yes DLSS will be coming very soon once I move to Unity 2021.2 which I will do when it moves out of beta in the next month or so.
With regard to VRSS NVIDEA says: Submit your games and applications to NVIDIA for VRSS consideration. Actually , I am thinking, that even current version might benefit from VRSS.
I submitted stuff to Nvidia previously and was ghosted by them, so I won't be doing that. I will support it when Unity adds support for it like they have done with DLSS and AMD's new solution.
That all sounds super good, I'm looking forward to testing the first version
I know you said you can't go much into detail about the timeline for this game and I understand. Would the new timeline allow more in-depth features that would make it easier for some to make short films that have camera angles, lighting and emotive models? This would probably be something more suited for desktop mode.
2.X should be easier to create content with. You could use a subscene for camera and lighting and load subscene presets to quickly change perspective and lighting. This won't be much different than 1.X other than faster load times. The 2.X character models have more features we have not explored yet, including a face rig. This should pave the way for making custom expressions. The included expression morphs are also of higher quality than 1.X. So I think the answer is yes.
Are you planning Supporting NVIDIA VRSS ? . Maybe this will make improvements for rendering. NVIDIA VRSS 2: Dynamic Foveated Rendering, No Assembly Required https://developer.nvidia.com/blog/nvidia-vrss-2-dynamic-foveated-rendering-no-assembly-required/
2.X can run in forward rendering mode with MSAA, so it sounds like it should just be supported out-of-the-box based on the article. Support for DLSS and AMD FidelityFX should also be coming as Unity 2021.2 now supports those. We are currently on 2021.1 but plan to move to 2021.2 as soon as it is out of beta.
I was wonder if in VAM 2 there will be possible to sth like cum on clothes ? Now as there is no real cum and we use Cloth like cums, that are somehow sub transparent if i put it "on cloth" via settings it render cloth disapear at that fragment so its useles.
Fluids will be supported in 2.X using a particle physics system. That combined with the increased collider density for improved collision accuracy should give realistic results.
Any idea when the first release is ready for public ??
Hi, Seem Great and hurry to test it ! Hope to see a better performance on 2.X version. I have a high end PC and difficult to exceed 30-35 fps in VR. Sure it could be possible to reach 50 fps but with many sacrifices and VAM is no longer of interest. I would like to use high quality physics, lights, animations, elaborate clothes and more without too many sacrifices...I build scenes and animations on my desktop and spend lot of time to optimize performances to keep 70 to 80 fps (35 fps in VR).
One of our top 4 major goals is to improve performance. We are tackling the major physics CPU performance bottleneck that occurred in 1.X by using a new custom skin physics system. We have not yet written this, but early experiments suggest we can improve both performance and accuracy of skin physics in 2.X. If your GPU is the bottleneck, 2.X will have more scalable graphics and performance settings that should be able to get you higher framerates without as much visual sacrifice.
This is just awesome! Such great work here. Question on 2.X performance improvements: how does this translate in VR? What are the minimum specs to see optimal vr performance with 2 models, all or most physics, soft body enabled, with 3 or more lights? Or rather, what are the target specs you're aiming for? For reference, I have 32GB RAM, an ryzen R9, and a nvidia 2080 super, and in 1.X with the scenario above typically dips me to 15 FPS. In desktop, performance is waaaay better (predictably).
In general 2.X is more demanding with rendering, so will need a higher end GPU to look best, but we are finding performance is a lot more scalable by changing performance options. 1.X has a major CPU bottleneck from physics that 2.X should solve. Unity still does a lot on the main thread, but we think we can overcome most of this and each additional character on screen won't cost as much, allowing for more with better performance. We don't have exact numbers because we have not implemented skin physics which will be the most demanding CPU feature.
SO happy to be a part of the future. Great, GREAT job peeps
Will it be possible to access muscle tension data? Thinking of dynamic muscle tension morphs.
