Lemma Soft Forums

You can take a look at YouTube, there are lots of example there. Just type "Live2D". If you want to see how it works (how you can make 2D images become 3D), search for "Live2D modeler". It's not that simple but it's still easier than actually making 3D... not making 3D is hard, but for people who only good at 2D, Live2D could be a godsend.

Basically, you tell which part will be stretch to make it look 3D. We already have this kind of thing before (2D looks 3D but never in real time like games or interactive apps, it's always been a movie/clip).

IDK how to implement it in a video game/apps though... may be they include a module, IDK, or something? Can we use it in various software or do we need to use included software only? I admit I haven't read about that from the site.

Now my question, is it worth it? Yes, like I said, Live2D could be a godsend for 2D artist. But cost-wise, it can be cheaper if we just make full 3D. Yes, there pros and cons for using full 3D and Live2D. But for ¥147,000... we need to re-think about the budget.

Though like I said before, I think we should wait for the technology to become common and mature. Who cares about those companies who already use it. They are big companies with lots of manpower and money. If we wait a bit longer, the price could be cheaper and may be other companies will release similar software (which hopefully, cheaper) that also make competition for Live2D (which will further push the price down).

LVUER wrote: IDK how to implement it in a video game/apps though... may be they include a module, IDK, or something? Can we use it in various software or do we need to use included software only? I admit I haven't read about that from the site.

It says that they have SDKs for most of the major game systems, although you have to pay an additional license fee for each system.

Honestly, I really don't see the use of this for idie development. Between the cost for both the modeler and SDK, and the amount of time it takes to draw, model and animate a character (3-5 days), I think it would be better to just have static images or use simple transforms.

@bunbun and LVUER: Thanks for the explanation.

The live2d webpage is amazing, but I've been looking at CrazyTalk (lots of videos in youtube too) and Anime Studio and I'm not sure if live2d adds too much to this existing programs (I mean, for us, VN creators).

I don't know what do you have in mind; I'm thinking in my characters bliking, moving slightly their heads and lips to speak, maybe walking a little. But I'm not going to do a full video, a full anime.

bunbun wrote:@xavimat : Never heard of CrazyTalk though, does it work with body movement too?

There are CrazyTalk and CrazyTalk animator. The CrazyTalk simple version only animates the face, and the animator version, the full body. In an animator's forum users say that the CT-animator version is not worth, it's only a toy for simple movements, they prefer the Anime Studio. Of course, they are doing full anime; IMO this "toy" could be interesting for us.

Also, the prices:
live2d: $1,500 (Studens & Teachers version: unknown yet)
CrazyTalk $30 (PRO: $113), the non-pro looks useful enough.
CrazyTalk Animator $38 (PRO $135)
Anime Studio: $300 (there is a "Debut" version, but too limited).

My working conclusion: (after the summer) I'll try with CrazyTalk. It can export the animations to a series of PNG images, and that could be really easy to implement in renpy.

Any of you has experience with this programs?

Seems like it'd be almost as much work to make a full 3D model, but with less options once you've finished it. I may be wrong, but I felt the info provided on the website side-stepped the problem of doing things like a 180 degree turn-around of a character, showing them from significantly different angles, or animation of shading. I'd speculate that's because you'd simply need to redraw.

That being said, I actually kind of liked the look of the video demo on the site. I think it really boils down to what your needs are and what your skill-set is. If you want the characters to be able to react to lighting (inc. the cels automatically changing, automatic shadows, as well as colouration from the light source), do full rotations, and feature them in a lot of varied situations then full 3D would be more efficient in my opinion, even if it's overlaid on 2D BGs. Not better or worse looking, just more efficient. This video is full 3D and shows the benefits well I think. Imagine drawing all that shading + shadows in addition to animating the movements. Just needlessly time consuming. But again, it depends. If it's a short project, there's no use in getting your artists to learn entirely new skill-sets just for it especially if the work is quicker in 2D. Like drawing a sprite with a simple blink animation would probably be quicker than modelling, rigging, and animating a 3D equivalent. But if it's a long term project with a lot of complex stuff going on, you'll see benefits in the long run by investing in the right production method in the beginning.

Tempus wrote:Seems like it'd be almost as much work to make a full 3D model, but with less options once you've finished it. I may be wrong, but I felt the info provided on the website side-stepped the problem of doing things like a 180 degree turn-around of a character, showing them from significantly different angles, or animation of shading. I'd speculate that's because you'd simply need to redraw.

