avatar


Consistent characters in Midjourney GenAI

It’s been a while since I posted anything for lots of reasons, that’s an offline conversation. However, this weekend something appeared on my feeds that was just too exciting not to do a quick test with and then share, which I already have on Twitter, LinkedIn, Mastodon and Facebook plus a bit of instagram. So, many channels to look at!

Back in August 2022 I dived into using Midjourney’s then new GenAI for images from text. It was a magical moment in tech for me of which there have been few over the 50+ years of being a tech geek (34 of those professionally). The constant updates to the GenAI and the potential for creating things digitally, both in 2D, movies and eventually 3D metaverse content has been exciting and interesting but there were a few gaps in the initial creation and control, one of which just got filled.

Midjourney released its character constant reference approach to point a text prompt at a reference image, in particular of a person, and to then use that person as a base for what is generated in the prompt. Normally you ask it to generate an image and try and describe the person in text or by a well known person, but accurately describing someone starts to make for very long prompts. Any gamers who have used avatar builders with millions of possible combinations will appreciate that a simple sentence is not going to get these GenAI’s to get the same person in two goes. This matters if you are trying to tell a narrative, such as, oh I don’t know… a sci fi novel like Reconfigure? I had previously written about the fun of trying to generate a version of Roisin from the book in one scene and passing that to Runway.ml where she came alive. That was just one scene, and to try any others would not have given me the same representation of Roisin in situ.

Midjourney experiments
Initial render in midjourney some months ago

The image above was the one I used to then push to runway.ml to see what would happen, and I was very suprised how well it all worked. However, on Friday I pointed the –cref tag in midjourney to this image and asked for some very basic prompts related to the book. This is what I got

epredator_in_a_computer_room_hacking_--cref_httpss.mj.runBEGdSJ_962db28e-f687-45b3-a6f1-dd3853750dc9
Another hacking version this time in a computer room not a forest
epredator_a_girl_running_from_a_car_--cref_httpss.mj.runBEGdSJ3_5d642f40-d948-4be8-93ae-0e5afc8c8254
A more active shot involving a car
epredator_a_girl_running_from_a_car_--cref_httpss.mj.runBEGdS_d3f394d0-6389-41e6-8050-cfe0d98a6827_1
A different style, running at the car but similar to the previous one
epredator_buying_marmite_in_the_supermarket_--cref_httpss.mj.ru_29a1e69f-65ad-47fe-aaae-794e3e23bbd5
Looking to buy some Marmite (a key attribute in the story and life of Roisin)
epredator_buying_marmite_in_the_supermarket_--cref_httpss.mj._35656c46-5f87-435a-ba91-b4d8c20eef0b_1
Another marmite shopping version
epredator_snowy_scene_with_girl_--cref_httpss.mj.runBEGdSJ3q2_5502223a-4746-4847-8218-ab1d13deea7b_2
A more illustrative style in the snow (sleeves are down)
epredator_snowy_scene_with_girl_--cref_httpss.mj.runBEGdSJ3q2_5502223a-4746-4847-8218-ab1d13deea7b_3
Another snow stylistic approach (notice sleave up and tattoos)
epredator_in_a_computer_room_hacking_--cref_httpss.mj.runBEGd_e3489682-baf7-4265-881e-0ab6aa8a6d95_2
A more videogame/cell shaded look to the computer use in the bunker.

As you can see it is very close to being the same person, same clothing, different styles and these were all very short prompts. With more care, attention and curation these could be made to be even closer to one another. Obviously a bit of uncanny valley may kick in, but as a way to get a storyboard, sources for a genAi movie or create a graphic novel this is great innovation to help assist my own creative process in ways that there is no other way for me to do this without some huge budget from a Netflix series or a larger publisher. When I wrote the books I saw the scenes and pictures vividly in my head, these are not exactly the same, but they will be in the spirit of those.

Metaverse and GenAI webinar for BCS

This month was the AGM for the BCS Animation and Games specialist group that I have been chairing for a very long while now. I gave a presentation from a personal view point (this is not a work presentation and I make that clear in the disclaimers, though it is what I work in too of course), on the advances in Metaverse and GenAI content creation. The full YouTube version is below but the link to the blurb and bio (and the video) at the BCS is here

We are always looking for presenters to come and share some ideas with our specialist group around all things games, animation, metaverse, esports etc, so if you are interested ping me there is a slot waiting for you. We sometimes get a big crowd, other times smaller ones but with the videos published like this it can be a useful thing to do and share.

