I have been starting multiple projects involving Hatsune Miku.
One is attempting to implement MikuMikuStudio code as a visualisation for XBMC, so that you could provide models and motion data, and when a specific track is played, it starts executing that model and motion. However, this has stalled slightly, when I found MikuMikuStudio written in Java, XBMC in C++ and everything having to be fired across JNI. While I got that working, and it very nearly worked. I am left with the slight issue of MikuMikuStudio and jMonkeyEngine upon which it is written trying to create windows as it starts.
Another, I mentioned on Google+ was to do with recreating the "holographic" projection of Hatsune Miku, similar to how it's done at the concerts, but "on the cheap".
This is the main topic of this post.
The original post on Google+ is basically based on something I saw when I was thinking about how it was done, finding out the massive expenses and then one of most recent inspirer's made my month about the same goal with a post on the same subject.
From that post, it turns out you don't need millions.
Initial prototype
What is needed for the initial prototype?
One is attempting to implement MikuMikuStudio code as a visualisation for XBMC, so that you could provide models and motion data, and when a specific track is played, it starts executing that model and motion. However, this has stalled slightly, when I found MikuMikuStudio written in Java, XBMC in C++ and everything having to be fired across JNI. While I got that working, and it very nearly worked. I am left with the slight issue of MikuMikuStudio and jMonkeyEngine upon which it is written trying to create windows as it starts.
Another, I mentioned on Google+ was to do with recreating the "holographic" projection of Hatsune Miku, similar to how it's done at the concerts, but "on the cheap".
This is the main topic of this post.
Hologram Concerts "On the Cheap"
The original post on Google+ is basically based on something I saw when I was thinking about how it was done, finding out the massive expenses and then one of most recent inspirer's made my month about the same goal with a post on the same subject.
From that post, it turns out you don't need millions.
Initial prototype
What is needed for the initial prototype?
- Projector.
- Plain white netting material, from local "dunelm mill" which does it fairly expensive, but wanted it asap
- computer
The projector was sourced from a friend, a sony vpl-cx21 that can do 1280x1024 apparently, calls itself a short throw projector, but it's not ultra-short. About 4ft before the focus can't do any better, but it's great for something I didn't have to pay for.
The netting I got causes a problem, I will explain when I show the images later, but it was the best I could get with the timescale of my impatience.
Computer? I have them coming out of the floorboards. But ultimately, it needs to render the movies on the projector. I found it best to render in MikuMikuStudio on a powerful machine, and use a less powerful machine to control the projector. Could ultimately end up just being one of my Raspberry PIs in the end ^_^
So. The initial prototype. I'll post the images I put from the G+ post up. I have video, but it's not great, it was shot in the dark, didn't really show it off much.
Now it looks quite good. I shall now highlight problems
- I built that frame. I'm rubbish at building things.
- The frame is heavy and cumbersome.
- I wouldn't want to dismantle it or put it back together.
- Because of that, it's still together, taking up space I don't really have to spare
- I even have a bigger one I don't want to dismantle either.
- The material is causing problems. The "cross" of light is caused by the regular spacing of the holes in the netting. If they were random it would be a fuzzy blur and would be less distracting, or I think it'd be less distracting anyway.
The good thing that it shows though is that the netting allows Miku to be projected onto a surface. That surface will then scatter the light allowing her to be seen. But that surface will also allow most other light to pass through unhindered. I should have shot a picture of the wall behind me as it was bathed in a blurry image of the pieces of Miku that didn't get captured by the netting.
So. It worked. Yay. Just so cumbersome... I was rather uninspired by this...
My flat is not so big. The largest room could barely contain the largest of the frames we had constructed. I wanted a full-size Miku to perform Weekender Girl in-front of me in all honesty. Not that the projector I currently have is really to to the task anyway.
But I was still thinking...
Then I saw a projector screen being rolled up. I had an idea. The screens they use in the concerts. Dilad? Still can't find a price on them btw.
Prototype 2
Anyway. Got sidetracked looking at info about Dilad. Where was I? Yeah. The netting. I thought about laminating the net curtains.
Yes. I don't know how weird that sounds. But actually putting the net curtains through a cheap laminator. I first thought they'd need a cold laminator, they are poly-cotton curtains so, a lot of plastic in there. Maybe they'd melt. But accidentally not paying attention to the laminator I was buying, I ended up with a hot-lamination machine. My impatience kicked-in and I tried it anyway.
I cut a piece of net curtain to A4, put it in the plastic sleeve and put it in the machine...
It was fine, the machine was fine, nothing melted and jammed, it all worked as it should and now I can wrap also images of Miku in plastic to preserve the beauty forever... ahem... sorry... sidetracked.
Projecting the same image, I now have a moving picture, we the following (no sound it's not my tune, just hum, its Weekender Girl)....
Now. Compared to the pure curtain, a few things are solved.
- The cross has turned more-or-less into a random blur. There is a slight pattern to it, but nothing immediately obvious (it's every-so-slightly bulgy in one dimension).
- it's very light and if you look carefully, it's being held up by blu-tac and it's own strength. No frame at all.
But it has it's own problems too.
- It's very light, it keeps falling over ^_^
- It's hard to make, but that's not a real major issue really
- The biggest problem? You can't see through it any more, real big problem, might as well be paper.
So, onto
Prototype 3
I realised something during the testing of prototype 2. The lamination sheets. They are very diffuse already, before they are put through the machine. They come out nice and shiny and smooth.
What if we found something that was very rough and diffuse and laminated that. The net is fairly diffuse, but at a very (at the scale I'm thinking) very large scale. It needs to be very much smaller.
Then it hit me. The sheets themselves. Just laminate a lamination sheet.
Erm? O_o
Well. I did that the first time. Got an A4 laminate wallet. Put it inside another. Put it through the machine, expected it to jam and it didn't. However there were a lot of bubbles.
So. I took a wallet tore it in half at the join on one end. Put half a wallet inside another and did the same. This time, it worked. It's not perfect, but it's a lot better.
So how well does this work?
To test it, I first made an A3 version. Those two above are A4. I wanted A3 simply because it was an A3 laminator. I want Miku big. There were a few issues with creating the larger version, a few more bubbles got through, but you don't really see them below.
So I created the laminated sheet added a few supports, it needed them this time and well, tell me what you think...
I may need to buy a decent camera at some point. The phone really is rubbish.
But the best thing, you can see through it again.
Future plans? 'faid not, that's it for now. I'm tired, it's nearly 1am.
