About to purchase a bunch of Reallusion products!


https://forum.reallusion.com/Topic475365.aspx
Print Topic | Close Window

By infoxiasma - 5 Years Ago
Hi all,
I wonder if anyone here has used a similar setup to that which I'm intending to buy this week?
We want to create avatars which will be exported to be used in Unity. For that side, we're thinking Character Creator 3 Pipeline or CC3 for iClone + 3DXcjange Pipeline. And we'll add in Headshot + Skingen.

We also want to motion capture.  We will use Leap Motion (actually, IR 170) + ARKit on an ipad Pro + Kinect 2.  A bit "domestic" rather than "pro" but this will do the job for 3-6 months and then we're likely to go Qualisys (we will have access to a studio and hardware).

On the trial products, I've got Leap Motion and ARKit working together at the same time, and I've got Kinect working but the trial won't let me try all three together.  Note that ARKIt and LeapMotion are through Motion Live Plugin whereas the Kinect is through the Kinect Mocap Plug-in.  Does anyone foresee a problem in this setup?

We will maybe hand-correct animations in Blender, unless Reallusion's suite will do a better job?  I say Blender because our animator is very familiar/skilled there but she's happy to learn new things.

And what about exporting the animations and characters? Is this everything I'll need to get my animated avatars into Unity and Blender?

Thanks all,


By planetstardragon - 5 Years Ago
I'd start small and slowly add things to what I can integrate to my current system,  there's a huge eco system of existing mocap that's either free of very inexpensive to acquire and modify. 

From my own experience in setting up systems in both music and now with my new cg life -  there's always new tech coming out and slowly ingesting new stuff to my existing workflow is a more efficient upgrade than stopping work just to learn a new system that may change in a month. I don't even want to think about the money i've spent in gear and software i've hardly used because by the time i learned it something newer and better just came out.  it never ends! 

If you plan to go hard with iclone products -  then you want cc3 - for character creation,   iclone for all things animation -  and 3DX for non standard importing and exporting -   by non standard I mean features not found in CC3    - 3DX allows you to do things manually that can't be achieved "Automatically" in cc3 and allows more custom detail production in the items -  ie importing sketchup models,  and breaking them down into small parts .....or importing some obscure model made in some obscure software with a triple rig that cc3 won't recognize and putting spring effects on it's nose or whatever.

from what you are talking about the mocap set up -  I'd focus on building a library of stock motions while you wait to have access to the larger better mocap system. My library is currently at around 10k mocaps,  there's not much left for me to mocap lol,  nor do I plan on learning monkey kung fu and buying a 2k mocap set to do something I can easily keyframe or piece together from my library.  I'd rather organize my library and then buy what i need to fill in the blanks after. 

Building a business these days is hard enough than to waste money on something that will also waste time in learning that will be replaced faster than it took in time for you to learn it lol

Good luck in your adventure -  sounds like a fun one!

my 2 cents.
By Peter (RL) - 5 Years Ago
Hi...

Character Creator 3 Pipeline will allow you to create characters and export them with animations to almost any other software. We have export pre-sets for software like Unity, Unreal, Maya, Max etc or you can choose custom export options if using other software not covered by a pre-set.

If you want to create animations for your character and do motion capture then you will also need iClone 7, the Motion LIVE 3D plug-in, and the Gear Profiles for the hardware you plan to use. Apart from the Kinect plug-in all mocap gear profiles work together in Motion LIVE. With Kinect you will have to record your body animation separately as this older device is not supported by the Motion LIVE system.

3DXchange 7 Pipeline is an optional extra and if you need it will depend if you also want to export props and other non-standard characters. It can be a helpful tool to have but if you only want to export characters and animations then you can do all exporting directly from Character Creator 3 Pipeline.
By infoxiasma - 5 Years Ago
Thank you @planetstardragon.

I think you speak a lot of sense and that route really would be my first choice, but much of the motions I need to capture are simply not in libraries where I can purchase them.  Cued Speech and various sign languages (not just American Sign Language), for instance.  I would love to be wrong about the availability of off-the-peg mocaps for this!  If I am wrong, please show me, I will be extremely grateful.
By infoxiasma - 5 Years Ago
Hi Peter,

Thank you for your guidance.
When you say "with Kinect you will have to record your body animation separately as this older device is not supported by the Motion LIVE system." how will this work in practical terms?  My expectation was that I would load up iClone 7 and run the Motion LIVE plugin for ARKit and Leap Motion, start it recording and then start and the Kinect Mocap plugin and start Device MoCap recording in iClone.  Is this not possible?  The trial of Motion Live won't record so I can't see how it operates.  I do see that the trial of Device Mocap (Kinect), when I start recording, the rest of iClone is "locked" (the other controls like Motion Live are disabled).  
Is there no reasonable way to run all three mocaps together (and see it on the avatar)?  I need to watch the mocap as the Leap Motion does lose sight quite often and then we'll have to retake. 
By planetstardragon - 5 Years Ago
For a specialized collection like that,  you may have some luck with contacting universities like
Max Plank institute of psycholinguistics
https://archive.mpi.nl/mpi/islandora/search/motion%20capture?type=dismax

or the CMU -

Carnegie Mellon University Motion Capture Department
http://mocap.cs.cmu.edu/

both of which would have an educational interest in such a collection,  some may have an existing collection that they would license to you for a fee and some may give you a chance to pitch your project and maybe even funding for mutual benefit on your project. - at least thats how I would think considering the uniqueness of your project as a selling point.