I will probably come up with something. Unfortunately there is no way to read that kind of information from Unity Physx for the type of joints we are using. But it could be estimated based on position of joint, the joint muscle power setting, and how well the joint is holding the target rotation. It won't be perfect. It would be much better if we could read back the actual forces being applied by Physx.
Greaaaat work guys!!! :))
Is there a chance to improve the passtrough view? Currently, we can choose a black background and we see the model quite poorly lit (a bit like a ghost), but is there anything to improve here? I've heard of problems with the passtrough API, but seeing the models in your own room is a fascinating prospect.
Longer term I want to support AR and passthrough, so yes I will be looking at it. I don't know when yet.
I’m wondering if the team has a sort of timeline for working on things like clothing/hair, soft body physics, and UI. To know what to expect for future updates, will they be worked on one at a time or all at their own pace?
We do have a timeline and planned order, but we are not sharing because it will set false expectations of when things will happen. We will report on what we will be working on next after each update or release, but not discussing the bigger plan. Too many times I have promised something at a specific time and been unable to deliver causing a lot of stress on me, and frustration to backers. Some things will be worked on in parallel. One example is soft body physics will take a long time to develop so it will be done in parallel with smaller things that can be released to the alphas as they come out.
will the plugins and assets from 1 x also work in 2 x?
No those will not work without porting. Plugin API will be vastly different so nothing from 1.X will work out-of-the-box. Assets could work by converting them to new custom atoms as Custom Unity Asset atom is going away.
I was hoping for a more playable form of VAM. It is mainly a design application, like Daz3D. But if you want to just play it like a GAME, it is not possible. This is because it doesnt have a separate game mode, where you can load animation into click icons. I seen some plugins where they try to make simple interface mode for gaming, but it doesnt really do this.
This would take over the world
Vam 1.x is going to be a 200gig relic on an old portable hard drive, visited with nostalgia like a super Nintendo. Looks great!
I was wondering if all that improvments on skin meshes and collisions will make part like vagina more natual, now as you try to insert eg penis, vagina open so unnaturaly , like there is some circle colider inside that enlarage on insert ,it totaly not depended on inserting shape, so if you try to put sth thinner or flat , vagina orifice open roundly, same for anus. :) also many times inserting items clipping through skin and orifices, jump from vagina to anus and back. it breaks immersion :)
Physics around the vagina, anus, and mouth are going to be the hardest to get right, but we are going to work hard to improve what is seen in 1.X. It will require a lot of experimentation and new solutions. We will be looking at this when we get to soft-body physics.
With these new systems it looks like animating on desktop will be way easier and much more user friendly. Nice job! I usually animate in VR because the current system feels a little clunky for me on desktop but this looks really promising.
Yeah - it is definitely easier to pose on desktop. Last night I spent time using both 1.X and 2.X on desktop to compare posing and physics a bit more. 2.X is way less frustrating on desktop. 2.X physics also keeps the model in a more natural position without fussing with the controls or pins as much. Pins should be useful in VR as well to help moving elbows and knees and not needing to open the UI to turn controls on/off. It is now to hover, click, and drag. Another different click to remove the pin. Vastly improved ease of use is in our top for goals for 2.X, and pins are just one step in that direction.
Looks awesome! Thank you for the update... Is there any more you can add to the "custom atom" mention above? Specifically, will the pipeline be same (or similar)? 3d modeling software (fbx, obj etc)->texturing->unity->VaM? Will there be other steps or tools external to Unity? Access to shaders (material creation) is major issue for asset creators at this point. I mean, we have Vammifier, but being able to create PBR materials with the in-game shader and preview them accurately in another package would be an amazing step forward for content creation 😉
Yeah. A big part of this last month's work I will detail more in the upcoming text progress update was the creator tools. We are using the same tools in house that we will be releasing to the community for building atoms, including the ones we use to make the character atoms. This will require Unity, but we have made tools to make this import process much, much simpler than it would be otherwise. The flow is basically 3d modelling tool -> fbx export -> unity import using our custom tools -> tweak as needed using Unity -> export to VaM using our custom tools -> drop the file created into a special folder in 2.X and you have a new custom atom that shows up in the menu. Materials are HDRP and you will be able to make custom shaders using Unity's shader graph with HDRP target output and they can be included in the atom.
the animation system in VAM is very primitive... it should be much simpler. Also should simply be able to click on animations like they are icons, rather than complex timelines
It might be daunting at first but acidbubble's timeline plugin is actually very user friendly and powerful. You just gotta either figure things out or check out some tutorials. Once you get familiar with it you can make anything you want.