It should be impossible to make a 180 degree turn. I think even perfect 90 degree (from front facing sideway) will be very difficult to do if not impossible. If the amount of work is almost the same as making a full 3D model, then it's pretty simple option. People who's better at 3D can stick with 3D software and people who's better at 2D can use Live2D.

The advantage (and main selling point) of Live2D is to be able to use 2D images as pseudo 3D model. As 3D models can be very stiff, can be difficult and time consuming to make, and can have that "uncanny valley" problem. And I think we still can't make 3D model to look like a hand-painting.

Tempus wrote:That being said, I actually kind of liked the look of the video demo on the site. I think it really boils down to what your needs are and what your skill-set is. If you want the characters to be able to react to lighting (inc. the cels automatically changing, automatic shadows, as well as colouration from the light source), do full rotations, and feature them in a lot of varied situations then full 3D would be more efficient in my opinion, even if it's overlaid on 2D BGs. Not better or worse looking, just more efficient. This video is full 3D and shows the benefits well I think. Imagine drawing all that shading + shadows in addition to animating the movements. Just needlessly time consuming. But again, it depends. If it's a short project, there's no use in getting your artists to learn entirely new skill-sets just for it especially if the work is quicker in 2D. Like drawing a sprite with a simple blink animation would probably be quicker than modelling, rigging, and animating a 3D equivalent. But if it's a long term project with a lot of complex stuff going on, you'll see benefits in the long run by investing in the right production method in the beginning.

This video is not about Live2D, right? Since if we're already use a full 3D model, we don't need Live2D anymore.

According to this page, live2d is allowing small indies and individuals to use the SDK for free upon agreement, as long as you make less that 10,000,000 Yen.
There is also a free version of the editor for modeling and animating now.
I don't know if that helps at all, though.

The academic student/teacher price for the Pro Editor is $300. Most of us here would qualify under individuals/small businesses/doujin, so we can use the Free Cubism Editor for commercial projects at no cost.

Since someone is going to bring up a Live2D/Emote technology comparison sooner or later, I THINK -- I haven't tried the free cubism so I could be wrong here -- Emofuri is likely going to be more popular with casual artists because of the gentler learning curve. Emofuri's work method goes like this: draw over one of their template psds, import it to the program, and animate it using a bunch of sliders and export as .png, .gif (any size) or .wav. Users are restricted to what kind of poses they cam draw (custom poses require additional tech support, hence the licenses) and the base mesh itself don't translate less moe/cartoony styles well. There's also a maximum file size. However, it was easy to use and pick up, which explains the Emofuri boom among artists on dA and Tumblr.

Live2D, on the other hand, is much more flexible but the work method is different. It seems that you can draw whatever pose you like first, livetrace the coordinates on Cubism to create the mesh, and then animate it using... I don't know yet lol. But the steeper learning curve comes from the fact that users need to livetrace the model first. The maximum file size for gifs for the free version is 640 x 480, and I'm not sure what Live2D exports to at the moment. However, there's no doubt that Live2D is more flexible than Emofuri in terms of pose dynamics.

At any rate, neither technology seems like they'll work well with Ren'py because of the large number of frames involved? I agree that waiting for the technology to mature some more is a great idea. I'll see if I can set aside some time to try the Cubism editor; otherwise I'd love to hear from any artists who's given it a spin.

Btw, from an artist's perspective, Emofuri took quite a while for me to figure out. Even if someone has the commercial license, it might not mean that they can find a skilled enough/affordable enough artist who can prepare good Emofuri .psds; because an Emofuri .psd is complex, the price of a commissioned Emofuri sprite should be higher than a normal static sprite commission. On the other hand, you can generate lots of animations with one great Emofuri sprite so I guess it all balances out?

I'm still really interested in Live2D and using it in some of my future projects, but I agree it may be more of a pain to do using RenPy; which is why I started working on my RenPy to Unity converter since Live2D comes with support for it built in.

Sadly, I don't have the art chops to make something look half decent, so I'm waiting til I can find an artist to work with the software to create some Sprites that I can fit into to create a tech demo aside from using provided assets.