if you want to rough it,  iclone itself has some useful hand puppet tools / plugin  that can help make the process faster and easier.....  the best feature,  for me at least is that iclone gives you tools to draft up quick animation scenes that normally are tedious and time consuming in other software ....so you draft something in iclone fast ...and iron it out in blender.   the RL tools have a good grip on taking the "tedious" and "Time consuming out of animation. They give you a chance to really make things really complex too,  but being able to draft ideas fast is priceless.




the pose manager is really cool too..

https://marketplace.reallusion.com/pose-manager


By infoxiasma - 5 Years Ago
That does look rather cool, but it means we'll have to animate everything, even if the animation process is significantly improved.   And I wonder how it would perform on movements like touching thumb tip to little finger tip or crossing fingers? 

Thank you for those links, much appreciated.
By planetstardragon - 5 Years Ago
my pleasure,  I enjoy answering questions and helping because I learn more and get more ideas for myself in the process hehe!

I suspect the pose manager may be a helpful tool for you if you can make a quick library of preset moves
So for example,  you grab an idle motion,   you tell iclone to keep the rest of the body motion but erase just arms,  then you just focus on the arm movements and paste your customized hand gestures from the pose manager plug in.  you could even get creative like removing arm motions from a dancing motion and have the character talk while dancing.  I had a friend that loved music even though he was deaf,  he felt the pulses of music and would love reading the lyrics.

the idea would be to make a library of presets you can cut / paste or merge with other existing motions.  between small saved clips and the pose manager plug in,  you should be able to develop a system that lets you create new unique and complex animations quickly by piecing existing segments together.



By infoxiasma - 5 Years Ago
Thank you again :)

I am now thinking that going with the Perception Neuron V2 (body and gloves) but only using the body (instead of Kinect) should do what we want? It's an extra $2500 I didn't want to spend but it will mean we still get good face and fingers capture and we can experiment with the gloves to see if we can ultimately get something as good/better in there.

This really is to last us 6 months or so until we get (hopefully) unfettered access to a Qualisys rig.
By planetstardragon - 5 Years Ago
I don't have that set up so I wouldn't be able to comment,  but I do know that capturing small details like fingers can be a challenge for most systems - I'd seek out other users that use the system and ask them, and ask them how much clean up after the capture. -  maybe even contact the mocap developer themselves and see if they have any specialized solutions for your needs.
By oonabe - 5 Years Ago
From my experience I'd say Headshot is not a necessary plugin if you have time to recreate the faces by using the sliders on CC3. I got the plugin and there is something that bothers me a little and it's that the plugin uses the original picture as texture and it's not easy to remove it later without having unwanted side effects.

And the problem is that it's specially noticeable now with the CC3+ and real human additions where you can add skin textures, capilarity, make up... to the character. The Headshot originated texture is mostly getting in the way when working with those new and more realistic textures so I've ended up importing the original pictures into the UI as a reference then modeling the character by myself from scratch and I got much better and customizable results than using Headshot.

Headshot would be a better help if removing the picture texture would be an easier task as it is now, because it does some of the job automatically, but you also have to consider it only works on a 2D plane, meaning, you can do a very quick portrait when looking the person on the frontal view, but you will have to do the side view working as I described above, by using a reference picture and tweaking the sliders manually to match the picture, so at the end Headshot isn't really that helpful, and in my opinion for most of the part a waste of money
By Lord_Dreadmoor - 5 Years Ago
@oonabe Headshot is just like any other tool, whereas the quality of the results are in direct proportion to the time a user spends learning how to use said tool in order to achieve the best results. I have a lot of characters in game and animation projects I am working on, that are based on various actors and films I hold in high regard. Granted, while I would always prefer to simply create a likeness inside of Zbrush for far more detailed results, closer to the source material, Headshot is a more than viable alternative for someone who might be just starting out, or someone who doesn't want to spend a month or more perfecting each character design. I also agree with you that using the plugin straight out of the box does produce some irregular skin textures, the fact is that humans themselves have some pretty irregular skin textures, so it's just a matter of tweaking the maps a bit in order to get as close to reality as humanly possible.