For those of you who don’t know, BCS (formerly British Computer Society) Chartered Institute for IT is a UK based (but worldwide membership) professional body for anyone in the tech industry. It exists at all levels from just getting going in the business to Fellows with vast amounts of experience and willingness to help. It was part of my professional certification whilst at IBM and I then also became a certifier whilst there too. Volunteering and sharing ideas, such as this presentation, is one of the many ways to get involved (you don’t have to do this). It benefits you as an individual but also elevates tech roles within enterprises and organizations you work in.

You can find more at BCS, The Chartered Institute for IT (bcs.org)

Digital humans evolve

Way back in the dim and distant era of 2009 I was exploring a lot of tools to help me build virtual environments with avatars and characters that could be animated, typically in Unity. 3D modelling is an art in its right and then rigging those models for animation and applying animations to those is another set of skills too. At the time I was exploring Evolver, which in the end was bought by Autodesk in 2011. A decade on and there is a new kid on the block from Epic/Unreal called Metahuman. This cloud based application (where the user interface is also a cloud streamed one) runs in a browser and produces incredibly detailed digital humans. The animation rigging of the body is joined by very detailed facial feature rigging allow these to be controlled with full motion capture live in the development environment of Unreal. Having mainly used Unity there is a lot of similarity in the high level workflow experience of the two, as they are the leading approaches to assembling all sorts of game, film and XR content. However there was a bit of a leaning curve.

I decided to generate a couple of characters and ended up making what to me feels like Roisin and the Commander from my Reconfigure novels. Creating them in the tool plays instant animations and expressions on the faces and is really quite surreal an experience. I installed Unreal on my gaming rig with its RTX gfx card and all teh trimmings and set about seeing what I needed to do to get my characters working.

First there is an essential application called Quixel Bridge that would have been really handy a decade ago as it brokers the transfer of file formats between systems, despite standards being in place there are some quirks when you move complex 3D rigged assets around. Quixel can log directly into the metahuman repository and there is a specific plugin for the editor to import the assets to Unreal. Things in Unreal have a data and control structure called a blueprint that is a kind of configuration and flow model that can be used in a no-code (but still complex) way to get things working. You can still write c++ is needed of course.

My first few attempts to get Roisin to download failed as the beta was clearly very popular. I only took a photo of the screen not a screen cap, a bit low quality but there is more to come.

Metahumans

However, eventually I got a download and the asset was there and ready to go. Unreal has a demo application with two MetaHumans in it showing animation and lip synching and some nice camera work. Running this on my machine was a sudden rush to the future from my previous efforts with a decade old tech for things such as my virtual hospitals and the stuff on the TV slot I had back then.

Roisin ?
Roisin from Reconfigure and Cont3xt in Unreal engine

The original demo video went like this

Dropping into the edits and after a lot of shader compilation I swapped Rosin with the first metahuman by matching the location coordinates and moving the original away. Then in the Sequence controller, the film events list I swapped the target actor from the original to mine and away we go.

This was recorded without the sound as I was just getting to grips with how to render a movie rather than play or compile an application to then screen cap instead. Short and sweet but proves it works. A very cool bit of tech.

I also ran the still image through the amusing AI face animation app Wombo.AI this animates stills rather than the above which is animating the actual 3d model. I am not posting that as they are short audio clips of songs and teh old DMCA takedown bots end to get annoyed at such things.

Now I have plan/pipe dream to see if I can animate some scenes from the books, if not make the whole damn movie/tv series 🙂 There are lots of assets to try and combine in this generation of power tooling. I also had a go at this tutorial, one of many that shows the use of a Live facial animation capture via an iPhone streamed to the metahuman model. I will spare Roisin public humiliation of sounded like me and instead leave it to the tutorial video for anyone wishing to try such a thing.

Lets see where this takes us all 🙂

Is Statue?

I had a all to rare go on Call Of Duty on the Xbox One yesterday. I downloaded the latest map pack and I was intrigued as I ran around an Aztec ruin against bots that I heard one of them shout “enemy over by the statue”. I had not really noticed specific voice prompts of that nature before. However after a bit of exploring, and obviously getting fragged a lot I found the statue and I was pleasantly surprised to see this.