This looks amazing !
Is vr still going to be an option?
Yes. VaM is rooted in VR. The 1st demo will have VR support, but it probably won't have and UI or UI interaction. Just the ability to manipulate the controls and pins to pose the model.
All this with a AI driven chatbot and maybe something like RealTimeVoiceCloning.ipynb Then you can make anyone come to life! Oh the future is beautiful.
Hopefully the AI driven muscle control system will bring something new to the table as nowadays only mocap solutions work somewhat realistic. It would be nice to see the model react and not just replay what has been previously recorded.
Yeah muscle hold allows much better interaction because the joint can still move in a natural way even if other parts are moved or there is a collision along that joint chain. The control pin method in 1.X tries to hold a specific part to a point in space which could cause an unrealistic look.
Hello, in 1.x I met unnatural gravity oriented soft body parts positions and shapes (breasts), i.e. these parts were not affected by vertical gravity force, according to the body position. Breasts shape was almost the same, no matter, if body is laying on back or abdomen. Maybe it is my fault, maybe it is a bad 1.x feature. If bad feature, will be improved in 2.x? Thanks for answer.
It is funny seeing the same question asked back to back! Please see my response to the post directly above this. Short answer is in 2.X gravity will affect position of breast joints which will affect the rendered shape, so yes.
Will there be improvements to gravity-based physics? For example, even with the aid of some morph-based mods out there the breasts in 1.x have some rather unrealistic deformations especially when the model is in various positions, such as lying down or upside-down. And the built-in spring-based 'fat' system is good for general jiggle effects, but doesn't accurately simulate the fluid dynamics of real breast motion with proper influence of gravity. Will the increased number of colliders and Genesis 8 mesh resolution also pave the way for more realistic soft body motion (in addition to collisions)?
Just want to add that breast physics definitely seem to be a tough thing to tackle, I didn't mean to sound overly critical of the current system. But there only seems like so much that can be done with bone-based squash and stretch, morphs, and jiggle/spring deformers. I was just wondering if the new technology might allow for something closer to fluid/particle based motion.
Great beginning foundation work on a massive undertaking. This will take us into almost real experiential 'suspension of disbelief'. Heed not the naysayers and impatient, Slow but steady as she goes. Bravo!
Thank you!
Sounds great! Hopefully my PC can handle it. Can't wait to see a demo video. Maybe a side-by-side comparison between 1.x and 2.x
One major goal of 2.X is improved performance. The rendering in 2.X is a bit more costly for both CPU and GPU, but the main CPU physics bottleneck from 1.X should be eliminated in 2.X. 2.X should also have more scalable performance with performance options that actually allow it to run on lower end hardware.
Will there be any improvement in the skin collisions within a models body to prevent for example breasts that overlap into eachother when pushed together or hands going a few inches through the skin in some areas of the body or more realistic body fat physics?
Yes. The plan for 2.X is for skin-accurate collision. 1.X uses about 1000 colliders per person atom if soft-body is on and most of these are in the soft-body areas. 2.X is going to use roughly 1 collider per skin vertex, or 18000, to bring the accuracy up. We are planning on a custom collision system to do this because this is well beyond what could be done with Physx in Unity. We will be diving into this soon to see what is feasible.
Video looks cool
I only saw a bit of it in the morning due to work, but I like what I see, and the following conversation looks promising as well. I'm more at ease now, this project is going along nicely. Well done, and I will patiently wait for alpha.