I can only speak from personal experience, but the way I like to go about it is as follows: First, make sure the photo being used is a pretty high quality, preferably centered in a square space around 2160x2160. If you don't have a large, HQ photo, you can increase size and resolution, clean up any pixel noise, as well as center, etc., in Photoshop. If you don't have Photoshop, you can use the visible grid function in GIMP, since it is freeware. You make sure to get the bridge of the nose as aligned as close to the center of the grid as humanly possible, keeping in mid that no photo is perfectly symmetrical. Then after you import the picture into the plugin and generate the character and skin texture, it is CRUCIAL that you do this next step BEFORE you activate the SkinGen editor, as doing this afterwards will give you wonky results. What you do is select your character in the scene tab, then head over your texture list and select the head, then exporting ans saving the diffuse map for the face/head to an external folder. Once you have the face diffuse map saved, you then open up the saved copy in your edito of choice, i.e. Photoshop, or GIMP. You create a grid overlay and expand the size and alignment so that there is a perfect center line aligned with your dimensions. The generated map will be 4K so keep that in mind. What I do then is use the rectangle select and select the half of the grid that has the least flaws and best details, then copy and paste as a new layer. Once you have your new layer, you flip horizontal and move into position so that it lines up perfectly with the opposite side of the face, thus giving you symmetrical details. The next step, you merge down the new half (after it is perfectly aligned of course) and then use whatever tools you need to clean up any extreme skin discoloration produced as a photo effect, i.e. excessive highlighting or shadows in the photo that don't translate well to skin textures, as well as the darkened sections on the eyelids generated from eyelashes in the photos, etc. The goal being to get the skin blended as best as humanly possible. Once you are remotely happy with the cleanup, you then export the finished map and then drag it back into the facial diffuse slot in your texture section. This way, once you go to finally activate your SkinGen editor, you will get a prompt telling you that an edit has been made and asking you if you would like to keep the results. Click "Yes". Now, when the SkinGen editor, activates, it will generate an overall skin texture reflective of your polished results. Now you can tweak any adjustments you need within the SkinGen editor itself and get much better results.

There are plenty of tutorials online detailing similar techniques, many offered by Reallusion themselves, or many of the experts who have spent enough time with the software to get their results pretty polished.

Like I said, I can only speak from personal experience, but here are some the results of my character designs and renders from models generated using Headshot and the techniques listed above....

https://forum.reallusion.com/uploads/images/8bdcbafc-70b5-409b-aece-e78e.png

https://forum.reallusion.com/uploads/images/00cf8966-86be-4dc8-989e-75e9.png

https://forum.reallusion.com/uploads/images/f8fb6ab7-520e-4a45-b8cd-f431.png

https://forum.reallusion.com/uploads/images/4a0a42ad-b654-4632-853f-7412.png

https://forum.reallusion.com/uploads/images/3a3c1143-a1ff-4c63-8efd-6f68.png

https://forum.reallusion.com/uploads/images/341d37a5-81b2-4c1d-9003-83bb.png

https://forum.reallusion.com/uploads/images/c58ca593-8041-49c7-b839-c092.png

https://forum.reallusion.com/uploads/images/142e48b1-66cf-4670-bb54-c262.png
By Am7add9 - 5 Years Ago
that shrink wrapping on the clothes:Wow:
By Peter (RL) - 5 Years Ago
infoxiasma (3/9/2021)
Hi Peter,

Is there no reasonable way to run all three mocaps together (and see it on the avatar)?  I need to watch the mocap as the Leap Motion does lose sight quite often and then we'll have to retake. 


Unfortunately not using Kinect. If you use Perception Neuron or any of the other full body Gear Profiles then you can run everything simultaneously in Motion LIVE. With Kinect being a separate plug-in you would need to record body animation in a separate session, save your work, and then record the other motion capture using Motion LIVE.
By oonabe - 5 Years Ago
@Lord_Dreadmoor yeah, I agree that at the end it's a matter of how much dedication the user puts on the work, and also what your project is, I think the textures can be good for games but I'm personally aiming for photo realistic animations and there the "weird" textures can really end up being a problem. I don't know what changed in some of latest updates, but I know I could remove the automatic generated textures that Headshot takes from the original picture and replace it with one of the default CC3 textures but I couldn't manage to do this in the latest CC3 version, which is kind of sad because right now with the new digital human and skingen tools it makes more sense than ever to be able to use Headshot only to get a quick result on the geometry side and then removing the auto generated skin texture to replace it with some SkinGen contents, that would be really helpful, specially in cases where the original picture doesn't have a very good quality and ends up looking very artificial on the characters face.

Maybe I'm just missing how to  get rid of that auto generated texture, I asked in this forum a while ago and the methods that I got recommended and were working on previous versions of CC3 aren't working anymore for me on the latest one. So I've ended up completely ignoring the tool and importing the reference pictures as image layers and work on a CC3 base character from scratch so there is no auto generated texture messing with the new SkinGen tools that I'm really enjoying.