I haven’t explored the other levels too much yet but I am as always intrigued by the level design and the spirit and atmosphere these design elements create.
Checking the other add-ons it I also noticed that COD has gone all Sat-Nav on us. There are additional voice packs for the game commentary and in particular you can have the dulcet tones of Snoop Dog keeping you up to date on the game. (‘big fan am I’ of the Yoda voice on my Iphone TomTom Sat-Nav app).
Whilst on the subject of persona, Kinect Sports Rivals arrived a few weeks ago. It has a number of sports to try some of which differ from the usual motion sports. There is go course tennis and ten pin bowling, but without the controller so no risk of launching one into the TV. There is Jet Bike riding across waves, football(soccer), shooting, and the intriguing rock climbing too. What I found even more interesting was the kinect being used to scan me and make an avatar. Often the web cam style face camera pastes textures on a standard rig. This however creates a cartoon character of you. So clearly there are a set of noses, eyes etc and it picks ones close to the parameters. It starts by scanning your body shape then asks you to get close in to look at your face. I ran this a couple of times as I was intrigued when it asked me to remove my glasses (which I thought was a generic message) but then put glasses on my avatar. When I took the glasses off before I started it did not ask me to remove them and I had a specs free avatar. You can of course customise the avatar once this scan has been done but it seemed to work very well.

As I mentioned its caricature 🙂

Talking Heads, Photospeak and easter eggs

I am always interested in ways we are able to talk one piece of content and do something with it. I recently started using Photospeak an iphone application you can find on the app store. It is somewhat like Crazy Talk that I have used over the years for things like my Second Life Wimbledon 2008 video and the Daz in foil (of pointawayfromface) and also these examples

Photospeak lets you take a 2d facial photo, it then sends it off to a server, does some magic and returns an instrumented and working interactive semi 3d version of you face. One that tracks movement and can also lip synch. It is always a little creepy, and very uncanny valley, in particular because of the eyes. However, when recording this video of the iphone app I happened to double tap the screen and discovered this add on to it. (it may be documented but who reads docs nowadays?) It also instantly made the face more convincing I found. Not least because I wear glasses normally.

Avatar portability – Evolver to FriendsHangout (unity3d)

I have continued to explore and be impressed by the avatar model creation site evolver.com
Having created an avatar and looked at the various ways of transporting that content to places I also tried the link to the unity3d powered Friends Hangout. I had tried this a little while ago, and it has changed drastically. There are now multiple game rooms and types to experiment with, and yes it does really import your evolver avatar into unity3d and the engine for you. Something that I can see growing as more platforms open up.
friends hangout evolver avatar
For those of your who have basic 3d packages but not the human clip art that you need it is worth trying the Evolver test transport. They have a basic avatar (to do a try before you buy on your own) that lets you create a sample with a certain resolution, number of bones and file format which you can then import. I am still experimenting with the ideal format for me using Cheetah and Unity3d but it looks very promising. The bones work, the mesh is bound, the textures are there. It costs $39 to export your own creation but that seems well worth the price I have to say.
I just have to decide if I want the more freaky avatar I created or not?
evolver epred
epredator unmasked

This is the Friends Hangout embed. It uses the Unity3d, but now that on Mac Snow Leopard Safari Unity is currently broken, works fine in Firefox though if you want to try it.

Evolving epredator

I have been experimenting with many avatar creation sites recently for some projects. I have two experiments I usually try, one is the predator themed version. This usually tests the creative flexibility of a system, with a mask or mandibles messing up most systems.
The other is some green (preferably spikey) haired version thats a caricature in some way of me.
Most recently I looked at Evolver.com a great looking avatar creation service that is designed as a web based wizard for rich 3d content which is then exportable in a variety of formats, from animations to yes, 3d prints of the avatar.
It is also interesting that there is a much more complex and richer export to 3d applications with rigged bones etc for Maya and alike. This seems a good addition though I suspect that many of the users of the high end packages are already skilled in design or have suitable models to work with. I have to try and see if it fits into my “not quite a graphic designer but with aspirations” tool chain and it would be great to grab the custom me and pop it into Unity3d.
evolver epredator
evolver epredator face
This is a sample video the site makes for you with animation of the character too.

The prize for innovation in avatar interopability goes to ….

Codemasters Colin McRae Dirt 2 on the Xbox 360
I was taking a look at the latest incarnation of off road console gaming. Codemasters Colin McRae Dirt 2 on the Xbox 360. The previous Dirt game was brilliant and it looks like this is going the right way too. The demo is even packaged with voice overs from Ken Block and Travis Pastrana indicating it is a demo (i.e. not just a disabled functions as with most demos).
Having had a blast around an excellent simulation of a Baja track and then an even more impressive and stunning rally stage set in Morroco I was thinking about how this mirror world, and the purity of the experience as a simulation blended with fun and some interesting game related features (such as rewind when you total the car) would not be one to have any quirky avatar or non car related expressions of personality in it. I often use Forza 2 (soon Forza 3)(as here in 2007) as an example of how expression and customization of things like cars are done in context and that interoperability is a social and branding activity more so than a technical challenge to move data from one environment to another. “We don’t want our avatar from X, wandering around in Y as it will break the atmosphere”, “yes but we can”.
I happened to pop into the demo customization options and selected something that said avatar as a cockpit customization. I was amazed, amused and impressed when on the replay of the hectic drive I saw this.
Dirt 2 - Embedded Avatar
My personalized xbox avatar dangling from the mirror, swinging around with ragdoll physics in car. It s such a subtle little idea, yet I was surprised. There in this real yet fun off road driving simulation, where mentally I was the driver of the car in overalls and helmet, hurtling and sliding through the desert, I was able to reference back to a little bit of my more global system wide visual persona. My predalike dreadlocked avatar with his new virtual t-shirt.
It did not break the spell, it fitted. The avatar was embedded in world yet not interfering with it. Well done to the Codies !