If the pose saves are that lightweight, will it be possible for us to create a keyframe system in a mod that could cycle though these quickly to make muscle based animations? Man the possibilities are endless, good luck! The JCMs look good too! edit: I just read the above post lol
Yeah 2.X will have a built-in keyframe (and more) animation system. Thank you!
It looks excellent. Thank you!!! I have a favorite character(appearance) in Vam 1.20 Can I import her or have her somehow in Vam 2.X? I am happy as long as I see her in 2.X That's my most concern. Please let me know.
Please see the post above. 2.X will not be able to load 1.X content. 1.X creators might be able to port some looks over with some effort.
This is looking awesome! Thank you for the update. My only question, which I may have missed if someone else asked already, what about backward compatibility? I have over 100GB of VAM content on my system. Will all my girls still work on 2.0?(and my guys and others as well.) I can’t wait to know more. Thank you again.
Sorry 2.X will not be backward compatible. There might be a path for some creators to port clothing and hair to work in 2.X, but almost everything else will not be compatible. 2.X was meant as a fresh start to break free from systems that are holding 1.X back in terms of performance, quality, and usability. 1.X still has a long life ahead as 2.X won't match feature set for quite a while.
Using genesis 8??? Omfg!!! It’s possible import custom morphs and textures too???
Yeah, custom morphs and textures will be possible. The process to get them into 2.X will be different. Morphs are more complex due to using HD mesh and the way grafts are going to be done. They can no longer be imported direct from the dsf files.
Looking good dude! The pin system will definitely be hard to get used to, but it appears to be a much more streamlined posing tool. Are the pins going to be marked with XYZ - RGB?
You can still use controls if you prefer, and they will be needed when you need to fix something in place, like a hand on a table. Pins can only pull position and don't control rotation. I will likely add a line that appears if the pin is pulled away from the skin so you can see what it is pulling. Not sure what you mean with XYZ - RGB.
That looks great! I can hardly wait for the alpha build.
The new version looks awesome in the video. I am extremely hyped up about 2.X 🔥🔥 😃 🔥🔥
Hey Meshed, looks really good. An important feature I would beg for would be a feature to have the joint controllers exactly fixed at their positions without any shifts or delayed followings to the model in animations like in VAM 1.x so far. Even with the xyz fix checkboxes active, the controllers always shift around. That would be urgently needed to have CUA assets like high quality shoes (instead of importing them with the cloth importer who always leads to weight failures and therefore distorded shoe meshes) or CUA props attached to the model's hands and feet without any shifting around during animation like currently in 1.x.
Yeah all of that will be fixed in 2.X. You will also be able to attach props to specific skin vertices in addition to direct to bones. Although not related to what you are mentioning, CUA is also actually going away to be replaced with full custom atoms.
Well I guess the future is here ...or will be when this gets finished :-) greatwork!
Awesome! Will we be getting a muscle based system so that we can contract/expand muscles for a guaranteed physically accurate simulation? (Toribash as an example)
I plan on exposing muscle controls per joint and per group (core, etc.) to allow controlling or animating the muscle strength. I'm planning to also allow capture muscle strength info in the pose itself so that muscles strength can be activated in bursts and then relax more once pose is established. This should allow for some interesting muscle based actions, like kicking, etc. I'm hoping longer term we can support active ragdoll simulation which could allow for AI to drive the muscles and perform balance and nature reactions.
That hair looks awesome, i know its a stand in but is it possible that maybe we can port many of the amazing DAZ hairs into VAM with great aplhas and good wrapping?
Yeah this is using the HDRP Hair shader which is giving better results than I expected. This isn't even a very high quality hair asset. What is shown will work for simple static hair that does not need to be simulated and I do plan on supporting this as an alternative. After seeing this, I have some ideas about how to possibly make these types of geometry drawn hairs into pseudo-simulated hair to allow some movement through physics.
How exciting !