I've had a hard time with both programs, but I can actually get cubism to work. Despite following tutorials and doing everything "correctly" emofuri still doesn't work to well for me. :C It's also just something I'm having a hard time learning.
I'd like to see what other programs come out of this movement, and if maybe they can work with Ren'Py. Also to see if the prices go down. XD
I wonder if it could work with GameMaker?

fioricca wrote:Since someone is going to bring up a Live2D/Emote technology comparison sooner or later, I THINK -- I haven't tried the free cubism so I could be wrong here -- Emofuri is likely going to be more popular with casual artists because of the gentler learning curve. Emofuri's work method goes like this: draw over one of their template psds, import it to the program, and animate it using a bunch of sliders and export as .png, .gif (any size) or .wav. Users are restricted to what kind of poses they cam draw (custom poses require additional tech support, hence the licenses) and the base mesh itself don't translate less moe/cartoony styles well. There's also a maximum file size. However, it was easy to use and pick up, which explains the Emofuri boom among artists on dA and Tumblr.

http://lemmasoft.renai.us/forums/viewto ... 31#p330501

IMO, Live2D have better result and can be used for interactive gaming (ie: apps). Also Emote is far more expensive since they don't have any free license for commercial project and they only have subscription model license (which is a big turn off for me).

And if I'm not mistaken, Live2D have been used by several big company to make games, like Namco Bandai's Summon Night 5 (PSP), Konami's Tokimeki Memorials Girls Side 3 (PSP), and Idea Factory/Compile Heart's Neptunia Android App (Android).

LVUER wrote:IMO, Live2D have better result and can be used for interactive gaming (ie: apps). Also Emote is far more expensive since they don't have any free license for commercial project and they only have subscription model license (which is a big turn off for me).

And if I'm not mistaken, Live2D have been used by several big company to make games, like Namco Bandai's Summon Night 5 (PSP), Konami's Tokimeki Memorials Girls Side 3 (PSP), and Idea Factory/Compile Heart's Neptunia Android App (Android).

Gyakuten Yoshiwara uses Emote technology and is out on both Android and iOS if I'm not mistaken.

I agree that Live2D is more flexible and gives you better results, but if you're not an artist you'd want to think about how feasible it is to find artists that have familiarity with Live2D vs those that are familiar with Emote. Needless to say that since the ability to prepare textures for this sort of 2.5D animation is currently a rare skillset among amateur artists, prices aren't going to be cheap. (OR they could be, since there's no set market price for such products yet? I don't see anyone offering Emote/Live2D sprite commissions.) The Cubism program also seems fairly heavy (they recommend 8GB memory) so you'd need a pretty powerful PC/machine to handle the load, which comes at an additional price.

Morhighan wrote:I've had a hard time with both programs, but I can actually get cubism to work. Despite following tutorials and doing everything "correctly" emofuri still doesn't work to well for me. :C It's also just something I'm having a hard time learning.
I'd like to see what other programs come out of this movement, and if maybe they can work with Ren'Py. Also to see if the prices go down. XD
I wonder if it could work with GameMaker?

I feel you -- drawing is the easy part of Emofuri, creating custom animations is the confusing and hard part. OTL I think it's because it requires knowing some kind of animation logic that I currently don't have as I've only ever drawn static images. Do you have any samples of Cubism work that you can share, if you don't mind?

netravelr wrote:I'm still really interested in Live2D and using it in some of my future projects, but I agree it may be more of a pain to do using RenPy; which is why I started working on my RenPy to Unity converter since Live2D comes with support for it built in.

Sadly, I don't have the art chops to make something look half decent, so I'm waiting til I can find an artist to work with the software to create some Sprites that I can fit into to create a tech demo aside from using provided assets.

/quietly goes to look up Live2D tutorials and videos, will let you know if I come up with anything... /o\

fioricca wrote:At any rate, neither technology seems like they'll work well with Ren'py because of the large number of frames involved?

If I had the chance to use this technology in my VN, I'd probably use it for making variations of poses, rather than movement due to the quoted issue. I love the idea that an artist could draw one character image, and create 30 variations from it. Of course I know it isn't as simple as that but the potential for greater variation would be there.

But like everyone else, still waiting to see someone from LSF make use of it and discuss it.

Inspired by the Twitter conversation I just had, I think many artists -- especially anime-style artists -- should be prepared to create sprites compatible with the Live2D / Emofuri or other system. This means lots of padding behind the scenes so that you can tilt the sprite's head etc. and not feel dislocated. Or just access to the full unflattened psds so they can be broken apart and entered into these programs... there are artists unwilling to send the full unflattened psds to their commissioners though.

Currently there's four engines that are Live2D compatible or workaround.

Unity, Cubism SDK | Fungus (2014, 2015)| Jokerscript [JP]
Tyranobuilder
Construct 2 [plugin]
Ren'Py - how to use live2d to make movie sprites

Sorry about the necroing.