Whilst this feature was on the 360 I am sure there are equally interesting features on all the other versions it is out mid September (which whilst i am on the subject all the games that get release pre-xmas are always in mid September which make it mighty annoying when the predlets want to by me a birthday present for the end of August!
Colin McRae: Dirt 2 (Xbox 360)
Colin McRae: Dirt 2 (PS3)
Colin McRae: Dirt 2 (PC DVD)
Colin McRae: Dirt 2 (Wii)
Colin McRae: Dirt 2 (PSP)

Avatars add up

Predlet 1.0 came home today with a username and password from school for a site called Mathletics. She was very keen to get on and have a look. She also was very keen to login herself and said the password was secret and not for me to see. (That’s a good start).

What I was struck with was how straight away the first task was customization of the character to represent you. I should not have been surprised but I was. The site is global and has elements of competition and scoreboards for completing the flash based maths games. It was very slick and very well done and she enjoyed it too. 

During the customization I got asked “dad what colour are my eyes?” and “does this hair look like mine?”. Entering work mode for a moment I said “why do you want it to look like you?”. The reply “It is me”. So we had certainly moved from the notion of the thing online being a toy to externalize.  She was showing a need to control her own brand effectively. 

I did point out that some of my characters online don’t look like me but are me, e.g. spikey green hair and that its ok to play with the look 🙂 However it is one of those stages we all have to go through. Start to be 100% accurate, then start tweaking. The difference here is that my daughter is starting this at age 5 going on 6. What will her self expression be like in the systems in 5 years time?. An interesting thought and great seeing a digital native evolve.

Mathletics

Latest merger news : Work and play combine

Last night I had the pleasure of speaking at a Nokia connecting worlds event in Second Life. The subject for the panel of us invited to talk was that of human and non human avatars. In particular we represented people with avatars in SL or multiple looks in SL that were not regular real world mirrors.
Nokia SLIt was a very interesting set of perspectives that we all brought to the event I think. The discussion of identity and the prejudice and also the power of using particular ways of interacting in virtual environments is a rich subject. It is also full of personal experience and though for each of us. 

Faced with an avatar based experience everyone who enters has to choose something, even if the choice is to do nothing, to represent them. However as they engage and act the avatar starts to not just be a visual but a repesentative of attitude and action. 

The panel was formed of Blue Linden, Myself, Digistar Brouwer of Nokia, Hiro Pendragon, Selkit Diller and Flea Bussy, moderated by Hydra Shaftoe.

Also for me the conversation and answering the questions ranged from my experience at a large corporation having represented both myself and their interests as epredator, flowed into how I show the fact I am clearly a science fiction fan with my AV and also how I try and act with the same strength and honour blended with the ethos of another character.

Clearly all that in conversation blends business and play in many complicated ways. The need for personal brand, for ego and for championing a business cause yet all being described to a group of people world wide when for me it was 8pm and I was at home having just put the predlets to bed.

It was captured by the newly forming Treet.tv (link to follow once the archive is up) 

So clearly work and play (which can be the same thing if you enjoy the work) have already merged. Sharing personal experience that some people may use for commercial advantages, or for personal enhancement crosses the boundaries. 

As I often say all this is about people. Thats the bottom line. People connecting and getting done what they need to do,  direct to one another with reduced overheads.
Nokia SLAs is also customary at any online gathering there is a post event mingle, this shows the great leveler that virtual worlds are as the social construct of audience, stage, panelists just stops and intermingles as it does at any good conference event. 

We ended up discussing WoW, Female/Male potential prejudices and sandwiches. 

There were a few eightbars (both current and former IBMers) who came along too so it was good to see everyone again. 

Thanks all for the discussion and look out for the video when its posted to hear everyones point of view.
The video is now ready

Or a direct link is here (Thank Digistar 🙂 )